Downloaded 19 times
![Editorial
The theme of the 99th Annual Meeting
of the Radiological Society of North
America is “The Power of Partnership”.
Nowhere is this concept better
exemplified
than in the cooperation between
academic medical centers
and industry
partners in the development and
improvement of diagnostic imaging.
This issue of MAGNETOM
Flash contains
a wealth of examples of how such
collaborations have advanced the discipline
of MRI.
As the world population’s healthcare
needs grow, so must diagnosis and
disease
management continue to
advance. Diagnostic imaging plays
an increasingly central role in detecting
and characterizing disease, and
guiding therapy. In particular, MRI
remains a cornerstone of neurologic,
orthopedic, oncologic, and cardiovascular
imaging.
MRI has long had advantages in leveraging
useful contrast mechanisms
for visualization of anatomy and pathology.
This is well-demonstrated in articles
describing visualization of diffusion-
weighted imaging data [Doring
et al. page 12], spectroscopic imaging
of prostate cancer [Scheenen et al.
page 16], susceptibility-weighted
imaging [Ascencio et al. page 52],
Mustafa R. Bashir, M.D., is an Assistant Professor of Radiology
at Duke University Medical Center, Durham, USA. He joined
the faculty at Duke in 2010, and serves as the Director of MRI
and Director of Body MRI. In 2012, he was named the Medical
Director of the Center for Advanced Magnetic Resonance
Development,
Duke Radiology’s MRI research and development
facility. He has been awarded several industry research grants
as principal investigator and serves as site imaging principal
investigator on several NIH-funded grants. Dr. Bashir serves as
a working group chair for the Liver Imaging Reporting and
Data System (LI-RADS) committee for the American College
of Radiology and is a site radiologist for the NIH-funded Non-
Alcoholic Steatohepatitis Clinical Research Network. His clinical
and research interests include abdominal MRI, liver imaging,
quantitative imaging, and novel contrast mechanisms.
and quantitative myocardial relaxivity
mapping [Moon et al. page 104].
However, gone are the days when
lengthy examinations producing
inconsistent image quality were considered
acceptable. In an atmosphere
of rising cost and diminishing
resources, all imaging is under pressure
to demonstrate consistent
examination quality, despite increasing
use in challenging populations,
such as the obese and those with
diminished breath-holding capacity.
In their article on the New York University-
Langone Medical Center experience
using Radial VIBE*, Tobias Block
et al. show the power of a motion-robust
non-Cartesian strategy to
obtain free-breathing, artifact-free,
volumetric T1-weighted image sets in
the body [page 6]. Such paradigms
can be applied to improve image
quality in patients unable to hold their
breath, and to enhance the MRI
experience
by providing healthier
patients with a more comfortable
examination with fewer breath-holds.
In addition, colleagues at the University
Hospital of Lausanne and Northwestern
University demonstrate the
feasibility of rapid cardiac acquisition
using compressed sensing* methods
[page 108 and 117]. Such techniques
are shown to produce high-resolution,
multiplanar acquisitions in single
breath-holds, which can both shorten
total examination time and provide
comparable or more accurate measure-ments
of left ventricular ejection
fraction and stroke volume, compared
with conventional methods.
The broad availability of commercial
wide-bore systems with high channel
counts makes clinical MR imaging a
reality in a larger portion of the population.
Particularly in the United States,
where over 35% of the population is
obese [http://www.cdc.gov/obesity/
data/adult.html], high-quality imaging
is now available to more patients than
ever, with greater physical comfort.
In addition to comfort, turnaround time
is also considered an important measure
of examination quality, and as
a first-line diagnostic modality, MRI
must provide rapid, definitive diagnosis
in order for appropriate treatment
to be rendered in a timely manner.
Working Together
Editorial
“In an atmosphere of rising cost and diminishing
resources,
all imaging is under pressure to
demonstrate
consistent examination quality,
despite
increasing use in challenging populations,
such as the obese and those with diminished
breath-
holding capacity.”
Mustafa R. Bashir, M.D.
This is exemplified in the article by
Kambiz Nael et al., who describe a six-minute
comprehensive acute stroke
protocol, combining brain structure
imaging, functional measures including
diffusion- and perfusion-weighted
imaging, and MR angiography [page 44].
This ‘one-stop-shop’ approach can
facilitate rapid triage of appropriate
patients to endovascular management
while avoiding unnecessary, and potentially
dangerous, delays in diagnosis.
Finally, Mark Griswold et al. and
Masahiro Ida address another important
element of a comfortable MRI
experience, as they discuss simple
methods for optimizing frequently-used
pulse sequences to reduce
acoustic noise [page 30 and 35].
In the following pages, fifteen high-quality
articles from a diverse group
of authors are presented. These highlight
important advances that build
on the excellent contrast/visualization
capabilities of MRI, strengthen image
quality and robustness, or that
improve the patient experience and
throughput. Importantly, they show
the success that can be realized by
bringing innovators from academia
and industry together into cooperative
teams.
Happy reading, and see you at RSNA!
Editorial Board
We appreciate your comments.
Please contact us at magnetomworld.med@siemens.com
Review Board
Lisa Chuah, Ph.D.
Global Segment Manager Neurology
Lars Drüppel, Ph.D.
Global Segment Manager Cardiovascular MR
Wilhelm Horger
Application Development Oncology
Michelle Kessler
US Installed Base Manager
Berthold Kiefer, Ph.D.
Oncological and Interventional Applications
Sunil Kumar S.L., Ph.D.
Senior Manager Applications
Reto Merges
Head of Outbound Marketing MR Applications
Heiko Meyer, Ph.D.
Neuro Applications
Edgar Müller
Cardiovascular Applications
Nashiely Sofia Pineda Alonso, Ph.D.
Global Segment Manager
Men’s and Women’s
Health
Silke Quick
Global Segment Manager Body Imaging
Heike Weh
Clinical Data Manager
Antje Hellwich
Associate Editor
Sven Zühlsdorff, Ph.D.
Clinical Collaboration
Manager, Chicago, IL, USA
Ralph Strecker
MR Collaborations Manager,
São Paulo, Brazil
Wellesley Were
MR Business Development
Manager Australia and
New Zealand
Gary R. McNeal, MS (BME)
Advanced Application
Specialist,
Cardiovascular
MR Imaging Hoffman
Estates,
IL, USA
Peter Kreisler, Ph.D.
Collaborations & Applications, Erlangen,
Germany
*WIP, the product is currently under
development and is not for sale in the US
and other countries. Its future availability
cannot be ensured.
2 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 3](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-2-2048.jpg)
![Clinical Head-to-Toe Imaging Head-to-Toe Imaging Clinical
Improving the Robustness of Clinical
T1-Weighted MRI Using Radial VIBE
Kai Tobias Block1; Hersh Chandarana1; Girish Fatterpekar1; Mari Hagiwara1; Sarah Milla1; Thomas Mulholland1;
Mary Bruno1; Christian Geppert2; Daniel K. Sodickson1
1 Department of Radiology, NYU Langone Medical Center, New York, NY, USA
2 Siemens Medical Solutions, New York, NY, USA
Introduction
Despite the tremendous developments
that MR imaging has made over
the last decades, one of the major
limitations of conventional MRI is its
pronounced sensitivity to motion,
which requires strict immobility of
the patient during the data acquisition.
In clinical practice, however,
suppression of motion is often not
possible. As a consequence, MR
images frequently show motion artifacts
that appear as shifted object
copies, which are well-known as
‘ghosting’ artifacts and which, depending
on the artifact strength, can
potentially obscure important diagnostic
information. Ghosting artifacts
pose a particular problem for abdominopelvic
exams that need to be performed
during suspended respiration.
Because many patients struggle
to adequately hold breath during
the scan, the number of exams with
suboptimal image quality is relatively
high. This has impaired the acceptance
of MRI as an imaging modality
of choice in many abdominopelvic
indications. Other widely utilized MRI
applications such as head and neck
imaging are also often affected by
motion-induced ghosting artifacts,
e.g., if patients are anxious, or if
wthey cannot suppress swallowing
or coughing during the exam.
Radial k-space
acquisition scheme
The high sensitivity to motion results
from the data-sampling strategy used
in conventional MR imaging to spatially
resolve the object. Conventional
sequences acquire the data space
Interestingly, although the advantages
for clinical applications seem clear and
although the idea of radial sampling
has been known since the early days
of MRI, the technique has not been
widely employed in clinical practice so
far. Radial sampling was first described
by Lauterbur in his seminal MRI paper
from 1973 [1]. However, because
practical implementation required coping
with a number of technical complexities,
it was soon replaced by the
Cartesian acquisition scheme which
could be more easily and more robustly
implemented on early MRI systems.
These technical complexities include
a more sophisticated image reconstruction,
higher required homogeneity
of the magnetic field, and the need for
much more accurate and precise generation
of time-varying gradient fields.
Consequently, radial sampling has
only been used sporadically in research
projects while clinically established
techniques are currently almost exclusively
based on the Cartesian scheme.
Over the last several years, however,
it has become possible to resolve the
majority of issues that prevented a
practical application of radial sampling,
in part through improvements of the
MR hardware designs and in part
through new algorithmic developments.
Therefore, it is now for the first time
feasible to utilize radial acquisitions
routinely on unmodified clinical MRI
systems, with sufficient reliability and
robustness for clinical applications and
with image quality comparable to that
of the conventional Cartesian scans.
Radial VIBE sequence
The Radial VIBE sequence* is the first
available works-in-progress sequence
for Siemens MR systems that integrates
these developments for volumetric
acquisitions and provides radial
k-space sampling in a fully seamless
way, aiming at achieving higher
robustness to motion and flow effects
in daily practice. It is based on the
conventional product VIBE sequence,
which is an optimized T1-weighted 3D
gradient echo sequence (3D FLASH)
with various fat-saturation options.
Radial sampling has been implemented
using a 3D ‘stack-of-stars’ approach,
which acquires the kx-ky plane along
radial spokes and the kz dimension
(k-space) using a sampling scheme
along parallel lines (Fig. 1A), which
is usually referred to as ‘Cartesian’
sampling. The acquired parallel lines
differ by a fixed difference in the
signal
phase, which is why the
scheme is also called ‘phase encoding’
principle. However, if the object
moves during the exam, phase offsets
are created that disturb the
phase-encoding scheme. In a simplified
view, it can be thought of as
jittering
of the sampled lines, which
causes gaps in the k-space coverage
and results in aliasing artifacts along
the phase-encoding direction from
improper data sampling. Hence, the
Cartesian geometry is inherently
prone to motion-induced phase
distortions. Even if navigation or
triggering techniques are used to
minimize phase inconsistencies
within the acquired data, a certain
amount of residual ghosting
artifacts is almost always present.
The situation can be improved when
changing the k-space acquisition to
a different sampling geometry. One
promising alternative is the ‘radial’
sampling scheme, which acquires the
data along rotated spokes (Fig. 1B).
Due to the overlap of the spokes in the
center, gaps in the k-space coverage
cannot occur if individual spokes are
‘jittered’ and, therefore, appearance of
ghosting artifacts is not possible with
this scheme. Furthermore, the overlap
has a motion-averaging effect. Data
inconsistencies can instead lead to
‘streak’ artifacts. However, in most cases
the streaks have only a mild effect on
the image quality, and they can easily
be identified as artifacts due to their
characteristic visual appearance (e.g.,
Fig. 3B). Because the artifacts appear
mainly as ‘texture’ added to the underlying
object, the likelihood that lesions
get obscured is significantly lower
than for the more dominant Cartesian
ghosting artifacts.
with conventional sampling, resulting
in cylindrical k-space coverage (see
Fig. 2). This trajectory design enables
use of time-efficient fat-saturation
methods, such as Quick FatSat or
SPAIR, with minimal artifact strength,
which is important as radial scans
should be performed with fat suppression
for most applications. Although
Cartesian acquisition steps are
employed along the kz dimension,
a high degree of motion robustness
is achieved due to the use of an incoherent
temporal acquisition order.
The Radial VIBE sequence can be
used on the full range of Siemens MR
systems, including systems from the
B-line generation (e.g., MAGNETOM
Avanto, Trio, Verio) and D-line
generation (e.g., MAGNETOM Skyra,
Aera), and it can also be used on the
Biograph mMR MR-PET system as well
as Siemens’ 7T** systems. Because
the sequence does not require any
2
modification of the MR hardware or
reconstruction system, it can be
deployed to installed systems and
used clinically for fat-saturated
T1-weighted exams as a motion-robust
alternative to 3D GRE, VIBE,
MPRAGE, or 2D TSE sequences.
Clinical applications
and results
Over the last two years, the sequence
has been tested extensively at NYU
Langone Medical Center to evaluate
the achievable image quality across
various MR systems in daily routine
applications. Radial VIBE scans were
added to clinical protocols under IRB
approval in more than 5000 patient
exams and compared to established
reference protocols. Several clinical
studies have been performed or are
ongoing that investigate the improvement
in diagnostic accuracy resulting
from the absence of ghosting artifacts.
Free-breathing
abdominal imaging
A key application of the Radial VIBE
sequence is imaging of the abdomen
and/or pelvis before and after injection
of a contrast medium, which is
conventionally performed during suspended
respiration. With Radial VIBE,
it is possible to acquire the data
during
continued shallow breathing,
which therefore can be the preferred
exam strategy for patients who
are unable to sustain the normally
1A 1B
(1A) Conventional ‘Cartesian’ MRI sampling scheme along parallel lines, and
(1B) radial sampling scheme along rotated spokes that overlap in the center
of k-space.
1
Stack-of-stars
trajectory as
implemented by
Radial VIBE, which
employs radial
k-space sampling
in the kx-ky plane
and Cartesian
sampling along kz.
2
kz
* Radial VIBE is a prototype for StarVIBE.
StarVIBE is now 510k released and is
available for 1.5T MAGNETOM Aera and
3T MAGNETOM Skyra.
Radial VIBE is work in progress.
** The product is under development and not
commercially available yet. Its future availability
cannot be ensured. This research
system is not cleared, approved or licensed
in any jurisdiction for patient examinations.
This research system is not labelled according
to applicable medical device law and
therefore may only be used for volunteer
or patient examinations in the context of
clinical studies according to applicable law.
6 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 7](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-4-2048.jpg)
![Clinical Head-to-Toe Imaging Head-to-Toe Imaging Clinical
7A 7B
required breath-hold time, such
as elderly or severely sick patients.
A blinded-reader study by Chandarana
et al. demonstrated that the average
image quality obtained with free-breathing
radial acquisition is comparable
to conventional breath-hold
exams and significantly better than
free-breathing exams with Cartesian
acquisition [2]. As an example,
figure 3 compares a free-breathing
Radial VIBE exam to a conventional
Cartesian exam of a patient with
insufficient breath-hold capability.
The Radial VIBE image is affected by a
certain amount of streak artifacts
but clearly depicts a lesion in the
right lobe of the liver, which is fully
obscured in the Cartesian scan.
High-resolution
abdominopelvic
imaging
The ability to acquire data during
continued
respiration also has advantages
for the examination of patients
with proper breath-hold capacity. With
conventional Cartesian sequences,
the achievable spatial resolution in
abdominopelvic exams is limited by
the amount of k-space data obtainable
within typical breath-hold durations of
less than 20 sec. Because Radial VIBE
eliminates the need for breath holding,
it is possible to sample data over
several minutes and, thus, to increase
the spatial resolution by a significant
factor. Figure 4 demonstrates this
possibility for an isotropic 1 mm
high-resolution scan of the liver
20 min after injection of Gadoxetate
Disodium, which provides clearly
sharper visualization of the biliary
duct compared to the corresponding
Cartesian protocol. In figure 5, the
achievable resolution improvement is
shown for the case of MR enterography,
which is another good candidate
for Radial VIBE due to the higher overall
robustness to the bowel motion.
Pediatric imaging
During the clinical evaluation phase,
Radial VIBE demonstrated particular
value for the application in pediatric*
patients. Pediatric exams are often
conducted under general anesthesia
or deep sedation, which makes active
breath holding impossible. Therefore,
conventional abdominopelvic scans
are in most cases affected by respiration
artifacts that impair the achievable
effective resolution and diagnostic
accuracy. Due to the inherent
motion robustness, much sharper and
crisper images are obtained with
Radial VIBE, as evident from the depiction
of small cysts in the kidneys
of a patient with Tuberous Sclerosis
shown in figure 6. A retrospective
blinded-reader study of our case collection
revealed that 8% of all lesions
were only identified with Radial VIBE
but missed in the corresponding
Cartesian
reference exams [3].
In young neonatal patients, sedation
is usually avoided due to the higher
risk of potential adverse effects.
Imaging these patients is challenging
because they often move spontaneously
in the scanner. Also in this
patient cohort Radial VIBE provides
improved image quality and reliability,
which is demonstrated in figure 7
for a brain exam of a 4-day-old
patient, in this case compared to
a Cartesian MPRAGE protocol.
*MR scanning has not been established as
safe for imaging fetuses and infants less
than two years of age. The responsible
physician must evaluate the benefits of the
MR examination compared to those of
other imaging procedures.
4
5
6
Abdominopelvic exam of
a sedated pediatric patient
with Tuberous Sclerosis. (6A)
Because suspending respiration
is not possible under
deep sedation, conventional
exams are affected by respiration
artifacts. (6B) Radial
VIBE provides significantly
sharper images with improved
spatial resolution, as visible
from the small cysts in the
kidneys.
6A 6B
Brain exam of a 4-day-old* patient using (7A) conventional MPRAGE and (7B) Radial VIBE sequence. Due to vigorous patient
activity, the MPRAGE scan is affected by strong ghosting artifacts, while Radial VIBE provides diagnostic image quality.
*MR scanning has not been established as safe for imaging fetuses and infants less than two years of age. The responsible physician must
evaluate the benefits of the MR examination compared to those of other imaging procedures.
7
(3A) Conventional VIBE
exam of a patient failing
to hold breath during the
acquisition and (3B) Radial
VIBE acquisition during
free breathing. Radial VIBE
provides significantly higher
image quality and reveals
a lesion in the liver (arrow)
not seen on the conventional
scan
3
3A 3B
(4A) Conventional breath-hold
VIBE exam of a patient
20 min after injection of
Gadoxetate Disodium. (4B)
Because Radial VIBE exams
can be performed during
continued respiration, data
can be acquired over longer
time, resulting in clearly
improved resolution (here
1.0 mm isotropic).
4A 4B
MR enterography using (5A)
conventional VIBE and (5B)
Radial VIBE acquisition. The
higher motion robustness
achieved with Radial VIBE
leads to sharper images and
improved resolution.
5A 5B
8 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 9](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-5-2048.jpg)
![Clinical Head-to-Toe Imaging Head-to-Toe Imaging Clinical
8A 8B 9B 9C
Examination of the neck and upper chest using (8A) a conventional 2D TSE sequence and (8B) Radial VIBE. Because of the
respiration and strong blood flow, the TSE scan shows drastic artifacts. Two suspicious lesions are more clearly visible on
the Radial VIBE exam (arrows).
Imaging of the neck
and upper chest
Although imaging of the head and
neck region appears less critical at
first glance, severe motion-related
artifacts occur quite often in routine
exams. Conventional neck protocols
usually include slice-selective
T1-weighted TSE sequences, which
are especially sensitive to motion and
flow. If patients are unable to suppress
swallowing or coughing during
the acquisition, images are rendered
non-diagnostic. Furthermore, adequate
examination of the upper chest
region is often not possible because
of drastic artifacts from respiration
and strong blood flow in the proximity
of the heart. Radial VIBE exams
are a promising alternative for this
application and are largely unaffected
by swallowing, minor head movements,
or flow, which is illustrated
in figure 8. The sequence also maintains
a convincing sensitivity to chest
lesions in the presence of respiratory
motion [4]. Because Radial VIBE
scans are immune to ghosting artifacts,
exams can be performed with
high isotropic spatial resolution,
which allows for retrospective reconstruction
in multiple planes (MPRs).
In this way, it is possible to substitute
multiple conventional slice-selective
protocols in varying orientation with
a single Radial VIBE high-resolution
scan. A representative example is
shown in figure 9.
Imaging of the orbits, inner
auditory canal, and full brain
Finally, the sequence also offers
improved sharpness and clarity for
the examination of the orbits. When
patients move the eyes or change
the position of the eyelids during the
exam, conventional protocols show
a band of strong ghosting artifacts
along the phase-encoding direction,
which can make identifying pathologies
a difficult task. Radial VIBE provides
cleaner depiction of the optic
nerves and improved suppression of
intra- and extraconal fat [5]. Flow
effects from surrounding larger blood
vessels can lead to mild streak patterns
but are less prominent than for
most Cartesian protocols and can
be additionally attenuated with the
use of parallel saturation bands. The
possibility to create high-resolution
MPRs is another advantage of using
Radial VIBE for this application, which
is demonstrated in figure 10 for a
patient with optic nerve sheath
meningioma. In a similar way, the
sequence can be applied for examinations
of the inner auditory canal (IAC)
or the full brain, in which a particularly
high sharpness of vessel structures is
achieved.
Conclusion
The large number of successful patient
exams of various body parts conducted
with Radial VIBE over the last two years
demonstrates that radial sampling is
now robust and reliable for routine use
on standard clinical MR systems. Due
to the higher resistance to patient
motion and the absence of ghosting
artifacts, improved image quality can
be obtained in applications where
motion-induced image artifacts are a
common problem. In particular, the
Radial VIBE sequence enables exams of
the abdomen and upper chest during
continued shallow respiration, which
can be a significant advantage for
patients that struggle to adequately
hold breath. Furthermore, the sequence
enables reconfiguring exam protocols
towards higher spatial resolution and
allows consolidating redundant acquisitions
into MPR-capable isotropic
scans. Because the sequence works
robustly on existing MRI hardware,
Radial VIBE has the potential to find
broad application as motion-robust
T1-weighted sequence alternative and
will complement the spectrum of clinically
established imaging protocols.
8
Sag Cor Tra
Neck exam using transversal Radial VIBE acquisition with 1 mm isotropic resolution. Due to the robustness to swallowing
and minor head motion, high quality 3D scans are possible that can be reconstructed in multiple planes (MPR). This enables
consolidating redundant 2D protocols with varying scan orientation.
9
9A
References
1 Lauterbur PD. Image formation by induced
local interactions: Examples employing
nuclear magnetic resonance. Nature
242:190–191, 1973.
2 Chandarana H, Block KT, Rosenkrantz AB,
Lim RP, Kim D, Mossa DJ, Babb JS, Kiefer B,
Lee VS. Free-breathing radial 3D
fat-suppressed T1-weighted gradient echo
sequence: a viable alternative for
contrast-enhanced liver imaging in
patients unable to suspend respiration.
Invest Radiology 46(10):648-53, 2011.
3 Chandarana H, Block KT, Winfeld JM,
Lala SV, Mazori D, Giuffrida E, Babb JS,
Milla S. Free-breathing contrast-enhanced
T1-weighted gradient-echo imaging with
radial k-space sampling for paediatric
abdominopelvic MRI. European Radiology,
September 2013.
4 Chandarana H, Heacock L, Rakheja R,
Demello LR, Bonavita J, Block KT,
Geppert C, Babb JS, Friedman KP.
Pulmonary Nodules in Patients with
Primary Malignancy: Comparison of
Hybrid PET/MR and PET/CT Imaging.
Radiology 268(3):874-81, 2013.
5 Bangiyev L, Raz E, Block KT, Hagiwara M,
Yu E, Fatterpekar GM. Contrast-enhanced
radial 3D fat-suppressed
T1-weighted gradient echo (Radial-VIBE)
10
Multiplanar reconstructions
of a transversal
Radial VIBE exam with
0.8 mm resolution. The
sequence achieves good
fat suppression and
provides sharp depiction
of the optic nerves
without artifacts from
eye motion. An abnormal
contrast enhancement
of the left optic nerve is
clearly visible (arrows),
which is indicative of
an optic nerve sheath
meningioma.
sequence: A viable and potentially
superior alternative to conventional
T1-MPRAGE with water excitation
and fat-suppressed contrast-enhanced
T1W sequence for evalu-ation
of the orbit. ASNR Annual
Meeting 2013: O-413.
Contact
Tobias Block, Ph.D.
Assistant Professor of Radiology
New York University
School of Medicine
Center for Biomedical Imaging
New York, NY 10016
USA
tobias.block@nyumc.org
10A 10B
10C
10 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 11](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-6-2048.jpg)
![3
Same patient as in figure 3 with curved reconstructed MPRs in the
coronal plane. (4A) Clearly visible the discontinuity in the spinal cord due
to composing errors. (4B) Significantly better registration of the stations.
4
4B
New Features of syngo MR D13 for Improved
Whole-Body Diffusion-Weighted MRI
Thomas Doring1; Ralph Strecker2; Michael da Silva1; Wilhelm Horger3; Roberto Domingues1;
Leonardo Kayat Bittencourt1; Romeu Domingues1
1 CDPI and Multi-imagem (DASA), Rio de Janeiro, Brazil
2 Siemens Ltda, São Paulo, Brazil
3 Siemens Healthcare, Erlangen, Germany
Backround
Whole-body diffusion-weighted
imaging (WB-DWI) is gaining in
clinical importance for oncological
imaging.
It has been shown to be a
promising tool principally for tumor
detection, tumor characterization,
and therapy monitoring of bone
metastases [1, 2]. The clinical implementation
of WB-DWI aims for standardization
of the acquisition protocol.
For this reason, further improvements
of data acquisition, analysis
and display of the results are
requested.
The recently launched new software
version syngo MR D13 provides several
new features for WB-DWI such
as variable averaging of the b-values,
inline composing, and a Bias Field
Correction (BiFiC) filter in order to
overcome previously existing limitations
such as long acquisition times,
mis-registration between, and intensity
inhomogeneities across image
stations.
New features in
syngo MR D13 b-value
specific averaging
One limitation of the broader clinical
usability is the long acquisition time
of over 25 minutes for head-to-pelvis
WB-DWI MRI. For more efficient scanning
the new feature b-value specific
averaging was developed: This allows
us to set the number of averages
(NEX) for each b-value individually.
The current product protocol uses a
b-value of 50 with NEX 2, and a
b-value of 800 with NEX 5, resulting in
a reduction of scan time of 30%, when
compared to the previously used NEX
5 for all b-values (Fig. 1). A similar
image quality can be achieved for lower
b-value images with lower NEX with
almost no impact on the signal-to-noise
ratio (SNR) for the calculated ADC.
New composing mode
diffusion
An inline composing filter can be activated
within the diffusion sequence
of syngo MR D13 on the diffusion
taskcard of the sequence (Fig. 2). The
composing itself is a fully automatic
process and creates for each b-value
a continuous stack of composed images
as an individual new series. Similarly
a new series for the optional calculated
b-value images is generated. Depending
on the local shim situation the
frequency
differences between neighbored
stations can lead to discontinuities
of anatomical structures like the
‘broken-spine’ artifact. During the composing
step a correction is applied
showing a much smoother transition
of local anatomy (Fig. 2).
In older software versions (syngo MR
D11) composing had to be done
manually
within the syngo.3D taskcard
by dragging and dropping all trace-weighted
series at once with no possibility
to correct for any discontinuities
in the anatomy. The new inline composing
feature significantly improves
the acquisition workflow of the technologist
as it allows to load the single
composed series to syngo.3D for the
generation of the 3D reformatted
maximum
intensity projection (MIP)
images.
Bias Field Correction
(BiFiC) filter
The BiFiC filter as a homomorphic filter
aims to normalize inhomogeneities
in image intensities from multi-station
measurements such as whole-spine
imaging. After completing the inline
composing step the filter is automatically
applied to the composed 3D continuous
image stack and saved as the
new composed series. The strength
of the filter (weak, medium, strong) can
be set within the diffusion task card
(Fig. 2, arrow).
In the Diffusion taskcard it is now possible to select the number of averages
for each b-value individually (Here: b50 NEX 2, b800 NEX 5, red arrows).
1
1
4A
64-year-old female patient,
after surgery, with endometrial
stromal sarcoma, that has
evolved with bony metastases.
Acquisition parameters: 1.5T
MAGNETOM Aera, echoplanar
imaging diffusion sequence
with fat suppression (STIR),
TR 14100 ms, TE 79 ms,
TI 180 ms, 4 stations, 50 slices
with 5 mm slice thickness,
no gap, voxel size
1.7 × 1.7 × 5 mm3, b50 with
2 and b800 with 5 averages.
(3A) b800 manually composed
images within the syngo.3D
tool shows the artifact in the
neck shoulder transition where
the stations are joined (arrows).
(3B) The new inline composing
feature demonstrates significantly
better composing of the
neck and shoulder transition.
3A 3B
By checking the Inline Composing box in the diffusion taskcard of the
sequence the automatic composing modus is activated. The strength of the
BiFiC filter can be set to weak, medium or strong (red arrows).
2
2
Head-to-Toe Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 13
Clinical Head-to-Toe Imaging
12 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-7-2048.jpg)
![Clinical Oncology Oncology Clinical
The Metabolite Ratio in Spectroscopic
Imaging of Prostate Cancer
Alan J. Wright; Thiele Kobus; Arend Heerschap; Tom W. J. Scheenen
Radboud University Nijmegen Medical Centre, Radiology Department, Nijmegen, The Netherlands
Introduction
Prostate cancer is the second leading
cause of cancer related death in
Western countries [1]. The prevalence
of the disease is very high, but many
men diagnosed with the disease will
die from unrelated causes. This is
because prostate cancer very often
is a disease of old age that grows
slowly. Common treatment for prostate
cancer in clinical practice involves
radical resection of the entire gland
or radiotherapy with a dose distributed
over the whole organ. Provided
that the cancer has not metastasized,
these therapies are curative, though
concern over their side effects has
led to patients and their doctors delaying
this treatment and, instead, entering
into active surveillance or watchful
waiting programs. In order for
patients to safely forgo curative treatment,
it is essential to characterize
their disease: to determine that it is
sufficiently benign that growth will
be slow and metastasis improbable.
Selecting these patients, with low risk
disease, that are appropriate for active
surveillance requires accurate diagnosis
of not just the presence of tumor,
but how aggressive it is: i.e. how fast
it is growing and how likely it is to
metastasise to the lymphatic system.
Magnetic resonance imaging (MRI)
is an emerging technique for making
this patient selection. It can diagnose
the presence of tumor, localize it in
the organ and provide information as
to how aggressive it is. The MRI exams
employed for this purpose usually
involve multiple imaging sequences
including a T2-weighted sequence,
diffusion-weighted imaging (DWI) and
one or more further techniques such as
dynamic contrast enhanced MRI (DCE-MRI)
or Proton Magnetic Resonance
Spectroscopic Imaging (1H MRSI) [2].
Radiologists can read the different
imaging modalities to decide the location,
size and potential malignancy
of the tumor which are all indicators
of its metastatic potential. Acquiring
and reporting imaging data in this
way is known as multiparametric (mp)
MRI. MRSI is the only mpMRI methodology
that acquires data from molecules
other than water [19]. A three
dimensional (3D) 1H MRSI data set
consists of a grid of spatial locations
throughout the prostate (see Fig. 1)
called voxels. For each voxel a spectrum
is available. Each spectrum
consists of a number of peaks on a
frequency axis, corresponding to
resonances from protons
with a certain
chemical shift in different molecules.
The size of a peak at a certain frequency
(chemical shift) corresponds to the
amount of the metabolite present in
the voxel. In this way MRSI measures
the bio-chemicals in regions of tissue
in vivo without the need for any external
contrast agent or invasive procedures.
Examples of spectra from two
voxels, acquired at a magnetic field
strength of 3 Tesla (3T), are given in
figure 1B, C, which clearly shows the
differing profiles that are characteristic
of benign prostate tissue and its tumors.
Important metabolites
in prostate MRSI
The initial papers on in vivo prostate
MRSI were performed at a magnetic
field strength of 1.5T [3-5], and three
assignments were provided for the
observed resonances: choline, creatine
and citrate (Fig. 2). The small number
of these assignments reflected the
simplicity of the spectrum, which contained
two groups of resonances: one
in the region of 3.3 to 3 ppm, which
will be referred to as the choline-creatine
region, and another at 2.55–2.75
ppm, which shall be called the citrate
group. These assignments related to
what were believed to be the strongest
metabolite resonances. People should
be aware however, that the assignments
are representative of multiple
similar molecules. The choline assignment
reflects the methyl resonances
from multiple compounds containing
a choline group (Fig. 4): choline,
phosphocholine and glycerophosphocholine.
Similarly, creatine refers to
both creatine and phosphocreatine.
In between the choline and creatine
signals another group of resonances
are present: the polyamines (mainly
spermine and spermidine). The citrate
resonances are from citrate only but
can have a complicated shape, although
in vivo at 1.5T they give the appearance
of a single peak. Nowadays a magnetic
field strength of 3T is used more
and more for prostate spectroscopic
imaging, which gives opportunities to
better resolve the choline, polyamines,
creatine resonances, but also changes
the shape of the citrate signal.
Larger choline signals are associated
with tumor in nearly all cancers [6].
High choline signals are interpreted
as being evidence of rapid proliferative
growth and, more directly, the
increased membrane turnover
required for cell division. Membranes
contain phospholipids: phosphatidyl
choline and phosphatidyl ethanolamine,
which are synthesised by
a metabolic pathway involving choline-
containing metabolites known
as the Kennedy pathway. It is in the
synthesis and catabolism of these
products, upregulated in proliferative
tumor growth, that causes the
increase in these signals.
The large amplitude of citrate resonances
observed in prostate tissue is
due to an altered metabolism particular
to this gland. Prostate tissue accumulates
high concentrations of zinc
ions which inhibit mitochondrial aconitase,
leading to a build up of citrate
in the prostate’s epithelial cells [7].
This citrate is further secreted into
the ductal spaces of the prostate as
(1A) T2-weighted MR image of a transverse section through a prostate
with an overlaid grid of MRSI spectra from voxels within the prostate.
1
(1B) One example spectrum shown on a ppm scale from a region of benign prostate tissue. (1C) A spectrum from
another voxel that, in this case, co-localises to a region of tumor.
1
Benign tissue
Choline-Creatine
Citrate
3.2 2.9 2.6
ppm
Tumortissue
Choline
3.2 2.9 2.6
ppm
Examples of 1H MRSI spectra acquired from benign prostate tissue and
tumor at 1.5T.
2
1A
1B 1C
2A 2B
16 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 17](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-9-2048.jpg)
![Clinical Oncology Oncology Clinical
T2-weighted MRI image of a transverse section through a prostate as shown in
figure 1 with an overlaid grid of MRSI-voxel data displayed as fitted spectra (5A).
5
1.5T
Cho
Ci
ppm
Choline Phosphocholine Glycero-phospholine
CH3
N
CH3 CH3
OH
CH3
N
CH3 CH3
OPO3H2
CH3
N
CH3 CH3
Creatine Phospho-creatine Citrate
CH3
N
NH2
NH
O
OH
CH3
N
NH
PO3H2 OH
NH
part of prostatic fluid, which has
a high concentration of this metabolite.
Prostate carcinomas do not accumulate
zinc ions, so they do not have
this high citrate concentration. The
increased presence of tumor cells
within a 1H MRSI voxel can, therefore,
have two diminishing effects on the
observed citrate signals: epithelial
cells that accumulate citrate can transform
into, or be replaced by, tumor
which has low citrate, or the lesion
can grow through the ductal spaces,
3T
Cho
Ci
ppm
OPO2HO
OH
OH
O
O OH O
OH OH
HH HH
OH
O
thus displacing the prostatic fluid.
The relative contribution of each of
these two physiological changes,
whether we are observing tumor formation
and malignant progression
or a histological change in tumor
invasion of ductal structure, is not yet
known. It is, however, clear that there
is an inverse correlation of the levels
of citrate metabolite and tumor cell
density with some evidence to support
a similar correlation with the aggressiveness
of the tumor as well [8].
The introduction of
a metabolite ratio
To transform the described changes
in choline and citrate signals between
benign (high citrate) and tumorous
tissue
(low citrate, high choline) into
a marker for prostate cancer, the
metabolite ratio was introduced [3-5].
The signal intensities of the different
spectral peaks were quantified by
simple
integration of the two groups
of resonances (the choline-creatine
region and the citrate group), and the
results were expressed as a ratio of
the two. This gave the choline plus
creatine over citrate ratio (abbreviated
to CC/C [4]) or its inverse (with citrate
as the numerator, [3, 5]). With choline
in the numerator and citrate in the
denominator, it became a positive
biomarker for the presence of cancer.
Acquiring the MRSI data sets
As the prostate is embedded in lipid
tissue, and lipids can cause very strong
unwanted resonance artefacts in prostate
spectra, the pulse sequence to
acquire proton spectra is equipped with
five properties to keep lipid signals out
and retain optimal signals-of-interest
in the whole prostate [9].
1. Localization of the signal with slice-selective
pulses. The point resolved
spectroscopy sequence (PRESS) is
a combination of one slice selective
excitation pulse and two slice
selective refocusing pulses leading
to an echo at the desired echo time.
The three slices are orthogonal,
producing an echo of the volume-of-
interest (crossing of three slices)
only.
2. Weighted acquisition and filtering.
Proton MRSI data sets are acquired
using a phase encoding technique
where the gradients across spatial
dimensions are varied with each
repeat of the pulse sequence. By
using weighted averaging of these
phase encoding steps (smaller
gradient
steps are averaged more
often than larger gradient steps)
and adjusted filtering of the noise
in these weighted steps, the resulting
shape of a voxel after the mathematical
translation of the signal
into an image (Fourier Transform)
is a sphere. Contrary to conventional
acquisition without filtering,
the spherical voxels after filtering
are not contaminated with signals
from non-neighboring voxels.
3. Frequency-selective water and lipid
suppression. The pulse sequence
has two additional refocusing pulses
Structural formulas of the key small molecule metabolites observed in the
spectra of prostate tissue. For each group, cholines and creatines, the common
moiety is highlighted in red. The protons that give the MR spectral resonances
present in the choline-creatine region are indicated in green. Choline-containing
metabolites have 9 co-resonant protons in the region 3.2–3.25 ppm. Creatines
have three co-resonant protons at 3.05 ppm. Citrate has four protons that resonate
at two chemical shifts (2.6 and 2.7 ppm), one for each proton in a pair bonded
to the same carbon. This pair also has a coupling between them and the symmetry
of the whole molecule ensures that two protons co-resonate at each frequency.
4
Simulated spectral shape of Cho and Ci for typical echo times (120 ms at 1.5T,
145 ms at 3T). Identical concentrations, i.e. scale factors, are applied, but
different line broadening of the signals (4 Hz at 1.5T; 8 Hz at 3T). Note the
difference in the spectral shape of Ci and the different peak amplitude ratios
for Cho/Ci.
3
3A
4
3B that only touch upon water and
lipid signals. Together with strong
crushing gradients, signals from
water and lipids are suppressed.
4. Outer volume suppression. Around
the prostate, slice-selective pulses
can be positioned to suppress all
signals in the selected slabs. These
slices can be positioned quite close
to the prostate, even inside the
PRESS-selected volume-of-interest.
5. Long echo time. To accommodate
all localization and frequency
selective pulses, the echo time of
1H MRSI of the prostate is around
120 ms at 1.5T and 145 ms at 3T.
At longer echo times, lipid signals
decay due to their short T2 relaxation
time.
The prostate is small enough
(< 75 cubic centimetres) to allow a
3D 1H MRSI data set to be acquired,
with complete organ coverage,
within 10 minutes of acquisition time.
The nominal voxel size is usually
around 6 × 6 × 6 mm, which after
filtering as described above results
in a true voxel size of 0.63 cm3.
Spectral patterns
Due to multiple different protons in
the molecule, a single metabolite can
have multiple resonances. If interactions
exist between protons within
a metabolite, the shape of a spectral
peak can be complicated. A resonance
group of protons that has a mixture
of positive and negative parts is said
to have a dispersion component in
its shape; a symmetrical-positive peak
is referred to as an absorption shape.
Choline and creatine resonances
appear as simple peaks (singlets),
5A
5B 5C
5 Two spectra (shown previously in figure 1) with their fitted model metabolite signals (5B, C).
18 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 19](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-10-2048.jpg)
![Clinical Oncology Oncology Clinical
Table 1: Typical integral values of the CC/C ratio in prostate tissue
at 1.5T [17] and pseudo integral values of CC/C at 3T [18]:
Tissue 1.5 Tesla* 3 Tesla**
Non-cancer peripheral zone 0.28 (0.21– 0.37) 0.22 (0.12)
Non-cancer central gland*** 0.36 (0.28–0.44) 0.34 (0.14)
Cancer 0.68 (0.43–1.35) 1.3 (3.7)
*median and 25th and 75th percentile **mean and standard deviation ***combined transition zone and central zone
T2-weighted MR image of a transverse section through a prostate as shown in figure 1 with an overlaid grid of MRSI-voxel data
displayed as spectra (6A) and as the pseudo integral CC/C ratio (6B) from metabolite fitting of the spectra. The ratio data points
are interpolated and shown on a color scale of values from 0 (blue) to 3 (red). The tumor containing region of the prostate
corresponds to the higher ratio values: the cyan-red area in the image.
although they very often cannot be
separated from each other as they
overlap within in vivo spectral linewidths.
The structure of citrate, given
in figure 4, results in protons at two
different chemical shifts, with coupling
between each proton and one
other (a strongly coupled spin system).
The spectral shape of these
protons depends on their exact chemical
shift, the coupling constant
between them, the pulse sequence
timing and the main magnetic field
strength. At an echo time of 120 ms
at 1.5T (and a very short delay
between excitation and first 180
degree refocusing pulse), the spectral
peaks of citrate are close to a
positive absorption mode. The spectral
shape consists mainly of an inner
doublet with small side lobes on
the outer wings. Together with line
broadening the citrate protons quite
closely resemble a single, somewhat
broadened peak. The small side lobes
around this peak are hardly detectable
over the spectral noise in vivo.
At 3T with an echo time of 145 ms
(examples given in Fig. 1), the negative
dispersion components of the
citrate shape cannot be ignored. Its
side lobes are substantially larger and
reveal also some negative components
[10]. Therefore the area under
the curve, the integral, is substantially
smaller at 3T than at 1.5T. Because
of its complicated shape, it is essential
at 3T to incorporate this shape in
quantification of the signal.
Signal quantification by
integration or metabolite
fitting
The size of the peaks of the individual
resonances represent the amount
of the metabolite present in the voxel.
Integration provides a simple method
to quantify the spectra, as long as
all signals have an absorption shape.
Although it cannot discriminate
between overlapping resonances, as
long as overlapping signals (choline
and creatine) are summed in a ratio
this does not matter. With clear separation
between citrate resonances
and the choline-creatine region, the
CC/C ratio can be calculated. However,
as pointed out earlier, the spectral
shape of citrate is not straightforward,
and ignoring the small
satellites at 1.5T, or simply integrating
the large dispersion parts of the
signal at 3T, would inevitably lead to
underestimation of the total citrate
signal intensity. An alternative is to
fit the spectra with models of the
citrate resonances with their expected
shape. The shape can either be
measured, using a solution of citrate
placed in the MRI system and a spectrum
acquired with the same sequence
as the in vivo data, or it can be calculated
using a quantum mechanical
simulation (Fig. 3). By this process of
spectral fitting, models of each metabolite’s
spectral peaks are fit to the total
spectrum and the intensities of each
fitted model are calculated. A linear
combination of the metabolite models
is found by the fitting routine such
that
Data = C1
*choline model + C2
*
creatine
model + C3
*citrate model
+ baseline Eqn1.
The coefficients C1-4 give the relative
concentrations of the individual
metabolites.
When fitting with syngo.via, the result
of a fit to a spectral peak can be
expressed in two ways: as an integral
value, which describes the area under
the fitted spectral peak, or as a relative
concentration (incorporating the number
of protons in the corresponding
peak) of the metabolite, called the scale
factor (SF) of the metabolite.
As noted earlier, the integral value
of citrate is different for 1.5 vs. 3 Tesla
due to the different spectral patterns
and would also change if pulse
sequence timing other than standard
6
would be used. If the scale factor is
multiplied with the number of resonating
protons (#H), it represents the
intensity of a signal, in relation to the
integral value of a pure singlet of one
resonating proton in absorption mode.
We call this entity pseudo integral,
which is calculated as A.
pseudo integral (Metabolite) =
#H * SF (Metabolite).
For citrate this pseudo integral is perhaps
best described as the numerical
integral of the magnitude (all negative
intensity turned positive) of the citrate
spectral shape, ignoring signal cancellations
of absorption and dispersion
parts of the shape.
The spectral fits are shown for the two
spectra in figure 5 with model spectra
of the three metabolites choline, creatine
and citrate. It can be seen from
these spectra that the relative amplitudes
of the metabolites vary between
the benign and the tumor spectrum.
As expected, the benign spectrum has
a higher citrate amplitude while the
tumor has a greater choline amplitude,
relative to the other metabolites. Combined
in the CC/C ratio, the positive
biomarker for the presence of tumor
in the prostate is calculated.
Depending on the used quantification
(spectral integration without fitting
(a), fitted relative concentrations (b)
or pseudo integrals (c)) the CC/C can
be calculated by:
(a) {Integral(Choline) + Integral
(
Creatine)} / Integral(Citrate)
(b) {SF(Choline) + SF(Creatine) } /
SF(Citrate)
(c) {9*SF(Choline) + 3*SF(Creatine)} /
4*SF(Citrate), respectively.
The numbers in the last equation
correspond
to the number of protons
of the different signals. Generally,
use of the pseudo integral ratio is
strongly preferred, as it is least sensitive
to large variations in individual
metabolite fits in overlapping signals
(choline and creatine). Note (again)
that this pseudo integral ratio does
not aim to provide a ratio of absolute
metabolite concentrations, as this is
very difficult with overlapping
metabolite signals, partially saturated
metabolite signals due to short TR
(T1 effects), and variation in signal
attenuation due to the use of a long
echo time (T2 effects).
Now, what could be the effect on
the ratio if further metabolites are
included in the fitting? Could even
polyamines be incorporated in the
analysis [11]? After separate fitting,
the main focus of the analysis could
just be on choline and citrate, which
have opposite changes in intensity
with cancer, to make a simpler and
potentially more sensitive choline/
citrate ratio. Various metabolite
ratios have been proposed [12, 13],
and there is certainly value in using
choline over creatine as a secondary
marker of tumor malignancy that can
give complementary information to
the CC/C ratio [14-16]. However, any
of these interpretations are limited
by how well the individual metabolite
resonances can be resolved. At 3T
the choline, polyamines and creatine
resonances all overlap (Figs. 1 and 5).
In practice this lack of resolution in
the spectrum translates to errors in
the model fitting
where one metabolite
can be overestimated at the
expense of another. For example a
choline over citrate ratio could be
underestimated if the polyamines fit
was overestimated and accounted for
some of the true choline signal.
While acquisition and fitting methods
are being actively researched to
improve the individual quantification
of these metabolites, it is more reliable
to stick to the pseudo-integral
CC/C ratio.
Once reliably calculated, the CC/C
ratio combines the essence of the
observable spectroscopic data into a
single quantity that can be displayed
on an image (Fig. 6), combining the
key information into a simple to read
form for radiological reporting.
Published values of the ratios for
tumor and benign tissue, which are
calculated in a similar way to the
syngo.via fitting, are listed in table 1.
Future perspective of MRSI
for prostate cancer
The CC/C ratio is the most used
method for interpreting 1H MRSI data
of prostate and prostate cancer. It
remains, essentially, the integral of
the choline-creatine region divided
by the citrate region, a simple combination
of the metabolite information
in a single-value marker that is sensitive
to the presence of tumor. The
use of areas under the resonances in
the ratio has the implication that the
absolute value of this biomarker is
largely dependent on the acquisition
sequence used. Any change in field
strength, the pulses or pulse timings
will change resonance amplitude and
shape due to T1 and T2 relaxations
and the scalar couplings of especially
citrate. Values of the ratio quoted
in the literature for tumor or benign
tissues depend strongly on how the
6A 6B
20 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 21](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-11-2048.jpg)
![Clinical Oncology Pediatric ImagHinogw C-lIi-ndioc-aitl
References
1 Siegel R, Naishadham D, Jemal A. Cancer
statistics, 2012. CA Cancer J Clin 2012;
62(1):10-29.
2 Hoeks CMA, Barentsz JO, Hambrock T,
Yakar D, Somford DM, Heijmink SWTPJ,
Scheenen TWJ, Vos PC, Huisman H, van
Oort IM, Witjes JA, Heerschap A, Fütterer
JJ. Prostate Cancer: Multiparametric MR
Imaging for Detection, Localization, and
Staging. Radiology 2011; 261: 46 –66.
3 Heerschap A, Jager GJ, van der Graaf M,
Barentsz JO, Ruijs SH. Proton MR
spectroscopy of the normal human
prostate with an endorectal coil and a
double spin-echo pulse sequence.
Magn Reson Med 1997;37(2):204-213.
4 Kurhanewicz J, Vigneron DB, Hricak H,
Parivar F, Nelson SJ, Shinohara K, Carroll
PR. Prostate cancer: metabolic response
to cryosurgery as detected with 3D H-1
MR spectroscopic imaging. Radiology
1996;200(2):489-496.
5 Kurhanewicz J, Vigneron DB, Nelson SJ,
Hricak H, MacDonald JM, Konety B,
Narayan P. Citrate as an in vivo marker
to discriminate prostate cancer from
benign prostatic hyperplasia and normal
prostate peripheral zone: detection via
localized proton spectroscopy. Urology
1995;45(3):459-466.
6 Glunde K, Bhujwalla ZM, Ronen SM.
Choline metabolism in malignant transformation.
Nat Rev Cancer;11(12):
835-848.
7 Costello LC, Franklin RB. Novel role of
zinc in the regulation of prostate citrate
metabolism and its implications in
prostate cancer. The Prostate 1998;
35(4):285-296.
8 Giskeodegard GF, Bertilsson H, Selnaes
KM, Wright AJ, Bathen TF, Viset T,
Halgunset J, Angelsen A, Gribbestad IS,
Tessem MB. Spermine and citrate as
metabolic biomarkers for assessing
prostate cancer aggressiveness. PloS
one;8(4):e62375.
9 Scheenen TW, Klomp DW, Roll SA,
Futterer JJ, Barentsz JO, Heerschap A.
Fast acquisition-weighted three-dimensional
proton MR spectroscopic imaging
of the human prostate. Magn Reson
Med 2004;52(1):80-88.
10 Scheenen TW, Gambarota G, Weiland E,
Klomp DW, Futterer JJ, Barentsz JO,
Heerschap A. Optimal timing for in vivo
1H-MR spectroscopic imaging of the
human prostate at 3T. Magn Reson Med
2005;53(6):1268-1274.
11 Shukla-Dave A, Hricak H, Moskowitz C,
Ishill N, Akin O, Kuroiwa K, Spector J,
Kumar M, Reuter VE, Koutcher JA, Zakian
KL. Detection of prostate cancer with
MR spectroscopic imaging: an expanded
paradigm incorporating polyamines.
Radiology 2007;245(2):499-506.
12 Garcia-Martin ML, Adrados M, Ortega MP,
Fernandez Gonzalez I, Lopez-Larrubia P,
Viano J, Garcia-Segura JM. Quantitative
(1) H MR spectroscopic imaging of the
prostate gland using LCModel and a
dedicated basis-set: correlation with
histologic findings. Magn Reson Med;
65(2):329-339.
13 Heerschap A, Jager GJ, van der Graaf M,
Barentsz JO, de la Rosette JJ,
Oosterhof GO, Ruijter ET, Ruijs SH.
In vivo proton MR spectroscopy reveals
altered metabolite content in malignant
prostate tissue. Anticancer research
1997; 17(3A): 1455-1460.
14 Jung JA, Coakley FV, Vigneron DB,
Swanson MG, Qayyum A, Weinberg V,
Jones KD, Carroll PR, Kurhanewicz J.
Prostate depiction at endorectal MR
spectroscopic imaging: investigation of
a standardized evaluation system.
Radiology 2004;233(3):701-708.
15 Futterer JJ, Scheenen TW, Heijmink SW,
Huisman HJ, Hulsbergen-Van de Kaa CA,
Witjes JA, Heerschap A, Barentsz JO.
Standardized threshold approach using
three-dimensional proton magnetic
resonance spectroscopic imaging in
prostate cancer localization of the entire
prostate. Investigative radiology
2007;42(2):116-122.
16 Kobus T, Hambrock T, Hulsbergen-van de
Kaa CA, Wright AJ, Barentsz JO, Heerschap
A, Scheenen TW. In vivo assessment of
prostate cancer aggressiveness using
magnetic resonance spectroscopic imaging
at 3 T with an endorectal coil. European
urology;60(5):1074-1080.
17 Scheenen TWJ, Fütterer J, Weiland E and
others. Discriminating cancer from
noncancer tissue in the prostate by
3-dimensional proton magnetic resonance
spectroscopic imaging: A prospective
multicenter validation study. Invest Radiol
2011;46(1):25-33.
18 Scheenen TW, Heijmink SW, Roell SA,
Hulsbergen-Van de Kaa CA, Knipscheer
BC, Witjes JA, Barentsz JO, Heerschap A.
Three-dimensional proton MR spectroscopy
of human prostate at 3 T without
endorectal coil: feasibility. Radiology
2007;245(2):507-516.
19 Kobus T, Wright AJ, Scheenen TW,
Heerschap A. Mapping of prostate cancer
by 1H MRSI. NMR Biomed. 2013 Jun 13.
doi: 10.1002/nbm.2973. [Epub ahead of
print] PMID:23761200.
ratio is actually calculated and are,
therefore, often not directly comparable.
However, using the Siemens-supplied
default protocols for acquisition
and syngo.via postprocessing
enables one to make use of published
values as given in table 1, and incorporate
1H MRSI of the prostate into
their clinical routine.
Contact
Tom Scheenen, Ph.D.
Radboud University Nijmegen
Medical Centre
Radiology Department
P.O. Box 9102
6500 HC Nijmegen
The Netherlands
Tom.Scheenen@radboudumc.nl
Alan Wright Arend Heerschap Thiele Kobus Tom Scheenen
22 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
Get your
free copy
of the
PI-RADS
Scoring
Image Atlas
Visit us at
www.siemens.com/
magnetom-world
Go to > Publications >
Subscriptions > MRI Poster
Clinical Men’s Health
PI-RADS Clinical Men’s Health
Classification:
Structured Reporting for MRI of the Prostate
PI-RADS Clinical Men’s Health
Classification:
Structured Reporting for MRI of the Prostate
PI-RADS Classification:
Structured Reporting for MRI of the Prostate
M. Röthke1; D. Blondin2; H.-P. Schlemmer1; T. Franiel3
1 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany
2 Department M. Röthke1; of Diagnostic D. and Blondin2; Interventional H.-P. Schlemmer1; Radiology, University T. Franiel3
Hospital Düsseldorf, Germany
3 Department of Radiology, Charité Campus Mitte, Medical University Berlin, Germany
1 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany
2 Department M. Röthke1; of Diagnostic D. and Blondin2; Interventional H.-P. Schlemmer1; Radiology, University T. Franiel3
Hospital Düsseldorf, Germany
3 Department of Radiology, Charité Campus Mitte, Medical University Berlin, Germany
1 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany
2 Department of Diagnostic and Interventional Radiology, University Hospital Düsseldorf, Germany
3 Department of Radiology, Charité Campus Mitte, Medical University Berlin, Germany
Introduction
Prostate MRI has become an increas-ingly
common adjunctive procedure in
a structured reporting scheme
(PI-RADS) based on the BI-RADS classi-fication
Introduction
Prostate MRI has become an increas-ingly
the detection of prostate cancer. In
Germany, it is mainly used in patients
with prior negative biopsies and/or
abnormal or increasing PSA levels.
The procedure of choice is multipara-metric
based on a Likert scale with scores
ranging from 1 to 5. However, it lacks
illustration of the individual manifes-tations
common adjunctive procedure in
the detection of prostate cancer. In
Germany, it is mainly used in patients
with prior negative biopsies and/or
abnormal or increasing PSA levels.
The procedure of choice is multipara-metric
MRI, a combination of high-resolution
T2-weighted (T2w) mor-phological
sequences and the
uniform instructions for aggregated
scoring of the individual submodali-ties.
MRI, a combination of high-resolution
multiparametric techniques of diffu-sion-
classification in daily routine difficult,
especially for radiologists who are
less experienced in prostate MRI. It is
therefore the aim of this paper to
concretize the PI-RADS model for the
detection of prostate cancer using
representative images for the relevant
scores, and to add a scoring table that
combines the aggregated multipara-metric
T2-weighted (T2w) mor-phological
sequences and the
weighted MRI (DWI), dynamic
contrast-enhanced MRI (DCE-MRI),
and proton MR spectroscopy (1H-MRS)
[1, 2]. Previously, there were no uni-form
multiparametric techniques of diffu-sion-
weighted MRI (DWI), dynamic
contrast-enhanced MRI (DCE-MRI),
and proton MR spectroscopy (1H-MRS)
[1, 2]. Previously, there were no uni-form
recommendations in the form
of guidelines for the implementation
and standardized communication of
findings. To improve the quality of
the procedure and reporting, a group
of experts of the European Society
of Urogenital Radiology (ESUR) has
recently published a guideline for MRI
of the prostate [3]. In addition to pro-viding
recommendations in the form
This makes use of the PI-RADS
of guidelines for the implementation
and standardized communication of
findings. To improve the quality of
the procedure and reporting, a group
of experts of the European Society
of Urogenital Radiology (ESUR) has
recently published a guideline for MRI
of the prostate [3]. In addition to pro-viding
recommendations relating to
indications and minimum standards for
MR protocols, the guideline describes
according to the Likert scale. In addi-tion,
reporting scheme is presented, which
enables accurate communication of
the findings to the urologist. Further-more,
techniques are described and critically
recommendations relating to
indications and minimum standards for
MR protocols, the guideline describes
for breast imaging. This is
assessed in terms of their advantages
and disadvantages.
a structured reporting scheme
(PI-RADS) based on the BI-RADS classi-fication
Materials and methods
The fundamentals of technical imple-mentation
for breast imaging. This is
based on a Likert scale with scores
ranging from 1 to 5. However, it lacks
illustration of the individual manifes-tations
and their criteria as well as
selected by the authors by consensus
on the basis of representative image
findings from the 3 institutions. The
scoring intervals for the aggregated
PI-RADS score were also determined by
consensus. The individual imaging
aspects were described and evaluated
with reference to current literature
by one author in each case (T2w: M.R.,
DCE-MRI: T.F., DWI: D.B., MRS: H.S.).
Furthermore, a graphic reporting
scheme that allows the findings to be
documented in terms of localization
and classification was developed,
taking into account the consensus
paper on MRI of the prostate published
in 2011 [4].
and their criteria as well as
uniform instructions for aggregated
scoring of the individual submodali-ties.
This makes use of the PI-RADS
classification in daily routine difficult,
especially for radiologists who are
less experienced in prostate MRI. It is
therefore the aim of this paper to
concretize the PI-RADS model for the
detection of prostate cancer using
representative images for the relevant
scores, and to add a scoring table that
combines the aggregated multipara-metric
scores to a total PI-RADS score
a standardized graphic prostate
scores to a total PI-RADS score
according to the Likert scale. In addi-tion,
a standardized graphic prostate
reporting scheme is presented, which
enables accurate communication of
the findings to the urologist. Further-more,
the individual multiparametric
the individual multiparametric
techniques are described and critically
assessed in terms of their advantages
and disadvantages.
were determined by con-sensus.
Materials and methods
The fundamentals of technical imple-mentation
The sample images were
were determined by con-sensus.
The sample images were
selected by the authors by consensus
on the basis of representative image
findings from the 3 institutions. The
scoring intervals for the aggregated
PI-RADS score were also determined by
consensus. The individual imaging
aspects were described and evaluated
with reference to current literature
by one author in each case (T2w: M.R.,
DCE-MRI: T.F., DWI: D.B., MRS: H.S.).
Furthermore, a graphic reporting
scheme that allows the findings to be
documented in terms of localization
and classification was developed,
taking into account the consensus
paper on MRI of the prostate published
in 2011 [4].
1
I: Normal PZ in T2w
hyperintense
II: Hypointense
discrete focal lesion
(wedge or band-shaped,
I: Normal PZ in T2w
hyperintense
ill-defined)
III: Changes not
falling into categories
1+2 & 4+5
II: Hypointense
discrete focal lesion
(wedge or band-shaped,
ill-defined)
IV: Severely hypo-intense
round-shaped, well-defined
III: Changes not
falling into categories
1+2 & 4+5
focal lesion,
without extra-capsular
extension
IV: Severely hypo-intense
focal lesion,
round-shaped, well-defined
V: Hypointense mass,
round and bulging,
with capsular
involvement or seminal
vesicle invasion
without extra-capsular
extension
V: Hypointense mass,
round and bulging,
with capsular
involvement or seminal
vesicle invasion
1 PI-RADS classification of T2w: peripheral glandular sections.
30 MAGNETOM Flash | 4/2013 | www.siemens.com/magnetom-world
1604_MAGNETOM_Flash_54_ASTRO_Inhalt.indd 30 09.09.13 16:20
Read the comprehensive article
“PI-RADS Classification:
Structured Reporting
for MRI of the Prostate”
by Matthias Röthke et al.
in MAGNETOM Flash 4/2013
page 30-38.
Available for download at
www.siemens.com/
magnetom-world
I Choline Citrate
II Choline Citrate
III Choline Citrate
IV Choline Citrate
V Choline Citrate
www.siemens.com/magnetom-world
PI-RADS SCORING Image Atlas
Prostate MRI
Answers for life.
DCE type 1 curve = 1 point
I: Cho << Citrate
II: Hypointense dull focal
lesion (wedge or band-shaped,
ill-defined)
III: Changes not falling into
categories 1+2 or 4+5
IV: Hypointense focal lesion
without extracapsular
extension or bulging
V: Hypointense mass with
extracapsular extension or
bulging
I: Normal, hyperintense PZ
in T2w, peripheral glandular
lesions
II: Hypointense lesion with
well-defined capsule; band-shaped
hypointense regions
III: Changes not falling into
categories 1+2 or 4+5
IV: Hypointense lesion without
capsular involvement with
ill-defined margins, “erased
charcoal sign”
V: Oval-shaped mass with
capsular involvement;
infiltrating mass with invasion
into anterior structures
I: Stromal & glandular
hyperplasia without focal
hypointense lesions T2w,
central glandular lesions
II: Diffuse hyperintensity on
DWI b ≥ 800 image, no focal
ADC reduction
III: Changes not falling into
categories 1+2 or 4+5
IV: Focal area with reduced
ADC but isointense SI on
DWI b ≥ 800 image
V: Focal area with reduced
ADC and hyperintense SI on
DWI b ≥ 800 image
I: No reduction in ADC
and no increase in SI on
DWI b ≥ 800 images
2 points = probably benign 3 points = indeterminate 4 points = probably malignant
5 points = highly suspicious of
1 point = most probably benign malignancy
DCE type 2 curve = 2 points DCE type 3 curve = 3 points
DCE-MRI – asymmetric, non-focal:
+ 1 point
DCE-MRI – asymmetric, unusual
location: + 2 points
DCE-MRI – asymmetric, focal
location: + 2 points
DCE-MRI – symmetric, non-focal:
+ 0 points
II: Cho << Citrate
III: Cho = Citrate
IV: Cho > Citrate
V: Cho >> Citrate
For details please refer to: M. Röthke, D. Blondin, H.-P. Schlemmer, T. Franiel: “PI-RADS Classification:
Structured Reporting for MRI of the Prostate”, MAGNETOM Flash issue 4/2013, ASTRO edition, page 30-38.
T2w,
peripheral
glandular
lesions
T2w,
central
glandular
lesions
DWI b ≥ 800
ADC
DCE
time curve /
parametric
color map
1H-MRS
Introduction
Prostate MRI has become an increas-ingly
common adjunctive procedure in
the detection of prostate cancer. In
Germany, it is mainly used in patients
with prior negative biopsies and/or
abnormal or increasing PSA levels.
The procedure of choice is multipara-metric
MRI, a combination of high-resolution
T2-weighted (T2w) mor-phological
sequences and the
multiparametric techniques of diffu-sion-
weighted MRI (DWI), dynamic
contrast-enhanced MRI (DCE-MRI),
and proton MR spectroscopy (1H-MRS)
[1, 2]. Previously, there were no uni-form
recommendations in the form
of guidelines for the implementation
and standardized communication of
findings. To improve the quality of
the procedure and reporting, a group
of experts of the European Society
of Urogenital Radiology (ESUR) has
recently published a guideline for MRI
of the prostate [3]. In addition to pro-viding
recommendations relating to
indications and minimum standards for
MR protocols, the guideline describes
a structured reporting scheme
(PI-RADS) based on the BI-RADS classi-fication
for breast imaging. This is
based on a Likert scale with scores
ranging from 1 to 5. However, it lacks
illustration of the individual manifes-tations
and their criteria as well as
uniform instructions for aggregated
scoring of the individual submodali-ties.
This makes use of the PI-RADS
classification in daily routine difficult,
especially for radiologists who are
less experienced in prostate MRI. It is
therefore the aim of this paper to
concretize the PI-RADS model for the
detection of prostate cancer using
representative images for the relevant
scores, and to add a scoring table that
combines the aggregated multipara-metric
scores to a total PI-RADS score
according to the Likert scale. In addi-tion,
a standardized graphic prostate
reporting scheme is presented, which
enables accurate communication of
the findings to the urologist. Further-more,
the individual multiparametric
techniques are described and critically
assessed in terms of their advantages
and disadvantages.
Materials and methods
The fundamentals of technical imple-mentation
were determined by con-sensus.
The sample images were
selected by the authors by consensus
on the basis of representative image
findings from the 3 institutions. The
scoring intervals for the aggregated
PI-RADS score were also determined by
consensus. The individual imaging
aspects were described and evaluated
with reference to current literature
by one author in each case (T2w: M.R.,
DCE-MRI: T.F., DWI: D.B., MRS: H.S.).
Furthermore, a graphic reporting
scheme that allows the findings to be
documented in terms of localization
and classification was developed,
taking into account the consensus
paper on MRI of the prostate published
in 2011 [4].
I: Normal PZ in T2w
hyperintense
II: Hypointense
discrete focal lesion
(wedge or band-shaped,
ill-defined)
III: Changes not
falling into categories
1+2 & 4+5
IV: Severely hypo-intense
focal lesion,
round-shaped, well-defined
without extra-capsular
extension
V: Hypointense mass,
round and bulging,
with capsular
involvement or seminal
vesicle invasion
1
1 PI-RADS classification of T2w: peripheral glandular sections.
30 MAGNETOM Flash | 4/2013 | www.siemens.com/magnetom-world
1604_MAGNETOM_Flash_54_ASTRO_Inhalt.indd 30 09.09.13 16:20
1
1 PI-RADS classification of T2w: peripheral glandular sections.
30 MAGNETOM Flash | 4/2013 | www.siemens.com/magnetom-world
1604_MAGNETOM_Flash_54_ASTRO_Inhalt.indd 30 09.09.13 16:20](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-12-2048.jpg)
![Evaluation of the CIVCO Indexed Patient
Position System (IPPS) MRI-Overlay
for Positioning and Immobilization of
Radiotherapy Patients
Th. Koch1; K. Freundl1; M. Lenhart2; G. Klautke1; H.-J. Thiel1
1 Klinik und Praxis für Strahlentherapie und Radioonkologie, Sozialstiftung Bamberg, Germany
2 Klinik für Diagnostische Radiologie, Interventionelle Radiologie und Neuroradiologie, Bamberg, Germany
Abstract
The emerging development in
modern radiotherapy planning (RTP)
requires sophisticated imaging modalities.
RTP for high precision requires
exact delineation of the tumor, but
this is currently the weakest link in
the whole RTP process [1]. Therefore
Magnetic resonance imaging (MRI) is
of increasing interest in radiotherapy
treatment planning because it has
a superior soft tissue contrast, making
it possible to define tumors and surrounding
healthy organs with greater
accuracy. The way to use MRI in radiotherapy
can be different.
The MRI
datasets can be used as secondary
images to support the tumor delineation.
This is routinely in use in many
radiotherapy departments. Two other
methods of MRI guidance in the RTP
process are until now only research
3 One index bar is latched to the
accuracy strongly depends on the
MRI scan position. Hanvey et al. [3]
and Brunt et al. [4] have shown that it
is indispensable for the MRI dataset
to be created in the treatment position
which is primarily defined by the CT
scan.
The MRI dataset can also feasibily
be used as the only dataset. Because
of the lack of electron density information,
which is required for dosimetric
calculations, bulk densities have
to be applied to the MRI images. For
this purpose the different anatomic
regions like bone, lung, air cavities
and soft tissue have to be overwritten
with the physical densities. With this
method it is possible to achieve dose
calculation results quite similar to the
calculation in the CT dataset in the
head and neck region [5, 6] as well
as in the pelvic region [7]. The advantage
of this method is that by avoiding
the CT scan you save some time
and money. In this case it is necessary
for the treatment position to be determined
during the MRI scan, hence the
MRI scanner has to be equipped with
the same positioning and immobilization
tools as the Linac. Further problems
to overcome are the evaluation
and correction of possible image distortions
and the determination of
accurate bulk densities.
After the RTP process there are a
lot of remaining uncertainties such
as set-up errors, motion of the target
structures and during the treatment
changes of the tumor volume and
shrinking. This problem can be overcome
with the so-called image-guided
radiotherapy (IGRT). IGRT involves
a periodical verification (weekly or
more frequent) of tumor position and
size with appropriate imaging systems.
It is evident that IGRT is only as
good as the accuracy with which
the target structures can be defined.
For this reason some groups try to
develop hybrid systems, where a Linac
or a cobalt treatment unit is combined
with an MRI scanner for a so-called
MR-guided radiotherapy [8-10].
Again: MR-guided radiotherapy can
only be successful when the reference
MRI dataset has been created in the
treatment position.
In any of the above three cases, where
MRI can be helpful to improve the
accuracy of radiotherapy, it is strongly
advised that one has a robust and
reproducible patient positioning and
immobilization system, mainly at the
MRI scanner, which is used for MR-guided
RTP. Siemens provides with
the CIVCO IPPS MRI-Overlay a suitable
solution. In our clinic we have introduced
and tested this MRI-
overlay,
especially for patients with tumors in
the pelvis and for brain tumors and
metastasis.
Method
Our 1.5T MAGNETOM Aera system
(Siemens Healthcare, Erlangen, Germany)
is located in the radiology
department and can temporarily be
used by the staff of the radiotherapy
department. For the purpose of MR-guided
RTP we have equipped the
MAGNETOM Aera with the CIVCO IPPS
MRI-Overlay. This overlay enables
the fixation of positioning and immobilization
tools necessary for radioprojects,
but interest in them is
increasing. The first method is to use
MRI data as the primary and only
image dataset and the second is the
application of the MRI data as reference
dataset for a so-called ’MRI-guided
radiotherapy in hybrid systems’
(Linear Accelerator (Linac) or Cobalt
RT units combined with MRI). For
all cases it is essential to create the
MRI datasets in the radiotherapy treatment
position. For this reason the
CIVCO Indexed Patient Positioning
System (IPPS) MRI-Overlay was introduced
and tested with our Siemens
MAGNETOM Aera MRI Scanner.
Introduction
Although computed tomography (CT)
images are the current gold standard
in radiotherapy planning, MRI
becomes more and more interesting.
Whilst CT has limitations in accuracy
concerning the visualization of boundaries
between tumor and surrounding
healthy organs, MRI can overcome
these problems by yielding superior
soft tissue contrast. Currently there are
three different possible strategies by
which MRI can help to improve radiotherapy
treatment planning:
The MRI datasets can be used as
secondary
images for treatment planning.
These MR images can be used
to delineate the tumor and the
surrounding
organs, whilst the CT
images – the primary planning data –
are necessary to calculate the 3D dose
distribution. The two image datasets
have to be co-registered thoroughly
to ensure that the anatomy correlates
(see for example [2]). The registration
therapy treatments. For our purpose
we have used an MR compatible mask
system for head and neck cases and
vacuum cushions for patients with
diseases in the pelvic region both from
Medical Intelligence (Elekta, Schwabmünchen,
Germany). These tools can
all be fixed with so-called index bars
(Figs. 4, 12) at the MRI-
Overlay. These
index bars are custom designed for
our purpose by Innovative Technologies
Völp (IT-V, Innsbruck, Austria) for
the MRI-Overlay and for use in the
high field magnetic environment. For
the correct positioning of the patients,
the laser system Dorado 3 (LAP,
Lüneburg,
Germany) was additionally
installed in the MRI room. The preliminary
modifications and the patient
positioning is described in the following
for two cases.
The first case describes the procedure
for a patient with a head tumor. The
first step is the removal of the standard
cushion of the MRI couch and the
mounting of the MRI-Overlay (Figs.
1–3). One index bar is necessary to
fix the mask system on the overlay
(Figs. 4, 5) to avoid movements and
rotations during the scan. Because
the standard head coil set cannot be
used with the mask system, two flex
coils (Flex4 Large) have to be prepared
(Figs. 6–8). In figure 8 one can see,
that the correct head angle could be
adjusted. Now the patient is placed
on the overlay and in the mask system.
The patient’s head can be immobilized
with the real and proper mask
made from thermoplastic material
called iCAST (Medical Intelligence,
Elekta, Schwabmünchen, Germany)
1 After the removal of the standard cushion the CIVCO
1.5T MAGNETOM Aera with the standard cushion on
the MRI couch.
IPPS MRI-Overlay can be mounted.
2
1 2
3
The lines indicate the position for
the index bars.
MRI-Overlay.
4
4
The mask system for head and neck fits
to the index bar to avoid movement.
5
5
Radiation Oncology Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 25
Clinical Radiation Oncology
24 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-13-2048.jpg)
![Technology Technology
2A
2D
2B
reference TSE qTSE qTSE-G
2E
2C
2F
2
Representative
slices from 2
patients acquired
with the reference
TSE sequence (left
column), the qTSE
sequence (center
column) and the
qTSE-G sequence
(right column).
Making MRI Scanning Quieter: Optimized TSE
Sequences with Parallel Imaging
Eric Y. Pierre1; David Grodzki2; Bjoern Heismann2; Gunhild Aandal3, 5; Vikas Gulani1, 3; Jeffrey Sunshine3;
Mark Schluchter4; Kecheng Liu6; Mark A. Griswold1, 3
1 Biomedical Engineering, Case Western Reserve University, Cleveland, Ohio, USA
2 Siemens Healthcare, Erlangen, Germany
3 Radiology, Case Western Reserve University, Cleveland, Ohio, USA
4 Division of Biostatistics, Case Western Reserve Univeristy, Cleveland, Ohio, USA
5 Haraldsplass Deaconess Hospital, Bergen, Norway
6 Siemens Medical Solutions, USA Inc., Malvern, Pennsylvania, USA
Introduction
Turbo Spin-Echo sequences at 1.5T
can generate noise at over 100dBA
inside the bore [1–3]. This noise is
equivalent to standing 5 meters away
from a jackhammer [3], and would
be even louder on higher field systems.
Despite the use of ear-protective
equipment, reducing the Sound
Pressure Level (SPL) generated by
these standard clinical sequences
could noticeably improve patient
comfort [4]. MRI pulse sequences
mostly generate acoustic noise
because of rapidly varying gradient
waveforms: The resulting Lorentz
forces applied on the gradient coils
make the entire scanner structure
vibrate [5]. To circumvent this issue,
several hardware solutions have been
proposed. For example, the whole
gradient coil can be enclosed in a
vacuum chamber [6–8], or gradient
field rotation can be performed
mechanically [9]. While these solutions
achieve significant noise reduction
for all types of sequences, they
can noticeably increase manufacturing
cost, and can even decrease
gradient
efficiency. Mechanical and
acoustic balanced designs of gradient
coil systems including windings performing
active acoustic control have
also been considered and investigated
[10, 11].
of acoustic noise in Echo Planar
Imaging
(EPI) [14, 15]. The reduction
was achieved by counterbalancing
lengthened gradient waveforms with
increased acquisition speed, thereby
reducing acoustic noise without
increasing acquisition time while maintaining
inter-echo spacing, only at
cost of signal-to-noise ratio (SNR).
By extending such principles to other
generally-used standard clinical MR
sequences, this article demonstrates
that with minor SNR reductions (≤ 10%),
effective reduction in acoustic noise
can be further achieved without
noticeable degrade of diagnostic information
or imaging time, as well as
without sacrificing gradient efficiency.
Two types of modifications in a
T2-weighted Turbo Spin-Echo (TSE)
sequence were investigated for acoustic
noise reduction: First by solely modifying
the gradient waveforms and second
by additionally using GRAPPA at
a reduction factor of two (R=2)*. Comparative
SPL measurements at the bore
were performed between standard
TSE, quiet TSE (qTSE)* and quiet TSE
with GRAPPA (qTSE-G)*. A statistical
analysis of comparative scores from
a reader’s study was conducted.
Methods
The gradient waveforms of the TSE
sequence were optimized with an
automatic gradient optimization
algorithm that extends any slope duration
to its maximum and reduces the
number of slopes to their minimum.
For instance, with minor changes in
protocols, spoiling and crusher gradient
lobes are replaced by long rising
or descending slopes, while maintaining
the crusher moment unchanged.
To keep the same total acquisition
time, the reduction of the gradient
slew rate is constrained by the fixed
inter-echo spacing. The decreased
slew rate of readout gradient will
slightly reduce readout sampling time
(Fig. 1). In consequence, the readout
bandwidth (BW) increases slightly,
with a tradeoff between reduction of
SPL and SNR loss.
In addition, parallel acquisition could
be further employed to reduce the
echo-train length, i.e. number of
echoes per train, by a factor of R.
Keeping the acquisition time constant,
the inter-echo spacing can be
extended by R, allowing further
stretching of the gradient moments.
This effectively represents a benefit
of parallel imaging acceleration in
acoustic noise reduction rather than
imaging time reduction.
The acquisition protocols changes
are as follows: The readout BW was
increased by about 10%, from
107 Hz/pixel in the standard protocol
to 125 Hz/pixel. The effective TR/TE
were increased from 5000/93 ms to
5180/85 ms, which resulted in only
a 3 second increase in acquisition
time, from 1:37 min to 1:40 min. The
qTSE-G parameters were identical to
the qTSE protocol, but with use of
GRAPPA with R=2. For both qTSE and
qTSE-G protocols, and the gradient
slopes were maximally stretched as
illustrated in figure 1.
Modifying and/or optimizing pulse
sequences can also reduce acoustic
noise effectively. One such solution is
to time the ramping up and ramping
down of the gradient waveforms so
that the induced scanner vibrations
cancel each other out [12]. Another
approach is to use lower gradient
amplitude and slew rates of the gradient
waveforms [13]. By low-pass filtering
the gradient, vibration frequencies
for which the acoustic response of the
gradient coil is high can be avoided.
Elaborate redesigns of gradient waveforms
coupled with parallel imaging
have demonstrated further reduction
1
Comparison of
conventional
sequence (dashed
lines) with quiet
sequence (solid
lines). The reduction
of ADC represents
a BW increase.
1
ADC RF /
ADC
GSlice
GRead
* WIP, the product is currently under
development and is not for sale in the US
and other countries. Its future availability
cannot be ensured.
30 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 31](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-16-2048.jpg)
![Technology Technology
Table 1: Comparison of dBA values
Sequence type Standard TSE qTSE qTSE-G Background
LAEQ (30 sec
average)
92.5 81.3 72.7 53.0
Max Peak 102.8 95.8 92.0 77.7
Comparison of dBA values for standard TSE, qTSE, qTSE-G sequences, and measured background noise. Measurements were performed
inside the bore at patient head position using a 2238 Mediator sound level meter (Brueel & Kjaer GmbH, Bremen, Germany).
Table 2: Ratings by readers
Sequence type
All techniques compared to
themselves
Reader #1
0.35 ± 0.40 (0.06, 0.64)
p = 0.02
Reader #2
-0.03 ± 0.11 (-0.11, 0.04)
p = 0.34
Reader #3
Average
0.11 ± 0.14 (0.01, 0.21)
p = 0.04
Mean and standard deviation, 95% confidence interval, and p-value of the scores given by each radiologist to the different types of
image volume pairs after self-bias correction. Positive score show preference of the right volume over the left volume, on a -10 to +10 scale.
In-vivo studies were performed on
a 3T MAGNETOM Verio MRI scanner
(Siemens Healthcare, Erlangen,
Germany)
with a 12-channel head
coil with patients admitted for head
examination. Informed consent was
obtained from the volunteer before
the start of the study in accordance
with IRB protocol. A total of 10 different
patient scannings were performed,
each comparing standard TSE images
with qTSE and qTSE-G images. The
image resolution (192 × 256 matrix),
number of slices (26), slice thickness
(5 mm) and slice orientation were
kept identical throughout the 3 different
acquisitions.
0 ± 0
–
To measure acoustic noise level LAEQ
(Equivalent Continuous Sound Level
in A-weighting) with 30 seconds average
and peak values, a professional
device, 2238 Mediator sound level
meter (Brueel & Kjaer GmbH, Bremen,
Germany), was used, which was placed
inside the bore at patient head position.
qTSE : TSE qTSE-G : TSE
-0.20 ± 0.26 (-0.38, -0.02)
p = 0.04
1.30 ± 1.96 (-0.10, 2.70)
p = 0.07
0.73 ± 1.59 (-0.41, 1.86)
p = 0.18
0.61 ± 1.17 (-0.23, 1.45)
p = 0.13
The background noise is mainly
generated by the cold-head pump and
the ventilation among other sources.
To evaluate the image quality, a total
of 7 image-volume-pairs were assembled
from each of the 10 patient
datasets. The first 2 pairs compared
qTSE with TSE volumes, alternatively
with qTSE on the left and TSE on
the right. Similarly, another 2 pairs
compared qTSE-G with TSE volumes
in both left-right orders randomly.
Finally, 3 pairs were assembled with
the same volume on the left and
right, which consist of TSE vs. TSE,
qTSE vs. TSE, and qTSE-G vs. qTSE-G
volumes, respectively.
All 70 volume pairs were presented
in the same random order to 3 trained
radiologists blinded to the acquisition
technique, who were asked the following
question: “On a scale from -10
to +10, how much better is the image
quality of the volume on the right
compared to the volume on the left,
0.20 ± 0.59 (-0.22, 0.62)
p = 0.31
3.95 ± 0.86 (3.33, 4.57)
p < 0.0001
3.08 ± 1.25 (2.18, 3.97)
p < 0.0001
2.41 ± 0.80 (1.83, 2.98)
p < 0.0001
with a positive score indicating
superiority
of the right volume, and
0 representing no difference in quality
between left and right?”. The graphical
user interface used for the reading
allowed user-navigation through the
paired-volume slices, and simultaneous
image windowing of the 2 displayed
images.
To avoid possible left-right bias, the
average of the qTSE vs. TSE score and
the TSE vs. qTSE score multiplied by -1
was then calculated for each reader’s
reading on each patient. The average
of the corrected scores across readers
was then computed for each patient.
Corrected scores were calculated in
the same way for the qTSE-G vs. TSE
comparison. One-sample t-tests were
used to test whether the mean average
reader scores differed from zero,
and 95% confidence intervals (CI) for
the mean scores were also calculated.
One-sample t-tests and CI were also
carried out using each reader’s scores
10
8
6
4
2
0
-2
-4
-6
-8
-10
preference for standard
A B C
preference for quiet
3
separately. A reader’s average rating
of these three self-comparisons using
images from each patient were averaged,
and then the three reader averages
were averaged for each patient.
A t-test was used to test whether the
average of the reader ratings across
patients differed from zero. One-sample
t-tests and CI were also carried out for
each reader separately.
Results
The respective average and peak SPL
in [dBA] measurements for standard
TSE, qTSE and qTSE-G protocols are
listed in table 1. The achieved reduction
of average SPL for qTSE and qTSE-G
were 10 dBA and near 20 dBA (30 seconds
average), respectively.
Discussion
Optimizing the gradient waveforms
alone with a 10% increase in bandwidth
achieves an 11 dBA SPL reduction
(Table 1), with little cost to image quality
(Fig. 3). These results are in accordance
with [16] though here the measurements
were made directly at the
bore. This cost might be more noticeable
with lower SNR systems, however
in this configuration, no statistically
significant difference in image quality
was observed (Table 2), making gradient
redesign a viable solution to make
TSE sequences quieter.
With additional use of Parallel Imaging,
the modified quiet TSE sequence
allows on average a 20 dBA reduction
in SPL (Table 1). The modified sequence
had an effect on in image quality
A. self image comparison (all methods)
(p < 0.05)
B. qTSE vs. standard TSE
Non significant difference for all 3 readers
(p = 0.20, p = 0.06, p = 0.18)
C. qTSE+GRAPPA vs. standard TSE
(p < 0.0001)
(Fig. 3): The average preference
score across readers for standard TSE
images over qTSE-G images was
+2.41 (p<0.0001, Table 2), and the
95% confidence interval places its
true value between +1.8 and +3.
However it should be noted that this
change in image quality is to be
expected as Parallel Imaging was
used. In compensation, the reduction
of acoustic noise was highly effective:
the SPL at the bore of the standard
TSE sequence was 39.5 dBA
higher than the background noise,
compared to 19.7 dBA for the modified
sequence.
Conclusion
In comparison with standard MR
sequences, gradient wave modifications
in TSE sequence coupled with
Parallel Imaging can achieve over a
factor 10 of acoustic noise reduction,
yielding an improved patient comfort
with nearly identical diagnostic information
and imaging time. Without
any hardware modifications or
upgrade, both proposals described
in this article, qTSE and qTSE-G, can
be easily implemented on a conventional
MRI system for routine clinical
applications. In addition, scanning
on a high field system with multiple
channel coils, such as the 32-channel
head coil, provides more flexibility
to make MRI scanning quieter.
* Work in progress: The product is still under
development and not commercially
available yet. Its future availability cannot
be ensured.
3
(A) 95% confidence
intervals for averages
scores by readers for
volumes compared to
themselves; (B) qTSE vs.
standard TSE; and (C)
qTSE-G vs. standard TSE.
Positive scores show
preference for standard
TSE in the last two cases.
References
1 Shellock FG, Morisoli SM, Ziarati M.
Measurement of acoustic noise during
MR imaging: evaluation of six “worst-case”
pulse sequences. Radiology
1994;191:91–93.
2 McJury M, Blug A, Joerger C, Condon B,
Wyper D. Acoustic noise levels during
magnetic resonance imaging scanning
at 1.5 T. Br J Radiol 1994;67:413–415.
3 McJury M. Acoustic noise levels
generated during high field MR imaging.
Clin Radiol 1995;50:331–334.
4 Quirk ME, Letendre AJ, Ciottone RA,
Lingley JF. Anxiety in patients undergoing
MR imaging. Radiology
1989;170:463–466.
5 Hedeen R, Edelstein W. Characterization
and prediction of gradient acoustic noise
in MR imagers. Magn Reson Med
2005;37:7–10.
6 Katsunuma A, Takamori H, Sakakura Y,
Hamamura Y, Ogo Y, Katayama R. Quiet
MRI with novel acoustic noise reduction.
MAGMA 2002;13:139–44.
7 Edelstein WA, Hedeen RA, Mallozzi RP,
El-Hamamsy SA, Ackermann RA, Havens
TJ. Making MRI quieter. Magn Reson
Imaging 2002;20:155–63.
8 Edelstein WA, Kidane TK, Taracila V, Baig
TN, Eagan TP, Cheng Y-CN, Brown RW,
Mallick J a. Active-passive gradient
shielding for MRI acoustic noise reduction.
Magn Reson Med 2005;53:1013–7.
9 Cho ZH, Chung ST, Chung JY, Park SH,
Kim JS, Moon CH, Hong IK. A new silent
magnetic resonance imaging using a
rotating DC gradient. Magn Reson Med
1998;39:317–21.
10 Mansfield P, Haywood B. Principles of
active acoustic control in gradient coil
design. MAGMA 2000;10:147–51.
11 Haywood B, Chapman B, Mansfield P.
Model gradient coil employing active
acoustic control for MRI. MAGMA
2007;20:223–31.
32 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 33](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-17-2048.jpg)
![Technology
Quiet T1-weighted 3D Imaging
of the Central Nervous System Using PETRA
Masahiro Ida, M.D.1; Matthew Nielsen, M.A.2
1 Dept. of Radiology, Tokyo Metropolitan Ebara Hospital, Tokyo, Japan
2 Research & Collaboration Dept., Healthcare Sector, Siemens Japan K.K., Tokyo, Japan
Introduction
Nearly all MRI sequences in routine
clinical use employ rapidly varying
magnetic field gradients that generate
considerable acoustic noise, one of the
primary causes of patient discomfort
and restlessness [1]. Eliminating such
noise would provide additional comfort
for all patients, and may provide
particular advantages for patients
with pediatric*, dementia and certain
psychiatric diseases who tend to have
difficulty relaxing or remaining still
during MR examinations.
Ultra-short echo time sequences such
as zero-TE [2], SWIFT [3] and PETRA [4]
require only limited gradient activity
and allow for inaudible 3D scanning.
However, due to ultra-short TEs, the
image contrast is given by the steady
state and is limited to the range of
PD- to T1-weighting unless pre-pulses
are used [5]. Similar to the MPRAGE
sequence [6-8], stronger T1-weighting
can be generated by applying an
inversion pre-pulse before every nth
repetition in the PETRA* sequence.
A study has shown that this quiet
inversion-prepared PETRA sequence
is capable of T1-weighting com-parable
to that of MPRAGE when
measured in the same time and
with the same spatial resolution [1].
In this article, examples of quiet
inversion-prepared PETRA images are
compared with conventional 3D
T1-weighted images (MPRAGE or
3D-FLASH) from the same patients.
All of the examples were obtained
during brain examinations, and all
but one of the examples employed
contrast enhancement.
* WIP, the product is currently under
development and is not for sale in the US
and other countries. Its future availability
cannot be ensured.
PETRA sequence principles
and noise reduction
In the PETRA sequence, gradients are
already on and stable at a certain
amplitude before the excitation pulse,
as shown in figure 1. At the end of
each repetition, the gradient strength
on each axis is altered only slightly
meaning that the required slew rate
is extremely low (e.g., < 5 T/m/s with
1A 1B
PETRA combines two different sequences, acquiring central k-space in a ‘point-wise’ fashion (one k-space point per repetition),
and the rest of k-space with radial trajectories. PETRA stands for Pointwise Encoding Time reduction with Radial Acquisition.
No hardware modifications or dedicated coils are needed [1].
(1A) Pulse sequence diagram for one repetition of the radial part of the PETRA sequence. Gradients are held constant during
almost an entire repetition and altered only slightly at the end of each repetition without being ramped down. This leads to
negligible deformation and vibration of the gradient coil. Thus, no acoustic noise is generated by the gradient coil. THW is the
time required to switch from transmission mode to receive mode (in the range of 10 to 100 μs on clinical scanners) [1].
(1B) During THW, a spherical volume (dots) at the center of k-space is missed by the radial part of the sequence. Each k-space
point inside that spherical volume is acquired separately in the Pointwise Encoding (PE) part of the sequence. The acquisition
time of the PE part is approximately 3 to 5% of the total measurement time [1].
1
TX
THW
RX
Tx/Rx
Gradients
TR
Technology
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 35
12 Shou X, Chen X, Derakhshan J, Eagan T,
Baig T, Shvartsman S, Duerk J, Brown R.
The suppression of selected acoustic
frequencies in MRI. Appl Acoust
2010;71:191–200.
13 Hennel F, Girard F, Loenneker T. “Silent”
MRI with soft gradient pulses.
Magn Reson Med 1999;42:6–10.
14 De Zwart J, Vangelderen P, Kellman P,
Duyn J. Reduction of Gradient Acoustic
Noise in MRI Using SENSE-EPI. Neuroimage
2002;16:1151–1155.
15 Witzel T, Wald LL. Methods for Functional
Brain Imaging. Massachusetts Institute
of Technology; 2011.
Contact
Eric Y. Pierre, Ph.D.
Department of Biomedical
Engineering
Case Western Reserve University
319 Wickenden Building
10900 Euclid Avenue
Cleveland, OH 44106-7207
USA
Phone: +1 (216)-368-4063
pierre@case.edu
16 Pierre EY, Grodzki D, Heismann B, Liu K,
Griswold MA. Reduction of Acoustic Noise
to Improve Patient Comfort Through
Optimized Sequence Design. In:
Proceedings of the 21st Annual Meeting
of ISRMRM. Vol. 42. Salt Lake City, USA;
2013. p. 256.
Eric Pierre Gunhild Aandal Vikas Gulani Jeffrey Sunshine
Mark Schluchter Kecheng Liu Mark Griswold
Answers for life.
www.siemens.com/quiet-suite
Quiet Suite
Imaging is to be seen, not heard.
Scan and
listen to
Quiet-Suite.](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-18-2048.jpg)
![PETRA) [1]. The resulting deformation
and vibration of the gradient coil is
negligible and produces almost no
audible sound. Completely unrelated
to the gradients however, transmit-mode-
to-receive-mode switching (and
vice versa) in receive-only RF coils
produces some noise [1], while PETRA
is essentially inaudible when used
with transmit-and-receive RF coils.
The acoustic noise levels generated by
PETRA and MPRAGE on a MAGNETOM
Trio A Tim System (3T) were measured
using a sound-pressure meter with
A-weighting. PETRA afforded a reduction
in acoustic noise of more than
25 dBA with both the 12-channel head
matrix coil and the 32-channel head
coil. Since both coils were receive-only,
an even greater reduction can be
expected with transmit-and-receive
coils.
PETRA versus routine
sequence image
comparisons
While MR angiography (MRA) is
undoubtedly the most commonly-used
3D sequence in brain MRI exams,
other 3D sequences are used in certain
circumstances at Tokyo Metropolitan
Ebara Hospital. The most common
one is contrast-enhanced (CE)
3D-FLASH which is employed for the
following indications because of its
high spatial resolution and short
echo time:
(1) To precisely diagnose head &
neck tumors, pre- & post-operatively
(with Quick FatSat),
(2) to inspect blood pools (AVM,
thrombosis, aneurysm, dissection),
and
(3) to detect cranial nerve
inflammation.
The second most common 3D scan
other than MRA is CE MPRAGE which
is employed to diagnose intracranial
brain tumors. Among those, the
most common indication is screening
for intracranial metastases.
MR imaging was performed on a
3 Tesla MAGNETOM Trio A Tim System.
PETRA was added to routine patient
exams that included either MPRAGE
or 3D-FLASH. The parameters of the
three sequences are shown in table 1.
The center of k-space for MPRAGE
(scan time: 5 min 56 sec) was acquired
approximately 6 to 11 minutes after
contrast media administration, while
the center of k-space for PETRA was
acquired approximately 3 minutes
later than that of MPRAGE, 9 to 14
minutes after contrast media administration.
PETRA acquires the central
portion of k-space first (in a pointwise
fashion as shown in figure
1) before
acquiring the rest of 3D k-space with
radial trajectories. A previous study
showed that for enhancing intracranial
lesions with a diameter of 5 mm or
larger, enhancement reached a plateau
in less than 10 minutes and lasted
until at least 20 minutes after contrast
media injection [9]. Thus, the 3 minute
difference in the acquisition of k = 0
between MPRAGE and PETRA would
not result in differing lesion enhancement,
and any difference in lesion
enhancement can be taken as primarily
due to sequence characteristics.
Protocols provided with the PETRA
sequence were designed to parallel the
contrast and spatial resolution (0.9
to 1.0 mm cubic voxels) typically available
with MPRAGE. Therefore, initial
work with PETRA at our hospital focused
on comparisons with MPRAGE. One
of the protocols placed the priority on
signal-to-noise ratio (SNR) (TI 900 ms),
and one placed the priority on contrast-
to-noise ratio (CNR) (TI 500 ms,
for higher contrast between gray matter
(GM) and white matter (WM)).
The scan times of both PETRA protocols
were adjusted such that they were
similar to that of the MPRAGE protocol
used in patient exams. In a pilot study,
volunteers and patients were scanned
with both PETRA protocols and with
Table 1: Sequence parameters
PETRA
Figs. 2, 3, 11, 12
PETRA
Figs. 4–10
MPRAGE
Figs. 2–8
3D-FLASH
Figs. 9–12
Voxel size
/ mm
(0.99 mm)3 for Figs. 2,3
(0.80 mm)3 for Figs. 11, 12
(0.99 mm)3 (0.94 mm)3 0.6 × 0.6 × 1.0 mm3
Matrix 288 × 288 288 × 288 256 × 256 346 × 384
Slices 288 288 176 144
TI / ms
500 for Figs. 2, 3
900 for Figs. 11, 12
700 900 n.a.
TR / ms 2.79 2.79 5.61 11
TE / ms 0.07 0.07 2.4 6.2
FA / deg 6 6 10 20
FatSat No Yes No Yes
Scan time 05:59 06:20
5:56
(3:54 for Fig. 2)
02:52
Technology
36 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
2A 2B
MPRAGE. The SNR on PETRA images
with TI 900 ms was visibly higher than
on MPRAGE images, while PETRA
images with TI 500 ms provided more
GM-to-WM contrast than necessary
for CE studies in the opinion of one
radiologist (MI). A decision was made
that some of the SNR could be ‘traded
for’ tissue CNR, and an intermediate
TI of 700 ms was chosen for further CE
studies. Statistical comparisons of contrast
enhancement and SNR between
PETRA and MPRAGE were performed
(that study is under review for publication
in a peer-reviewed journal).
PETRA was also compared with
3D-FLASH while remaining conscious
of the fact that, compared to the
PETRA implementation discussed in
this article, 3D-FLASH was capable of
higher spatial resolution.
Clinical observations
Comparisons of PETRA
and MPRAGE
Comparisons between PETRA and
MPRAGE with similar spatial reso-lution
and without fat suppression
are shown in figures 2 and 3.
GM-to-WM contrast is seen to be
similar, both without (Fig. 2) and
with (Fig. 3) contrast enhancement
(CE). In the latter CE case, a small
enhancing lesion appears to have the
2
3
Technology
Contrast-enhanced
screening for brain metastases
was indicated for a
57-year-old male who had
lung cancer. A tiny metastasis
(diameter 3.5 mm)
was detected in the medial
portion of the left temporal
lobe (arrows) on both
sequences.
(3A) CE MPRAGE,
(3B) CE PETRA with an
inversion time of 500 ms.
same size and enhancement on
the PETRA image as on the MPRAGE
image.
The remaining comparisons between
PETRA and MPRAGE, figures 4 through
8, were contrast-enhanced studies
of intracranial primary or metastatic
tumors with the same resolution
parameters as above. However, two
PETRA parameters were modified:
TI was changed to 700 ms, sacrificing
some GM-to-WM contrast for a gain
in SNR, and fat-suppression was
added (to PETRA only, avoiding
a change to the hospital’s routine
MPRAGE protocol).
3A 3B
MR imaging was indicated
for a 19-month-old* female
experiencing seizures with
no associated fever. Neither
sequence revealed any brain
abnormality. (2A) MPRAGE,
(2B) CE PETRA with a TI of
500 ms demonstrated
excellent gray-to-white-matter
contrast comparable
with that of MPRAGE.
* MR scanning has not been
established as safe for imaging
fetuses and infants less than
two years of age. The responsible
physician must evaluate
the benefits of the MR
examination
compared to those
of other imaging procedures.
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 37
1Repetition time of RF excitation pulses, which for MPRAGE is displayed on the MR console as ‘Echo spacing’.](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-19-2048.jpg)
![Acute MR Stroke Protocol in Six Minutes
Kambiz Nael; Rihan Khan; Kevin Johnson; Diego Martin
University of Arizona, Department of Medical Imaging, Tucson, AZ, USA
Background
Stroke is a common and serious
disorder,
with an annual incidence
of approximately 795,000. Based on
American Heart Association statistics
update in 2010, approximately
610,000 of these are first attacks,
and 185,000 are recurrent attacks. On
average, every 40 seconds, someone
in the United States has a stroke with
an estimated mortality rate of 5.5%,
claiming approximately 1 of every
18 deaths in the United States [1].
cranial hemorrhage including tissue
viability, site of occlusion, and collateral
status. While computed tomography
(CT) is the most widely available
and faster imaging modality,
some comprehensive stroke centers
favor streamlined MR protocols over
CT in the acute stroke setting due
to the higher specificity and superior
tissue characterization afforded by
MRI. The success of CT in initial evaluation
of AIS is due, in part, to fast
acquisition time, widespread availability
and ease of interpretation in
the emergency setting. The introduction
of multi-slice technology has
dramatically increased the speed and
simplicity of CT techniques and has
set a high standard for alternative
Neuroimaging plays a central role in
the evaluation of patients with acute
ischemic stroke (AIS). With improved
technology over the last decade, imaging
now provides information beyond
the mere presence or absence of intraimaging
techniques. A comprehensive
CT stroke algorithm including
parenchymal imaging (non-contrast
head CT), CT angiography (CTA), and
perfusion/penumbral imaging by CT
perfusion can now be acquired and
processed in less than 10 minutes
[5, 6].
MRI has been demonstrated to be
more sensitive for the detection of
acute ischemia and more specific for
delineation of infarction core volume
when compared to CT [7, 8]. However,
due to longer acquisition time
and limited availability; it has been
mainly used in large institutions
and comprehensive stroke centers.
A comprehensive MR protocol including
parenchymal imaging, MRA and
MR perfusion can now be obtained
in the order of 20 minutes as demonstrated
in several clinical trials [9–13].
If MRI is to compete with CT for
evaluation of acute stroke, there is
need for further improvements in
acquisition speed.
In this article we describe our modified
acute stroke MRI protocol that
can be obtained in approximately
6 minutes rivaling that of any comprehensive
acute stroke CT protocol.
We describe the technical aspects
and review a few clinical examples
based on our preliminary results.
92-year-old man with sudden onset of right-sided weakness and aphasia presented to our
emergency department after receiving IV-tPA at an outside institution. The acute stroke protocol
was performed after 9 hours from the onset in our institution and selective images are shown.
1A 1B
1C
Serial aligned DWI, ADC, EPI-FLAIR and EPI-GRE images
are shown. There is acute infarction of the left MCA
distribution involving the left operculum and insula.
Small focus of hemorrhagic conversion is present
within the area of infarction seen on both EPI-FLAIR
and EPI-GRE images.
1A Aligned DSC-Tmax, DSC-CBF and DSC-CBV images are shown.
DSC maps show a heterogeneous pattern of perfusion deficit
containing a small perfusion defect in the region of hemorrhage
and predominant luxury perfusion along the left MCA territory
seen on Tmax and CBF maps.
1B
Coronal MIP from CE-MRA of the entire supra-aortic arteries and cropped volume-rendered
reconstruction
of the intracranial arteries show no evidence of hemodynamic significant arterial
stenosis nor occlusion involving the proximal arteries. Note the high diagnostic image quality
of the CE-MRA images which are obtained after administration of 8 ml of contrast.
1C
44 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
How-I-do-it
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 45
How-I-do-it](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-23-2048.jpg)
![How-I-do-it How-I-do-it
Both FLAIR and GRE images have
been used to detect intraarterial clot
with variable sensitivity and specificity
[16, 17].
Introduction of fast imaging techniques
such as parallel acquisition [18]
and EPI [19, 20] has significantly
enhanced the performance of MR
imaging in terms of acquisition speed.
The main advantage of EPI, as in the
case of DWI imaging, is rapid acquisition
time, which is made possible by
rapid gradient switching which permits
the acquisition of all frequency
and phase encoding steps during a single
pulse cycle. The addition of parallel
imaging can further enhance the
acquisition speed and may also serve
to mitigate the geometric distortion
and susceptibility artifacts commonly
associated with long echo-train
sequences such as EPI [21, 22]. If their
potential is realized, the application
of EPI and parallel imaging techniques
to the FLAIR and GRE sequences can
result in reduction of image acquisition
time of the entire brain to less than
a minute, a three-fold reduction in
scan time over conventional imaging
[23, 24].
2. MR Angiogram
An important aspect of the workup
of patients with AIS is the imaging of
both the intracranial and extracranial
vasculature. Precise imaging of the
vascular tree is required during the
initial
assessment of patients with
acute stroke to accurately detect the
site of arterial disease, which in turn
can be crucial in determining the
type of acute therapy they are given.
Intravenous thrombolysis has been
shown to be more effective in small
distal vessels than in the large vessels
[25]. Larger vessel occlusion may be
more effectively treated with intra-arterial
thrombolysis or clot retrieval
devices while associated with fewer
complications [26, 27]. In addition,
MRA of the extracranial circulation
(neck arteries) is essential to establish
the mechanism of ischemia and
to prevent subsequent episodes.
Extracranial tandem stenoses with
plaque involving the carotid or vertebral
arteries can be the source of
disease
that triggers an acute stroke.
Time-of-flight MRA (TOF-MRA) has
been traditionally used in routine
stroke protocols to evaluate the
status of neck and brain arteries.
Despite its promising results [28],
TOF-MRA has significant disadvantages
including spin saturation and
phase dispersion due to slow or turbulent
flow [29, 30]. This can result
in overestimation of arterial stenosis
and increase false positive rates,
usually
due to slow flow distal to
a subocclusive thrombus or clot.
Most importantly the acquisition
time usually
is long, typically lasting
5–7 minutes.
The general consensus is that
contrast-enhanced MR angiography
(
CE-MRA) provides more accurate
imaging of extracranial vessel morphology
and of the degree of stenosis
than TOF-MRA techniques [31–33].
However, CE-MRA has not been
widely incorporated into acute stroke
protocols for several reasons. First,
CE-MRA has lower spatial resolution
relative to TOF-MRA, since the competing
requirements of coverage and
acquisition speed generally force a
compromise in spatial resolution for
CE-
MRA [34]. A second potential limitation
to incorporation of CE-MRA
into clinical stroke protocols is related
to the requirement of an extra contrast
dose, which would be in addition
to the intravenous contrast
bolus normally utilized for perfusion
imaging. With introduction of high
performance MR scanners and recent
advances in fast imaging tools such
as parallel acquisition (GRAPPA) [18],
high matrices can now be spread out
over a large field-of-view encompassing
the entire head and neck, resulting
in acquisitions with submillimeter
voxel sizes and acquisition times on
the order of 20 seconds [35, 36].
3. MR Perfusion
MR perfusion imaging has been used
broadly in the identification of potentially
salvageable tissue to determine
the best treatment strategy in patients
with acute ischemic stroke. Although
the concept of perfusion-diffusion
mismatch remains controversial
[37, 38], it has been used with some
success
to identify patients who may
respond favorably to revascularization
therapies in several clinical trials
[12, 13, 39].
Faster image acquisition combined
with higher signal-to-noise ratio
(SNR) resulting from the use of gadolinium
contrast agents has helped
dynamic susceptibility contrast (DSC)
perfusion become a more robust and
widely accepted technique in comparison
to arterial spin labeling (ASL)
to identify the presence of perfusion
abnormalities in patients with AIS.
A refined MR stroke protocol that
can combine both CE-MRA and DSC-perfusion
with improved acquisition
time and diagnostic image quality
as previously suggested [47, 48] may
have important therapeutic and
prognostic implications in the management
of patients with acute
stroke. Higher inherent SNR of higher
magnetic fields such as 3T with
improved multi-coil technology has
resulted in acquisition of low dose
CE-MRA of the supra-aortic arteries
with contrast dose as low as 8 ml
[40, 41]. A modified 2-phase contrast
injection scheme [46] can be used
to perform both CE-MRA and DSC
perfusion imaging, without the need
for additional contrast. The influence
of contrast dose reduction on DSC
perfusion has been evaluated by
several investigators [42, 43] and
contrast dose as low as 0.05 mmol/kg
has been used to perform DSC perfusion
with promising results [44, 45].
Advances in MR technology including
hardware and software, faster gradient
performance of MR scanners,
improved sequence design and fast
imaging tools such as EPI and parallel
Technical consideration
A comprehensive MR stroke protocol
has three essential components:
1) Parenchymal imaging that identifies
the presence and size of an irreversible
infarcted core and determines
the presence of hemorrhage;
2) MR angiogram to determine the
presence of proximal arterial occlusion
and/or intravascular thrombus
that can be treated with thrombolysis
or thrombectomy;
3) Pwerfusion imaging to determine
the presence of hypoperfused tissue
at risk for subsequent infarction if
adequate perfusion is not restored.
Below we describe each of these
components in detail and explain
how recent technical advances can
be used to enhance the performance
of the different aspects of acute
stroke imaging.
1. Parenchymal imaging
This encompasses three parts:
1) DWI (diffusion-weighted imaging)
that can detect ischemic tissue within
minutes of its occurrence and has
emerged as the most sensitive and
specific imaging technique for acute
ischemia, far beyond NECT or any
other type of MRI sequences [14].
2) FLAIR that helps to age the infarction
and permits the detection of
subtle subarachnoid hemorrhage;
3) GRE to detect parenchymal hemorrhage
with comparable accuracy for
the acute intraparenchymal hemorrhage
to CT [15].
2A 2B 2C
Serial aligned DWI, ADC, EPI-FLAIR and EPI-GRE
images are shown. There is acute right hemispheric
infarction involving both the ACA and MCA territories.
The EPI-FLAIR images demonstrate corresponding
hyperintensity suggestive of completed infarction.
There is associated mass effect. No hemorrhage is
identified on corresponding EPI-GRE images.
2B Coronal MIP from CE-MRA of the
2A Aligned DSC-Tmax, DSC-CBF and DSC-CBV
images are shown. There is a matched
perfusion defect with the region of infarction.
entire supra-aortic arteries shows
complete occlusion of the right
cervical ICA shortly after the origin.
There is some reconstitution of
flow signal at the supracliniod ICA
likely via collaterals.
2C
68-year-old man with left sided weakness and altered level of consciousness of unknown onset.
46 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 47](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-24-2048.jpg)
![How-I-do-it How-I-do-it
acquisition have promised the
potential
for a fast but comprehensive
MR stroke protocol that can be
performed in approximately 6 minutes
rivaling those of CT protocols.
Next we review our stroke protocol
in terms of image acquisition and
sequence parameters and show
some of the clinical examples that
were performed at our institution.
How we do it
At our institution, absent contraindication,
MR is the default imaging
modality for AIS. An MR safety questionnaire
is administered, and MR
compatible ECG leads are placed in
the emergency department as
patients are being evaluated by the
neurology team. Patients are then
placed onto an MR compatible table
and wheeled to the MR magnet for
rapid imaging.
We use both 3T and 1.5T MR scanners
(MAGNETOM Skyra and MAGNETOM
Aera, Siemens Healthcare, Erlangen,
Germany), with 3T the default scanner
for acute stroke imaging when available.
For signal reception, a combination
of a 16-element array coil [head
(n = 12), neck (n = 4)] will be used.
The coil design allows for application
of parallel acquisition in both the
phase and slice encoding directions.
Our 6-minute MR imaging protocol
consist of DWI, EPI-FLAIR, EPI-
GRE,
Table 1: Imaging protocol
CE-MRA and DSC perfusion. The clinical
indications
for using this acute
MR stroke protocol are patients with
acute (< 9 hours) presentation from
the onset of symptoms, unknown
onset of symptoms, NIHSS > 4, or
aphasia. Table 1 shows the sequence
parameters of our acquisition
protocol.
A modified 2-phase contrast injection
scheme [46] is used to perform both
CE-MRA and DSC perfusion imaging,
without the need for additional contrast.
To accomplish this, the total
volume of 20 ml of gadolinium
(
Multihance, Bracco Diagnostics Inc.,
Princeton, NJ, USA) that is used routinely
for MR perfusion is diluted with
normal saline to a total 50 ml volume.
Using a timing bolus, a total of
3 ml of contrast solution (1.2 ml of
gadolinium) is injected at 1.5 ml/s to
determine the transit time from the
arm vein to the cervical carotid arteries.
Then, a total of 22 ml contrast
solution (8.8 ml of gadolinium) is
injected at the same flow rate as the
timing injection for the CE-MRA
acquisition. A centric ordering k-space
is used for CE-MRA to minimize intracranial
venous contamination. Subsequently,
the remaining 25 ml of
contrast
solution (10 ml of gadolinium)
is injected at 5 ml/s for the MR
perfusion scan which is performed
at the end.
Image analysis
Following data acquisition, CE-MRA
image processing is performed on the
scanner console with standard commercial
software using a maximum
intensity projection (MIP) algorithm.
All of the reconstructed data, as well
as the source images are available on
the workstation for image analysis.
Perfusion
analysis will be performed
off-line on a dedicated FDA-approved
workstation (Olea-sphere, Olea Medical
SA, France). The arterial input function
is selected automatically and multiparametric
perfusion maps including
time-to-peak (TTP), time-to-maximum
(Tmax) cerebral blood flow (CBF) and
cerebral blood volume (CBV) are then
calculated using a block-circulant
singular
value decomposition technique
[49].
Our initial results using the described
stroke MR protocol have been promising.
We have scanned more than 600
patients with ASI since January 2013.
More than 97% of our studies have
been rated with diagnostic image quality.
The EPI-FLAIR sequence has been
used in parallel to conventional FLAIR
in a subset of patients with comparable
qualitative and quantitative results
[24]. In a study of 52 patients with
AIS, the mean ± SD of the signal intensity
ratios on EPI-FLAIR and FLAIR for
DWI positive lesions were 1.28 ± 0.16
and 1.25 ± 0.17 respectively with significant
DWI EPI-FLAIR EPI-GRE CE-MRA DSC
TR (ms) 4600 10000 (TI:2500) 1860 3.36 1450
TE 65 88 48 1.24 22
FA (degrees) – 90 90 25 90
Matrix 160 192 192 448 128
FOV 220 220 220 340 220
Slices (n × thickness) 30 × 4 30 × 4 40 × 3 120 × 0.8 30 × 4
Bandwidth (Hz/pixel) 1250 1488 964 590 1502
Parallel acquisition (GRAPPA) 3 3 – 3 3
Acquisition time 58 sec 52 sec 56 sec 20 sec 1 min and 30 sec
correlation (r = 0.899, z value
= 8.677, p < 0.0001). The EPI-GRE
sequence has been also used in parallel
to conventional GRE in a subset
of patients with comparable results
in terms of detection of hemorrhage
(Fig. 1) and blood clot in proximal
arteries.
The combination of CE-MRA and
DSC has been successfully tested in
our institution [48] with diagnostic
image quality. In a cohort of 30
patients with acute stroke, the specificity
of CE-MRA for detection of
arterial
stenosis > 50% was 97% compared
to 89% for TOF-MRA when compared
to DSA as the standard of reference
[48]. DSC perfusion imaging with
reduced contrast dose is feasible with
comparable quantitative and qualitative
results to a full-dose control group
[48]. Importantly, the presence of
contrast in the circulating blood of the
CE-
MRA half-dose group does not negatively
impact the image quality nor
the quantitative analysis of perfusion
data when compared to the control
full-dose group.
Conclusion
Described multimodal MR protocol
is feasible for evaluation of patients
with acute ischemic stroke with total
acquisition time of 6 minutes rivaling
that of the multimodal CT protocol.
References
1 Lloyd-Jones D, Adams RJ, Brown TM,
et al. Heart disease and stroke statistics
– 2010 update: a report from the
American Heart Association. Circulation.
Feb 23 2010;121(7):e46-e215.
2 Saver JL. Time is brain-quantified.
Stroke. Jan 2006;37(1):263-266.
3 Michel P, Bogousslavsky J. Penumbra is
brain: no excuse not to perfuse. Ann
Neurol. Nov 2005;58(5):661-663.
4 Gonzalez RG. Imaging-guided acute
ischemic stroke therapy: From “time
is brain” to “physiology is brain”. AJNR.
American journal of neuroradiology.
Apr 2006;27(4):728-735.
15 Kidwell CS, Chalela JA, Saver JL, et al.
Comparison of MRI and CT for detection
of acute intracerebral hemorrhage. JAMA :
the journal of the American Medical Associ-ation.
Oct 20 2004;292(15):1823-1830.
16 Flacke S, Urbach H, Keller E, et al. Middle
cerebral artery (MCA) susceptibility sign
at susceptibility-based perfusion MR
imaging: clinical importance and
comparison with hyperdense MCA sign at
CT. Radiology. May 2000;215(2):476-482.
17 Assouline E, Benziane K, Reizine D, et al.
Intra-arterial thrombus visualized on T2*
gradient echo imaging in acute ischemic
stroke. Cerebrovasc Dis. 2005;20(1):6-11.
18 Griswold MA, Jakob PM, Heidemann RM,
et al. Generalized autocalibrating partially
parallel acquisitions (GRAPPA). Magnetic
resonance in medicine : official journal of
the Society of Magnetic Resonance in
Medicine / Society of Magnetic Resonance
in Medicine. Jun 2002;47(6):1202-1210.
19 Mansfield P. Real-time echo-planar imaging
by NMR. Br Med Bull. Apr 1984;40(2):
187-190.
20 DeLaPaz RL. Echo-planar imaging. Radio-graphics.
Sep 1994;14(5):1045-1058.
21 Pruessmann KP. Parallel imaging at high
field strength: synergies and joint
potential. Topics in magnetic resonance
imaging : TMRI. Aug 2004;15(4):237-244.
22 Wiesinger F, Van de Moortele PF, Adriany
G, De Zanche N, Ugurbil K, Pruessmann
KP. Potential and feasibility of parallel MRI
at high field. NMR in biomedicine.
May 2006;19(3):368-378.
23 Kinoshita T, Okudera T, Tamura H,
Ogawa T, Hatazawa J. Assessment of
lacunar hemorrhage associated with
hypertensive stroke by echo-planar
gradient-echo T2*-weighted MRI.
Stroke. Jul 2000;31(7): 1646-1650.
24 Meshksar A, Khan R, Carmody R, Nael K.
The Role of Echo-planar Fluid-Attenuated
Inversion Recovery (EPI-FLAIR) in Acute
Stroke Setting: A Feasibility Study.
Paper presented at: ASNR; May 22, 2013,
2013; San Diego, CA.
25 del Zoppo GJ, Poeck K, Pessin MS, et al.
Recombinant tissue plasminogen activator
in acute thrombotic and embolic stroke.
Ann Neurol. Jul 1992;32(1):78-86.
26 Furlan A, Higashida R, Wechsler L, et al.
Intra-arterial prourokinase for acute
ischemic stroke. The PROACT II study:
a randomized controlled trial. Prolyse in
Acute Cerebral Thromboembolism.
JAMA : the journal of the American
Medical Association. Dec 1 1999;282(21):
2003-2011.
5 Zhu G, Michel P, Aghaebrahim A, et al.
Computed tomography workup of
patients suspected of acute ischemic
stroke: perfusion computed tomography
adds value compared with clinical
evaluation, noncontrast computed
tomography, and computed tomography
angiogram in terms of predicting
outcome. Stroke; a journal of cerebral
circulation. Apr 2013;44(4):1049-1055.
6 Schaefer PW, Roccatagliata L, Ledezma C,
et al. First-pass quantitative CT perfusion
identifies thresholds for salvageable
penumbra in acute stroke patients treated
with intra-arterial therapy. AJNR.
American journal of neuroradiology.
Jan 2006;27(1):20-25.
7 Jauch EC, Saver JL, Adams HP, Jr., et al.
Guidelines for the early management
of patients with acute ischemic stroke:
a guideline for healthcare professionals
from the American Heart Association/
American Stroke Association. Stroke.
Mar 2013;44(3):870-947.
8 Chalela JA, Kidwell CS, Nentwich LM,
et al. Magnetic resonance imaging and
computed tomography in emergency
assessment of patients with suspected
acute stroke: a prospective comparison.
Lancet. Jan 27
2007;369(9558):293-298.
9 Kang DW, Chalela JA, Dunn W, Warach S,
Investigators NI-SSC. MRI screening
before standard tissue plasminogen
activator therapy is feasible and safe.
Stroke. Sep 2005;36(9):1939-1943.
10 Hjort N, Butcher K, Davis SM, et al.
Magnetic resonance imaging criteria for
thrombolysis in acute cerebral infarct.
Stroke. Feb 2005;36(2):388-397.
11 Schellinger PD, Jansen O, Fiebach JB,
Hacke W, Sartor K. A standardized MRI
stroke protocol: comparison with CT in
hyperacute intracerebral hemorrhage.
Stroke; a journal of cerebral circulation.
Apr 1999;30(4):765-768.
12 Albers GW, Thijs VN, Wechsler L, et al.
Magnetic resonance imaging profiles
predict clinical response to early reperfusion:
the diffusion and perfusion
imaging evaluation for understanding
stroke evolution (DEFUSE) study. Ann
Neurol. Nov 2006;60(5):508-517.
13 Davis SM, Donnan GA, Parsons MW, et al.
Effects of alteplase beyond 3 h after
stroke in the Echoplanar Imaging
Thrombolytic Evaluation Trial (EPITHET):
a placebo-controlled randomised trial.
Lancet neurology. Apr 2008;7(4):
299-309.
14 Fiebach JB, Schellinger PD, Jansen O,
et al. CT and diffusion-weighted MR
imaging in randomized order: diffusion-weighted
imaging results in higher
accuracy and lower interrater variability
in the diagnosis of hyperacute ischemic
stroke. Stroke; a journal of cerebral
circulation. Sep 2002;33(9):2206-2210.
48 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 49](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-25-2048.jpg)
![Clinical Neurology Neurology Clinical
at the same time, but is generally
ignored. A filter is applied to the phase
image (High-pass Hamming Window
Filter) on a 64 × 64 matrix to reduce
aliasing artifacts. A new phase mask is
created which, when added to the
magnitude image, creates the susceptibility
image. In order to obtain a
better
interpretation, minimum intensity
projections (minIP) are used [6].
During post-processing the phase contrast
image is filtered to reduce undesirable
low spatial frequency components,
leaving the high frequency field
variations. The phase mask created
can be ‘positive’ or ‘negative’. The
phase mask is multiplied using the
original magnitude image to produce
images that maximize the negative
intensity of the mineralization of the
parenchyma. Minimum intensity projection
(usually from 2 to 4 slices) is
used to display the processed data [1].
MAGNETOM ESSENZA 1.5 Tesla MRI
unit with the following settings:
TR 49 ms, TE 40 ms, FA 15°, number
of slices 60, slice thickness 2 mm,
acquisition matrix 256 ×157.
Susceptibility-weighted imaging
takes advantage of the loss of signal
intensity created by alterations in
a homogenous magnetic field; these
disturbances can be caused by several
different paramagnetic or diamagnetic
substances. The loss of signal
intensity in the T2*-weighted
sequence is a result of the difference
in the precession rate of the spins
[5].
The susceptibility image is obtained
during the acquisition process by
combining the magnitude and phase
of the images. Routine MR images
are magnitude images where the signal’s
intensity is converted to a gray
scale. Phase information is obtained
Introduction
Susceptibility-weighted imaging
(SWI) is a sequence that utilizes a
phenomenon in which the phase and
change in the local magnetic field of
the tissues are proportional to one
another, provided the echo time is
constant [1]. It uses magnitude and
phase images, as well as a summation
of these in a three-dimensional
gradient echo sequence with flow
compensation [2]. It offers very high
sensitivity for visualizing calcium,
non-heme iron (ferritin) and hemoglobin
degradation products (deoxyhemoglobin
and hemosiderin) [3, 4].
Initial experience
By means of a series of cases we
will illustrate the clinical usefulness
of SWI with certain neurological
conditions. The studies reviewed
were performed in the Neurological
Scanography Magnetic Resonance
Imaging Service using a Siemens
The method is highly sensitive for purposes
of visualizing venous circulation,
blood products and iron content, and
is also useful for evaluating the vascularization
of tumors and for identifying
brain tissue that has been compromised
by a stroke, vascular dementia
or trauma, and can also be used in
functional imaging [1, 4, 7-9] (Fig. 1).
Hemorrhage
Oxyhemoglobin, formed by the binding
of an oxygen and an iron atom
contained in the Hem group, is a diamagnetic
substance. When the oxygen
is released from the iron atom it forms
deoxyhemoglobin, which is paramagnetic
because of its unpaired electrons.
Metahemoglobin is produced when
deoxyhemoglobin oxidizes, making it
less stable; in this state there is little
susceptibility effect and thus it is more
easily visualized in T1w images.
Hemosiderin is the final product of the
degradation of hemoglobin when it
degrades within phagocytic cells, and
is a highly paramagnetic [3, 4, 10]
substance. Diamagnetic substances
produce a weak local magnetic field,
while paramagnetics generate a stronger
magnetic field that leads to a
signal
de-phase and therefore a signal
reduction in the T2*w sequence [4].
The ferritin produced by different
metabolic processes also has para-magnetic
characteristics and is
associated with Parkinson’s disease,
Huntington’s disease and Alzheimer’s
disease [9-11].
Trauma
In the detection of diffuse axonal
damage, this approach is more sensitive
than conventional imaging for
detecting microhemorrhages in the
deep and subcortical white matter,
which can be obscured in computed
tomography (CT) scans [12, 13]. It is
three to six times more sensitive than
gradient echo images for detecting
the number, size, and location of the
lesions associated with this clinical
status of the patient [1, 13-16]. It is
equally useful in detecting brain-stem
lesions, subarachnoid and intraventricular
hemorrhage, as well as
other types of hemorrhagic lesions
of different origins [17] (Fig. 2).
Calcifications
Calcium is also diamagnetic and can
lead to changes in the susceptibility
image [12, 18]. SWI differentiates
iron from calcium based on their diamagnetic
or paramagnetic characteristics
in the filtered-phase image.
Calcium
appears brilliant in this latter
image, while the hemorrhage and
its derivative products have low signal
intensity. This differentiation is
important when dealing with neurodegenerative
and metabolic diseases,
trauma, and tumors [12, 18].
Vascular malformations
Venous blood causes non-homogeneity
in the magnetic field due to the
paramagnetic effect of the deoxygenated
blood due to T2* reduction,
depending on the oxygen saturation,
the hematocrit and the condition of
the erythrocytes; thus, the deoxyhemoglobin
present in venous blood
allows for the visualization of the latter
[4] as well as the phase difference
between the vessels and surrounding
structures [19]. The susceptibility
image provides contrast similar to that
of a functional image (BOLD blood
oxygen level-dependent).
SWI is more sensitive in the detection
of vascular structures that are hidden
to T2* and low-flow malformations
that are not detected by MR angiography,
such as venous development
malformations, telangiectasias and
cavernomas, as well as vascular abnormalities
and calcifications related to
Sturge-Weber Syndrome, since it is
not affected by flow velocity or direction
[20-24]. In dural sinus thrombosis
they show venous statis and collateral
flow, as well as early detection
of venous hypertension before infarcts
or hemorrhages occur [7, 8, 19]
(Figs. 3–5).
Susceptibility-Weighted Imaging.
Initial Experience
José Luis Ascencio L.1; Tania Isabel Ruiz Z.2
1 Escanografia Neurologica, Medellin, Colombia
2 Universidad CES, Radiology, Medellín, Antioquia, Colombia
1A
Patient with bilateral frontal hemorrhagic contusion. (1A) T2w axial; no lesions observed. (1B) Axial gradient echo shows a low-signal
lesion in left frontal lobe with a slight blooming effect. (1C) SWI magnitude, two bilateral frontal hemorrhagic contusions are observed.
1
1B 1C 2A 2B 2C
Patient with diffuse axonal lesion. (2A) T2w axial; no lesions observed. (2B) Low-signal, puntiform lesions. (2C) SWI minIP
makes the multiple microhemorrhagic lesions more apparent.
2
52 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 53](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-27-2048.jpg)
![Clinical Neurology Neurology Clinical
3A 3B 3C
Venous development anomaly. (3A) Axial gradient echo; anomaly not visible. (3B) Axial contrast-enhanced image shows right
frontal venous development anomaly that is more evident in the susceptibility image (3C).
3
4A 4B
4 Left frontal cavernoma. (4A) Axial proton density-weighted image; (4B) minIP SWI.
5A 5B 5C
Left parietal arteriovenous malformation. (5A) Axial PDw show serpinginous images with absence of flow signal. (5B) mIP SWI.
(5C) MIP TOF shows the AVM and the cortical drainage vein.
5
Brain tumors
This approach provides information
that supplements T1 with contrast for
detecting margins, internal architecture,
hemorrhage and vascularization
of a tumor that are not visible with
conventional sequences. This aids in
differentiating between a recurring
6A 6B
7A 7B 7C
Metastatic melanoma. (7A) T2w axial; large mass displacing the midline, with major edema and hypointense zone due to
hemorrhage
in the medial portion. (7B) Magnitude image, (7C) MIP SWI shows a greater hemorrhagic component of the mass,
on the contralateral side, as well as intraventricular hemorrhaging.
7
6C
Hemorrhagic metathesis. (6A) T1w axial gadolinium-enhanced, (6B) T2w axial show a left parietal mass with heterogeneous
enhancement, perilesional edema and mass effect on the lateral ventricles. (6C) MIP SWI shows hypervascularity and hemorrhage
in the interior of the mass.
6
tumor and post-operative changes.
The use of susceptibility imaging
before and after the administration
of gadolinium can differentiate areas
of enhancement of the vessels.
Because of its suppression of cerebrospinal
fluid, it enhances contrast
between edema and normal tissue,
similarly to what is provided by
FLAIR, thus facilitating the detection
of space-occupying lesions [4, 7, 25]
(Figs. 6–8).
8A 8B 8C
Oligodendroglioma. (8A) T1w axial gadolinium shows mass with enhanced foci and a cystic component, (8B) MIP SWI right parietal
hypervascular mass with increased relative flow (8C).
8
54 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 55](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-28-2048.jpg)
![Clinical Neurology Neurology Clinical
9A 9B 9C
CVD with hemorrhagic transformation. (9A) Axial T2w, (9B) gradient echo in patient with left parietal hemorrhagic infarct with
surrounding edema and hemorrhage. (9C) SWI makes the greater hemorrhagic component more obvious.
9
10A 10B 10C
Right MCA aneurism with bleeding. (10A) Axial T2w, (10B) TOF demonstrating aneurysm with bleeding, (10C) SWI aneurism with
greater bleeding than that shown in the T2w sequence.
10
Children: CT versus MRI versus Susceptibility
Weighted Imaging (SWI). Journal of
Neurotrauma. 2011;28(6):915-27.
14 Wang M, Dai Y, Han Y, Haacke EM, Dai J,
Shi D. Susceptibility weighted imaging in
detecting hemorrhage in acute cervical
spinal cord injury. Magnetic resonance
imaging. 2011;29(3):365-73.
15 Tong KA, Ashwal S, Holshouser BA,
Shutter LA, Herigault G, Haacke EM, et
al. Hemorrhagic Shearing Lesions in
Children and Adolescents with Posttraumatic
Diffuse Axonal Injury: Improved
Detection and Initial Results1. Radiology.
2003 May 1, 2003;227(2):332-9.
16 Babikian T, Freier MC, Tong KA, Nickerson
JP, Wall CJ, Holshouser BA, et al. Susceptibility
weighted imaging: neuropsychologic
outcome and pediatric head injury.
Pediatr Neurol. 2005;33(3):184-94.
17 Wu Z, Li S, Lei J, An D, Haacke EM. Evaluation
of Traumatic Subarachnoid Hemorrhage
Using Susceptibility-Weighted
Imaging. AJNR Am J Neuroradiol. 2010
August 1, 2010;31(7):1302-10.
18 Wu Z, Mittal S, Kish K, Yu Y, Hu J, Haacke
EM. Identification of calcification with MRI
using susceptibility-weighted imaging:
A case study. Journal of Magnetic
Resonance Imaging. 2009;29(1):177-82.
19 Tsui Y-K, Tsai FY, Hasso AN, Greensite F,
Nguyen BV. Susceptibility-weighted
imaging for differential diagnosis of
cerebral vascular pathology: A pictorial
review. Journal of the neurological
sciences. 2009;287(1):7-16.
20 Hu J, Yu Y, Juhasz C, Kou Z, Xuan Y,
Latif Z, et al. MR susceptibility weighted
imaging (SWI) complements conventional
contrast enhanced T1 weighted MRI
in characterizing brain abnormalities of
Sturge-Weber Syndrome. Journal of
Magnetic Resonance Imaging.
2008;28(2):300-7.
21 Deistung A, Dittrich E, Sedlacik J,
Rauscher A, Reichenbach JR. ToF-SWI:
Simultaneous time of flight and fully
flow compensated susceptibility weighted
imaging. Journal of Magnetic Resonance
Imaging. 2009;29(6):1478-84.
22 Koopmans P, Manniesing R, Niessen W,
Viergever M, Barth M. MR venography
of the human brain using susceptibility
weighted imaging at very high field
strength. Magnetic Resonance Materials
in Physics, Biology and Medicine.
2008;21(1):149-58.
23 de Champfleur NM, Langlois C, Ankenbrandt
WJ, Le Bars E, Leroy MA, Duffau H,
et al. Magnetic Resonance Imaging
Evaluation of Cerebral Cavernous Malformations
With Susceptibility-Weighted
Imaging. Neurosurgery. 2011;68(3):
641-8 10.1227/NEU.0b013e31820773cf.
24 Jagadeesan BD, Delgado Almandoz JE,
Moran CJ, Benzinger TLS. Accuracy of
Susceptibility-Weighted Imaging for the
Detection of Arteriovenous Shunting in
Vascular Malformations of the Brain.
Stroke. 2011 January 1, 2011;42(1):
87-92.
25 Hori M, Ishigame K, Kabasawa H,
Kumagai H, Ikenaga S, Shiraga N, et al.
Precontrast and postcontrast susceptibility-
weighted imaging in the assessment
of intracranial brain neoplasms at 1.5 T.
Japanese Journal of Radiology. 2010;
28(4):299-304.
26 Cherian A, Thomas B, Kesavadas C,
Baheti N, Wattamwar P. Ischemic hyperintensities
on T1-weighted magnetic
resonance imaging of patients with
stroke: New insights from susceptibility
weighted imaging2010 January 1, 2010
Contract No.: 1.
27 Mittal P, Dua S, Kalia V. Pictorial essay:
Susceptibility-weighted imaging in
cerebral ischemia2010.
28 Santhosh K, Kesavadas C, Thomas B,
Gupta AK, Thamburaj K, Kapilamoorthy
TR. Susceptibility weighted imaging: a
new tool in magnetic resonance imaging
of stroke. Clinical Radiology.
2009;64(1):74-83.
29 Hermier M, Nighoghossian N. Contribution
of Susceptibility-Weighted Imaging to
Acute Stroke Assessment. Stroke. 2004
August 1, 2004;35(8):1989-94.
30 Haacke EM, Makki M, Ge Y, Maheshwari
M, Sehgal V, Hu J, et al. Characterizing
iron deposition in multiple sclerosis
lesions using susceptibility weighted
imaging. Journal of Magnetic Resonance
Imaging. 2009;29(3):537-44.
31 Niwa T, de Vries L, Benders M, Takahara T,
Nikkels P, Groenendaal F. Punctate white
matter lesions in infants: new insights
using susceptibility-weighted imaging.
Neuroradiology. 2011:1-11.
32 Vinod Desai S, Bindu PS, Ravishankar S,
Jayakumar PN, Pal PK. Relaxation and
susceptibility MRI characteristics in
Hallervorden-Spatz syndrome. Journal
of Magnetic Resonance Imaging.
2007;25(4):715-20.
33 Kirsch W, McAuley G, Holshouser B,
Petersen F, Ayaz M, Vinters HV, et al.
Serial susceptibility weighted MRI
measures brain iron and microbleeds in
dementia. Journal of Alzheimer’s disease:
JAD. 2009;17(3):599-609.
References
1 Haacke EM, Xu Y, Cheng Y-CN, Reichenbach
JR. Susceptibility weighted imaging (SWI).
Magnetic Resonance in Medicine.
2004;52(3):612-8.
2 Haacke EM, Mittal S, Wu Z, Neelavalli J,
Cheng Y-CN. Susceptibility-Weighted
Imaging: Technical Aspects and Clinical
Applications, Part 1. AJNR Am J Neuroradiol.
2009 January 1, 2009;30(1):19-30.
3 Mittal S, Wu Z, Neelavalli J, Haacke EM.
Susceptibility-Weighted Imaging: Technical
Aspects and Clinical Applications, Part 2.
AJNR Am J Neuroradiol. 2009 February 1,
2009;30(2):232-52.
4 Sehgal V, Delproposto Z, Haacke EM,
Tong KA, Wycliffe N, Kido DK, et al.
Clinical applications of neuroimaging with
susceptibility-weighted imaging.
Journal of Magnetic Resonance Imaging.
2005;22(4):439-50.
5 Tong KA, Ashwal S, Obenaus A, Nickerson
JP, Kido D, Haacke EM. Susceptibility-
Weighted MR Imaging: A Review of Clinical
Applications in Children. AJNR Am J Neuroradiol.
2008 January 1, 2008;29(1):9-17.
6 Matsushita T AD, Arioka T, Inoue S, Kariya
Y, Fujimoto M, Ida K, Sasai N, Kaji M,
Kanazawa S, Joja I. Basic study of susceptibility-
weighted imaging at 1.5T. Acta
medica Okayama. [Journal Article]. 2008
Jun;62(3):159-68.
7 Thomas B, Somasundaram S, Thamburaj K,
Kesavadas C, Gupta A, Bodhey N, et al.
Clinical applications of susceptibility
weighted MR imaging of the brain –
a pictorial review. Neuroradiology.
2008;50(2):105-16.
8 Ong BC, Stuckey SL. Susceptibility weighted
imaging: A pictorial review. Journal of
Medical Imaging and Radiation Oncology.
2010;54(5):435-49.
9 Robinson RJ, Bhuta S. Susceptibility-
Weighted Imaging of the Brain: Current
Utility and Potential Applications.
Journal of Neuroimaging. 2011:no-no.
10 Goos JDC, van der Flier WM, Knol DL,
Pouwels PJW, Scheltens P, Barkhof F, et al.
Clinical Relevance of Improved Microbleed
Detection by Susceptibility-Weighted
Magnetic Resonance Imaging. Stroke.
2011 May 12, 2011:STROKEAHA.
110.599837.
11 Gupta D, Saini J, Kesavadas C, Sarma P,
Kishore A. Utility of susceptibility-weighted
MRI in differentiating Parkinson’s disease
and atypical parkinsonism. Neuroradiology.
2010;52(12):1087-94.
12 ZHU Wen-zhen QJ-p, ZHAN Chuan-jia, SHU
Hong-ge, ZHANG Lin, WANG Cheng-yuan,
XIA Li-ming, HU Jun-wu, FENG Ding-yi.
Magnetic resonance susceptibility weighted
imaging in detecting intracranial calcification
and hemorrhage. Chinese Medical
Journal. [Journal Article]. 2008 oct
20;121(20):2021-5.
13 Beauchamp MH, Ditchfield M, Babl FE,
Kean M, Catroppa C, Yeates KO, et al.
Detecting Traumatic Brain Lesions in
Contact
Jose Luis Ascencio L.
Escanografia Neurologica
Medellin
Colombia
jotaascencio@yahoo.com
Cerebrovascular disease
The susceptibility image can be used
together with diffusion images to
detect the hypoperfused region, the
presence of hemorrhaging within the
infarct (which could affect the treatment),
detect acute thrombus and
predict the likelihood of hemorrhagic
transformation and hemorrhagic
complications during and after thrombolysis
treatment, as well as microbleeding
due to amyloid angiopathy
and lacunar infarcts in patients
with hypertensive encephalopathy
[19, 26-28] (Figs. 9, 10).
Vascular occlusion can change the
susceptibility of the tissue as a
result of reduced arterial flow and
an increase in the accumulation
of deoxygenated blood, which
increases the amount of deoxy-hemoglobin
that can be detected
by SWI [27, 29].
Neurodegenerative
illnesses
Certain disorders, such as Parkinson’s
Disease, Huntington’s Disease,
Alzheimer’s, multiple sclerosis
and amyotrophic lateral sclerosis
(Lou Gherig’s Disease) present with
abnormal iron deposition, which can be
detected and quantified using susceptibility
imaging [11, 30-33]. SWI can
show chronic demyelinating plaques
with iron depositions that are hidden in
conventional sequences, as the iron content
makes the lesions more visible. It
can also determine the iron content of
the nucleii of deep gray matter that can
also be observed in patients with multiple
sclerosis, as well as the perivenular
distribution of the demyelinating lesions
[30].
56 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 57](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-29-2048.jpg)
![How-I-do-it How-I-do-it
Curve Fitting of the Lipid-Lactate Range
in an MR Spectrum: Some Useful Tips
Jamie Ho Xiu Mei; Helmut Rumpel
Department of Diagnostic Radiology, Singapore General Hospital, Singapore
Purpose
In the course of a neurological spectroscopic
application, questions about
the lipid-lactate signals often occur:
How can we accurately interpret an
asymmetric pattern with ‘humps’ and
partially inverted signals? How can
we recognise some fundamental
patterns
in order to differentiate
between, for example, an early membrane
degradation and necrosis?
In addition, questions always arise
about the proportion of lipid and
lactate
in an overlapping pattern.
This article offers some hints to curve
fitting of the lipid-lactate range and
to avoid incorrect or misleading labelling
of the peaks of an MR spectrum.
Background
The lipids are a large and diverse
group of naturally occurring molecules
with various functions, from
storing energy to being components
of membranes. They include miscellaneous
subgroups, such as triglycerides
of subcutaneous fat or bone
marrow
and glycerophospholipids of
membranes (Fig. 1). They have in
common fatty acids comprising four
different proton groups: olefinic protons
at 5.3 ppm, allylic protons and
protons adjacent to the carboxyl
group at 2.0 ppm, aliphatic methylene
groups (main peak) at 1.3 ppm,
and the terminal methyl group at
0.9 ppm. However, their T2 values
differ according to whether the fatty
acids are part of triglycerides or cell
membranes. Unlike glycerophospholipids,
fatty acids in membranes are
embedded in the interior of the membrane
resulting in efficient spin-spin
interactions, thus revealing short T2
Line curve fitting in the
NUMARIS software
Within the NUMARIS software, line
curve fitting is done in the frequency
domain [2]. It is the last of the post-processing
steps. Provided that the
phase correction is optimal, the
adequate line curve fitting protocol
has to be selected out of a set of
three ‘customised’ protocols, namely:
1. lipid signal only, e.g. TE 30_lip
2. lactate signal only, e.g. TE 30_lac
3. lactate and lipid signals e.g.
TE 30_lip_lac
They are easily derived from the
Siemens
protocol (CSI or SVS) TE 30
using the interactive post-processing
environment and adding new peak
parameters and peak restrictions (see
User Manual). The lipid-containing
protocols fit both the lipid_1.3 and
the lipid_0.9 peaks with parameters
as shown in figure 3. An analogous
set of protocols shall be customised
for TE 135.
How do we identify lipids and lactate?
Lipid-only peak (Fig. 3A): Here, the
assignment method is straightforward.
Firstly, we look at the line
width of the peaks. Their full width
at half maximum is reciprocally proportional
to T2. As a rule of thumb,
T2 of lipids is shorter than that of
lactate, and thus the lipid peak is
broader. Secondly, one can also
look at the symmetry of the peak
(whether it has a symmetrical
Gaussian / Lorentzian shape).
Thirdly, lipids should appear as two
peaks. Their relative intensities
may vary depending on the stage
of degradation (Fig. 4).
Lactate-only peak (Fig. 3A): The
scalar coupling constant JIS and
phase dependency of the lactate
signal, S ~ [cos (JIS TE)] can be used
as a kind of lactate editing: a doublet
signal of 7 Hz and a 180 degree
phase shift at TE 135 ms. In contrast
to the lipid peak, the lactate peak
reveals a sharp doublet pattern, or
at least it appears foreshadowed.
H1 H H2 H H
H H3 H4 H
1
(1A) Triglycerides,
(1B) Phospholipid
structures. Note that
in MR terminology,
peaks arising from
methyl-/methylene
protons of fatty
acids are referred
to as ‘lipid peaks’.
Schematic MR spectrum of normal brain tissue. Myo-inositol, choline, and lipid
signals are ‘iceberg-like’ as most signal is invisible due to short T2. In the event of
hydrolysis of inositol phospholipids or high cellular membrane turnover,
myo-inositol, choline, and lipids become more ‘MR visible’ due to change in T2
towards higher values.
2
values. As such, even in short TE
spectra, signals from intact membranes
are hardly visible, whereas the
freely tumbling fatty acids in triglycerides
and also the less restricted
‘fragments’ of fatty acids from membrane
degradation produce strong
signals due to a longer T2. In drawing
a comparison with an iceberg, protons
of membranes namely those of
methyl-, methylene-groups, choline,
and myo-inositol, are MR-invisible
unless they ‘surface’ due to degradation
processes (Fig. 2). For example,
depending on the grade of degradation
of membranes in brain tumours,
the MR signals of the said membrane
fragments are raised in a characteristic
way [1].
Lipid-lactate peak (Fig. 3A): In the
case of lipid-lactate overlapping,
this approach of Lorentzian-Gaussian
curve fitting will be inaccurate
in differentiating between lipid and
lactate proportions, because the
best fit is driven by the method of
least squares rather than taking
2
Schematic diagram of the peak pattern in the lipid-lactate range. (3A) Broad lipid
peak due to short T2 (blue), sharp lactate doublet due to J coupling and long T2
(orange), superposition of the lipid and lactate peak (black) for short TE.
Depending on slight resonance offsets and relative intensities, often a ‘hump’ is
visible (arrow). (3B) Superposition of lipid and lactate peaks for long TE (135 ms).
3
pre-knowledge of different shapes
into account. In fact the operator
will identify asymmetric patterns
(showing a hump on one side) or
if they partially invert on TE 135
spectra (Fig. 3B).
Representative cases are shown in
figures 5 and 6.
1B
3A 3B
1A
O
Fatty acid Glycerol Fatty acid
O
O
O
O
C
O
C
H H H H
H
H H H
H
H H
Choline Choline
fatty acid
fatty acid fatty acid
fatty acid
Choline Choline
fatty acid
fatty acid fatty acid
fatty acid
Choline Choline
fatty acid
fatty acid fatty acid
fatty acid
myo-Ins myo-Ins
fatty acid
fatty acid fatty acid
fatty acid
myo-Ins myo-Ins
fatty acid
fatty acid fatty acid
fatty acid
myo-Ins myo-Ins
fatty acid
fatty acid fatty acid
fatty acid
INS CHO CRE NAA LAC LIP
MR visible
@TE 135
MR invisible
@TE 135
58 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 59](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-30-2048.jpg)
![How-I-do-it
Lipid
I: 19.3
Lac
I: 6.4
Illustration of a representative case of lipid-lactate overlap. (5A) Metabolite maps. In (5B), the lactate doublet peak dominates the
pattern which also shows an unequal doublet with higher signal and broadening of the right peak of the doublet. Therefore, the
TE_30_lip_lac protocol has been used. Due to T2 relaxation, the lipid component became negligible at TE 135, and only the lactate
has been labelled with the TE_135_lac protocol in (5C). In 5D–F, lipids overwhelm the pattern. Since lactate is neither detected on
TE 135, nor a ‘hump’ is observable on TE 30, the protocol TE_30_lip_lac is inappropriate.
C C H
n Lipid_1.3
5 Line curve fitting in the syngo.via software. Lipid-lactate compositions are
slightly different in (7A) and (7B).
7
Line curve fitting in the NUMARIS software. (4A) Entire spectrum, (4B) lactate component, (4C) lipid _1.3 component,
(4D) lipid_0.9 component. Note that the proportions of lipid and lactate should be regarded with caution.
4
Line curve fitting in syngo.via
In syngo.via the approach of curve
fitting
in the time domain has been
chosen [3]. It is based on PRISMA [4]
using the basis set of metabolic time
signals of brain metabolites together
with published values of chemical
shifts and coupling constants.
The delineation of a lactate-lipid overlap
is based on free induction decays
(FIDs), in which the distinct T2 values
make the segmentation more accurate.
Figure 7 depicts two examples:
4A
5A
Spectra of a glioblastoma multiforme,
(6A) from the peri-lesional
area, (6B) from the solid
enhancing area. Note the relative
intensities of lipid at 1.3 ppm
and 0.9 ppm as in an early stage
of membrane degradation (6A)
these signals do not reflect a
proton density even at TE 30,
rather they are T2-weighted,
whilst in a more advanced stage
(6B) T2 becomes long enough
for both signals resulting in a
proton density spectrum (6C).
6
6A
6B
TE 30
TE 30_lip_lac
TE 135
TE 135_lac
TE 30
TE 30_lip
TE 30
TE 30_lip_lac
TE 135
TE 135_lip
Lactate area Lipid area
a) with a clearly visible hump at half
maximum of the Lip_1.3 peak, and
b) with only an adumbrated hump
(arrow).
It demonstrates that delineating
highly non-uniform signal compositions
is possible with only one
protocol for all cases. The protocol
can be easily linked together by
selecting the appropriate lactate and
lipid templates as shown in figure 8.
7A
7B
4B
4C
4D
Lactate map Lipid map
5B 5C 5D 5E 5F
6C
Methylene – Group
@ 1.3ppm
Methyl – Group
@ 0.9ppm
H
H
H
H
I: 29.34
Lac
I: 1.19
Lipid_1.3
I: 18.46
Lac
I: 4.84
How-I-do-it
60 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 61](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-31-2048.jpg)
![How-I-do-it
8 Lipid-lactate protocol in syngo.via.
References
1 Li X, Vigneron DB, Cha S, Graves EE,
Crawford F,Chang SM, Nelson SJ. 2005.
Relationship of MR-Derived Lactate,
Mobile Lipids, and Relative Blood Volume
for Gliomas in Vivo. AJNR Am J
Neuroradiol
26:760–769.
2 Mierisová Š, Ala-Korpela M.2001.MR
spectroscopy quantitation: a review of
frequency domain methods. NMR in
Biomedicine;14, 247-259.
3 Vanhamme L, Sundin T, Van Hecke P,
Van Huffel S. 2001. MR spectroscopy
quantitation: a review of time-domain
methods NMR in Biomedicine;
14,233–246.
4 http://elib.suub.uni-bremen.de/
publications/diss/html/E-Diss1066_HTML.
html
Contact
Helmut Rumpel, Ph.D.
Department of
Diagnostic Radiology
Singapore General Hospital
helmut.rumpel@sgh.com.sg
Conclusion
Carefully labelled spectra in the lactate–
lipid region are a prerequisite
to sending them to a PACS system as
otherwise they can be misleading
for further examinations and treatment
planning by clinicians.
Curve fitting based on chemical
shift assignments should be made
with caution. The technologist is
required to identify ‘what is what’,
as erroneous lipid-lactate discrimination
is inherent to frequency
domain curve fitting of overlapping
peaks. By following
basic steps,
the lactate-lipid overlap within the
region from 0.9 to 1.3 ppm can be
reliably delineated. However, the
proportions of lipid and lactate
remain ambiguous as various sets
of model peaks, i.e. half-width,
signal
intensity and chemical shift,
may lead to the same result of
curve fitting.
Incorporation of prior knowledge
such as supportive model spectra
automates the curve fitting of a
lactate-
lipid overlap. The protocol
TE_30_lip_lac can be made standard
practice.
8
62 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
Head-to-Toe Imaging Clinical
Relevant clinical information
at your fingertips
From technology to clinical applications, you will find
all the latest news on Siemens MR at
www.siemens.com/magnetom-world
Don’t miss the talks of international experts on
all aspects of Magnetic Resonance Imaging.
Go to
Clinical Corner > Clinical Talks
The centerpiece of the MAGNETOM World Internet
platform consists of our users’ clinical results. Here you
will find case reports and clinical methods.
Go to
Clinical Corner > Case Studies
Just a mouse click away you will find application videos
and useful tips allowing you to optimize your daily MR
examinations.
Go to
Clinical Corner > Application Tips
Liver Imaging Today
Tobias Heye, M.D.1; Mustafa R. Bashir, M.D.²
1Department of Radiology, University Hospital Basel, Switzerland
2Department of Radiology, Duke University Medical Center, NC, USA
Introduction
Liver disease is a global burden with a
growing incidence and prevalence. The
World Health Organization recently esti-mated
that there are 800,000 cirrhosis-related
deaths per year world-wide [1].
Chronic liver disease has a great impact
on public health care costs with thera-peutic
options ranging from antiviral
treatment for viral hepatitis to orthotopic
liver transplant for end stage cirrhosis. A
variety of pathogens, which can be toxic,
viral, metabolic or autoimmune in nature,
can induce fibrosis which may progress to
cirrhosis if the disease is not detected and
treated. An estimated 150 million people
world-wide are chronically infected with
hepatitis C virus, approximately 350,000
people die due to hepatitis C related liver
disease [2]. Liver fibrosis may be revers-ible
at an early stage, which indicates the
1A 1B
1089_Flash_52_Inhalt_CC.indd 111 02.04.13 10:45
How-I-do-it
Abdominal Imaging Clinical
importance of screening and detection of
liver disease. Many forms of liver fibrosis
and cirrhosis especially secondary to viral
hepatitis increase the risk for the devel-opment
of liver cancer, namely hepato-cellular
carcinoma.
Non-alcoholic steatohepatatis is emerg-ing
as a major pathway into chronic liver
disease and is closely related to other
metabolic disease entities such as diabe-tes
and morbid obesity. The incidence
and prevalence of these diseases has
risen steadily over recent years.
In a clinical context, liver disease is often
reflected by a combination of several
contributing factors, fibrosis, hepatic
steatosis and iron overload, each with
different forms of manifestations.
Although these diseases are considered
‘diffuse’, actual hepatic parenchymal
involvement by any of these can be
irregular and patchy, leaving other
parenchymal areas unaffected.
Clinical management of patients with
diffuse liver disease requires tools to
accurately detect and classify the various
forms of liver disease. Even with decades
of experience in imaging, liver biopsy and
the histological workup of the specimens
have traditionally been the reference
standard in the characterization of liver
disease [3]. However, biopsy is prone to
sampling errors if less affected paren-chyma
is sampled and may not reflect the
true disease severity and distribution in a
particular organ due to the variance in
the heterogeneous pattern of histological
changes on a macroscopic scale [4, 5].
Biopsy, associated with the risks of an
invasive procedure, is employed for dis-ease
detection and staging, but periodi-cally
repeated biopsy is not a practical
1 Results of the
Screening Dixon tech-nique
which produces
color coded maps to
visualize the distribution
of detected abnormal
metabolites in two dif-ferent
clinical examples.
(1A) A patient with dif-fuse
hepatic steatosis as
indicated by the yellow
hue. (1B) A patient with
diffuse iron overload as
marked by blue overlay
to the affected liver.
MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world 111
Faster Abdominal
MRI Examinations by
Limiting Table Movement
Mustafa Rifaat Bashir1; Brian Marshall Dale2; Wilhelm Horger3;
Daniel Tobias Boll1; Elmar Max Merkle1
1 Radiology, Duke University Medical Center, Durham, NC, USA
2 Siemens Healthcare, Cary, NC, USA
3 Siemens Healthcare, Erlangen, Germany
1 How to create a minimized
shimming protocol.
Modifications will be made to pulse
sequences following the initial
localizers (1A).
For the second sequence in the
scan protocol, ‘Shim mode’ is set to
‘Standard’, and ‘Adjust with body
coil’ is selected (1B).
For the third and subsequent se-quences,
‘Positioning mode’ is set
to ‘FIX’ (1C). ‘Shim mode’ is set to
‘Standard’, ‘Adjust with body coil’ is
selected, and ‘Adjustment Tolerance’
is set to ‘Maximum’ (1D).
Finally, for the third sequence,
a reference is created to the table
position of the second sequence
(1E). The same reference is created
for all subsequent sequences (1F).
1A
1D
118 MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world
1089_Flash_52_Inhalt_CC.indd 118 02.04.13 10:45
Clinical Abdominal Imaging
6
configurable
delay times
6 Example of the Abdomen Dot Engine user interface showing the guidance view that allows global planning of
delay times within a dynamic contrast enhanced liver MRI protocol.
ity. Additionally, multiple scan types
which differ by only a few minor compo-nents
(e.g., with or without MR Cholan-giopancreatography
(MRCP), with or
without diffusion-weighted imaging)
can be combined into a single, efficient
protocol with a few key decision points,
reducing redundancy and allowing for
simpler base protocol maintenance and
modification when necessary.
Summary
MRI examinations face serious competi-tion
compared to sonography and CT
when categories such as robustness,
acquisition time, patient comfort and
health care costs are considered. An
abundance of information may be
acquired through high resolution imag-ing
and dedicated quantitative MRI
sequences, but images and measure-ments
should be reproducible and reli-able
in their diagnostic value. The redun-dancy
of preparatory steps for the
operator within an MRI protocol is an
opportunity for more efficient and less
time consuming imaging. In addition, the
image acquisition process can be
improved by means of faster imaging at
higher resolution with the implementa-tion
of new parallel imaging acceleration
techniques, to reduce the risk of motion
116 MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world
1089_Flash_52_Inhalt_CC.indd 116 02.04.13 10:45
How-I-do-it
Methods
Automated algorithms to minimize table
movement have already been incorpo-rated
into MAGNETOM MRI systems under
syngo MR D11 and later software versions.
Under earlier software versions, a few
simple steps can be performed to convert
a standard MRI protocol into a minimized
shimming protocol, in order to realize the
time savings previously described. These
changes can all be made via the Exam
Explorer (Fig. 1A).
Pulse Sequence #1 – localizer
The first pulse sequence of an examination
is a localizer, typically utilizing either a
three-plane TrueFISP or HASTE technique.
At the MRI console, under the ‘Sequence’
card, the Shim is set to ‘None’ (typically the
default value), and precalibrated prescan
data is used with no need to acquire new
prescan data. No additional modification
of this sequence is required.
Pulse Sequence #2 – first and only
table move
Using the image data from the localizer
sequence, the image volume for Pulse
Sequence #2 is prescribed. This volume
should be centered on the area of interest
and rather large, covering the volume of
interest for the entire examination; at our
in patients with limited breath-hold capa-bilities.
which self-optimize during the course of
the examination or use initial pulse
sequences to tailor subsequent
sequence selection, can provide faster
and more efficient examinations, which
include quantitative data when appro-priate.
improvements may equip liver MRI
examinations with sufficient tools to
remain unique in delivering disease spe-cific
their diagnostic value.
2B 2E
2D
institution, we typically use a coronal
HASTE sequence for this purpose. The
prescan data acquired in this step, includ-ing
shim data, will be carried forward for
the remainder of the examination. The
following modifications are made to this
pulse sequence:
1. In the ‘System’ card, under the ‘Adjust-ments’
tab, set ‘Shim mode’ to ‘Stan-dard’.
Check the ‘Adjust with body coil’
box (Fig. 1B).
Pulse Sequences #3 and higher – no
further table movements
For all subsequent pulse sequences, table
movement is disallowed, and prescan
adjustment data from Pulse Sequence #2
is carried forward, so that as little time
as possible is spent acquiring new adjust-ment
data. The following modifications
are made:
1. In the ‘System’ card, under the ‘Miscel-laneous’
tab, set ‘Position mode’ to ‘FIX’
(Fig. 1C).
2. In the ‘System’ card, under the ‘Adjust-ments’
tab: set ‘Shim mode’ to ‘Stan-dard’;
select ‘Adjust with body coil’; and
set ‘Adjustment Tolerance’ to ‘Maxi-mum’
(Fig. 1D).
3. From the Exam Explorer, right-click on
the sequence and select ‘Properties’.
120 MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world
Intelligent imaging protocols,
Combining all of the described
quantitative data while expanding
Under the ‘Copy References’ tab, check
the ‘Copy reference is active’ box, then
select Pulse Sequence #2 in the left-hand
window and ‘Table position’ in
the right-hand window (Fig. 1E). In
combination with step #2 above, this
ensures that the MRI system table will
not move when progressing to later
pulse sequences in the examination,
despite different prescriptions for the
imaging volume.
4. Repeat steps 1–4 for all subsequent
sequences (Figure 1F).
Discussion
Preparatory adjustments made by an MRI
system are essential to realize excellent
image quality. In particular, adequate
shimming is necessary to ensure magnetic
field homogeneity. Shimming is a process
whereby the main magnetic field (B0) is
fine-tuned to compensate for field fluctu-ations
and inhomogeneities introduced
by the presence of the human body within
the scanner. These adjustments are applied
specifically to a volume within the bore
of the magnet (based on the anticipated
imaging volume), attempting to optimize
magnetic field homogeneity within that
volume while sacrificing field homogene-ity
outside of the volume.
2A
2C
1089_Flash_52_Inhalt_CC.indd 120 02.04.13 10:46
For the whole range of clinical MR information visit us at
www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-32-2048.jpg)
![Clinical Neurology Neurology Clinical
T1-weighted Phase Sensitive Inversion
Recovery for Imaging Multiple Sclerosis
Lesions in the Cervical Spinal Cord
MS lesions in the brain. The FLAIR
technique is a T2-weighted sequence
with a long TR and TE and is used to
demonstrate the changes in T2 relaxation
times in lesions when compared
to normal tissue.
As the name indicates,
the signal
of cerebro-spinal
fluid (CSF) is attenuated.
CSF has a long T1 relaxation times
compared to the other tissues in both
the brain and cervical spine. Therefore,
a rather long inversion time is
needed to null the signal of CSF
(~ 2500 ms). Hence, the contrast in
T2-weighted FLAIR images allows
for easier assessment of (MS) lesions,
especially when the lesions are close
to CSF, as compared to normal
T2-weighted images. However, while
the FLAIR technique works well in
the brain, it is hampered by flow and
motion artifacts when used in the
cervical spine.
Double Inversion recovery
The Double Inversion Recovery technique
has been implemented in the
SPACE-DIR sequence in the Siemens
syngo MR D13 software. A protocol
optimized for brain imaging is also
provided. SPACE-DIR is a T2-weighted
technique which uses two inversion
pulses, combined with a fat saturation
pulse, to null both the signal of
CSF and normal white matter. Similar
to the FLAIR technique, this sequence
is used to exploit the changes in T2
relaxation times in lesions when compared
to normal tissue. In the brain,
SPACE-DIR improves visualization of
Bart Schraa, MSc., Senior MR Application Specialist
Inversion Recovery
Sequences used for imaging
Multiple Sclerosis
Several inversion recovery techniques
are used for imaging lesions in MS.
Among these are Fluid Attenuated
Inversion Recovery (FLAIR), Sampling
Perfection with Application optimized
Contrasts using different flip-angle
Evolutions Double Inversion Recovery
(SPACE-DIR), and T1-weighted Phase
Sensitive Inversion Recovery (PSIR).
Fluid Attenuated
Inversion Recovery
FLAIR is commonly used to assess
white matter lesions and in particular
Siemens LTD Canada
Introduction
Multiple sclerosis (MS) is an inflammatory
disease in which the insulating
covers of nerve cells in the brain
and spinal cord are damaged. Magnetic
resonance imaging (MRI) was
first used to visualize multiple sclerosis
(MS) in the upper cervical spine in
late 1980 [1]. Spinal MS is often
associated with concomitant brain
lesions; however, as many as 20%
of patients with spinal lesions do not
have intracranial plaques [2]. This
article describes the experiences with
a T1-weighted phase sensitive inversion
recovery sequence for the detection
of MS lesions in the cervical
spinal
cord using the MAGNETOM
Skyra with syngo MR D13A software.
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Magnitude Reconstruction
110 160 210 260 310 360 410 460 510 560 610
Normal Cord Lesion Inversion time
Relative signal intensity
the cortex and reveals cortical lesions
as hyperintense relative to normal
surrounding
gray matter. It also provides
a high contrast between white
matter lesions and the surrounding
normal white matter. Initial studies
have investigated the applicability of
DIR for lesion imaging in the spinal
cord with positive results [3]. Nevertheless,
while SPACE-DIR provides a
high contrast and isotropic voxels, its
rather long acquisition time (~8 min)
may prove challenging within a clinical
setting.
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
T1-weighted phase sensitive
inversion recovery
A promising potential alternative for
imaging MS lesions in the cervical
spinal
cord [4], is the T1-weighted
true or phase sensitive inversion
recovery (PSIR) sequence. This technique
has been used to detect MS
lesions both in white and cortical gray
matter in the brain [5, 6]. This
sequence exploits the differences in
T1 relaxation times of tissues rather
than the differences in T2 relaxation
times as for both FLAIR and SPACE-DIR.
Contrast
Since the inversion time used is
chosen
such that it nulls the signal of
normal white matter (~350–400 ms
@ 3T), normal white matter is displayed
as intermediate gray. All other
tissues will have either lower or higher
signal intensity than normal white
matter depending on their T1 relaxation
time relative to normal white
matter. This provides a high contrast
between MS lesions and surrounding
tissue. Moreover, because PSIR uses
a short TE, it is less sensitive to flow
artifacts. High resolution imaging can
also be achieved within reasonable
scan times.
Based on these advantages,
T1-weighted PSIR is now being
explored for the detection of MS
lesions in the cervical spinal cord.
Reconstruction methods
The T1-weighted PSIR images can
be reconstructed as a magnitude or a
phase sensitive (real) image (Fig. 1).
Magnitude reconstruction
The magnitude reconstruction does
not consider the sign of the signal.
Therefore, the tissue which is nulled
by the inversion time will have a
signal
intensity of zero and all other
tissues will have higher signal intensity
(ranging from 0 to +4096), regardless
of whether they have shorter or
longer T1 relaxation time than the
nulled tissue (Fig. 2). However, there
is a range of inversion times where
the contrast between two different
Signal behavior in an inversion recovery sequence using
magnitude reconstruction.
2
Parameter Card for choosing magnitude or phase sensitive (real)
reconstruction method in an Inversion Recovery sequence.
1
1
2
Contrast behavior in an inversion recovery sequence
using magnitude or phase sensitive (real) reconstruction.
3
4
T1-weighted
PSIR images
using (4A)
magnitude and
(4B) phase
sensitive (real)
reconstructions.
4A 4B
Magnitude Real Inversion time
Relative signal intensity
110 160 210 260 310 360 410 460 510 560 610
3
64 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 65](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-33-2048.jpg)
![Pictorial Essay
Benign and Malignant Bone Tumors:
Radiological Diagnosis and
Imaging Features
Katharina Grünberg, M.D.; Christoph Rehnitz, M.D.; Marc-André Weber, M.D., M.Sc.
Section Musculoskeletal Radiology, Diagnostic and Interventional Radiology,
University Hospital Heidelberg, Germany
Topics
The learning objectives of this review
article are to identify benign vs. malig-nant
criteria in bone tumor diagnosis
and also to differentiate the types of
bone tumors and their characteriza-tion.
Based on the Lodwick classifica-tion
an overview of the three main
types of bone destruction patterns
visible on radiographs will be given
with many examples. Typical examples
of benign and malignant bone
tumors will be demonstrated, the
various imaging modalities will be
compared, and their utility will be dis-cussed.
The image gallery comprises
pearls and pitfalls. Presentation of
standardized magnetic resonance
imaging (MRI) protocols will be given.
Of course, this pictorial essay does
not have the focus of comprehensively
presenting all bone tumor entities.
Introduction
Primary bone tumors are categorized
according to their tissue of origin
into cartilage, osteogenic, fibrogenic,
fibrohistiocytic, haematopoietic, vas-cular,
lipogenic tumors and several
other tumors, like Ewing sarcoma and
giant cell tumor [1]. Thy are also
classified as either benign, malignant
or semi-malignant, as well as tumor-like
lesions [2]. They are rare, but found
on radiographs during an investigation
of a painful skeletal region or incidentally,
e.g. when performing a joint or whole-body
MRI. You will need four diagnostic
columns to make a diagnosis of a bone
tumor.
1. Malignant vs. benign?
X-rays:
Aggressiveness: Analysis of growth rate
(Lodwick classification),
periosteal reaction?
Further imaging modality CT, MRI?
Make a specific diagnosis:
2. Analysis of tumor matrix: X-rays, CT:
Osteolytic,
osteoblastic, mixed
3. Location within the tumor-bearing bone:
Epi-, meta-, diaphysis
4. Patient’s age, (affected bone)
in 80% correct specific diagnosis [4]
Four diagnostic columns
Four diagnostic columns (Fig. 1)
Tumor’s aggressiveness
The radiograph is the first method to
distinguish benign from malignant
lesions: at first by analysing the aggres-siveness
(analysis of growth rate) of
a lesion according to the classification
of Lodwick [3]. In radiographs there is
a correlation between bone tumor‘s
growth rate and dignity. If you identify
an aggressive growth pattern and/or
malignant periosteal reaction another
imaging modality like computed
tomography (CT) or magnetic reso-nance
imaging (MRI) is needed.
MRI is important for defining the
extension of tumor before biopsy.
Tumor matrix
In a second step, it is essential to
analyze the mineralisation of tumor
matrix in radiographs or CT. The
matrix may be osteolytic, osteoblastic,
or mixed, i.e. osteolytic with matrix
mineralisation.
Lodwick classification (Fig. 2)
Based on the Lodwick classification,
an overview of the three main types
of bone destruction patterns visible
on radiographs are given with a repre-sentative
example:
Type 1: geographic (with a: well-defined
border with sclerotic rim, b:
well-defined and sharp border but
without sclerotic rim, c: ill-defined
and blurred border);
Type 2: geographic with moth-eaten
or permeated pattern (patchy lysis);
Type 3: small, patchy, ill-defined
areas of lytic bone destruction with
moth-eaten or permeated pattern
(patchy lucencies) [3, 6].
Lodwick classification
IA IB IC II III
Non-ossifying fibroma Aneurysmal bone cyst Giant cell tumor Ewing’s sarcoma Osteosarcoma
Lodwick classification: An o 2 verview of the three main types of bone destruction patterns with representative image examples.
1
Tumor’s aggressiveness
Tumor matrix
Tumor location
Patient’s age
1 The four diagnostic columns needed to achieve a correct, specific diagnosis in about 80% of cases [4].
Tumor location and patient age
To make a specific tumor diagnosis,
the location within the tumor-bear-ing
bone (epi-, meta- and diaphysis)
and the patient’s age are also impor-tant.
With an optimized combination
of the different parameters, the expert
achieves a correct, specific diagnosis
in about 80% of cases [4, 5]. In other
words, even a dedicated musculo-skeletal
radiologist fails to predict the
correct histological diagnosis in one
fifth of all cases.
2
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 71
Clinical Orthopedic Imaging
70 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
Geographic,
well-defined
& sclerotic rim
Geographic
well-defined
& sharp border
but without
sclerotic rim
Geographic but
blurred border
Geographic
& moth-eaten
damage
with patchy lysis
Permeated lytic
damage with
small patchy
lucencies](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-36-2048.jpg)
![Criteria of malignancy (Fig. 3)
Periosteal reactions are also indicators
of lesion aggressiveness and can
be differentiated according to a benign
(thick, dense, wavy) type or an
aggressive (lamellated, amorphous,
sunburst) type. Figures
3A–D show
an example of an 80-year-old man
with a cloudy inhomogeneous tumor
of the distal humerus with perpendicular
periosteal reaction of a malignant
sunburst type, partial cortical destruction
and a big soft tissue
component,
best seen in MRI. All these criteria
suggest a malignant process. Differential
diagnoses are osteosarcoma
or bone metastasis. Biopsy results in
the diagnosis of metastasis of rectal
cancer (adenocarcinoma).
Types of bone tumors
According to their type of matrix
(osteolytic, osteoblastic, or osteolytic
with matrix mineralization) and to
their tissue of origin, bone tumors
are categorized into different types:
osteoid, chondroid, fibrous, lipoid/
fatty, other, cystic (solitary bone cyst,
aneurysmal bone cyst), vascular
(hemangioma), special cell type: Giant
cell (osteoclastoma), small cell
(Ewing’s sarcoma), histiocytes (eosinophilic
granuloma), plasma cells
(multiple myeloma), notochordal
cells (chordoma) and metastases.
Osteoid type
Osteoid osteoma
and Osteoblastoma (Fig. 4)
This entity is frequent: around 13.5%
of all benign bone tumors are osteoid
osteomas. The patients are usually
younger than 30 years and suffer „night
pain relieved by aspirin“ and other
platelet aggregation inhibitors. The
main location is in more than 50%
within diaphysis of long bones and in
10% within the vertebral column with
painful scoliosis. Osteoid osteomas
show in CT and X-ray a perifocal sclerotic
lesion with a central lucency
(nidus) that is cortically based in 80%.
Medullary, subperiosteal and articular
locations also occur. Calcification
of
the nidus is possible. The nidus is
extremely vascular in contrast-enhanced
MRI and it is important to
identify the nidus as the tumor itself;
surrounding sclerosis and bone marrow
edema pattern is just reactive. It
should be noted that lesions may
have less or no sclerosis if the nidus
is located in the marrow or in/adjacent
to a joint (Fig. 5). Osteoid osteoma
resembles osteomyelitis: For
example if a Brodie’s abscess is in an
eccentric position, e.g. cortically
located, it is difficult to differ Brodie’s
abscess from osteoid osteoma. The
differentiation can then only be done
by biopsy or radionuclide bone scan:
Osteoid osteoma shows – in contrast
to osteomyelitis – the ‘double density
sign’ (i.e. a high intense central activity
surrounded by an area of medium
activity). A lesion larger than 1.5 cm
is called osteoblastoma [7, 11].
Radiofrequency ablation (RFA) is
a successful treatment [8, 9, 10].
Keys to diagnosis: Sclerotic lesion
with a small lucency in X-ray.
The nidus shows a high signal on
T2-weighted MR images and has a
strong contrast-enhancement.
3
80-year-old man
with a cloudy
inhomogeneous
tumor of the left
distal humerus.
(3A) Radiograph
with lateral
projection shows
the perpendicular
periosteal reaction
of a malignant
sunburst type of
the humerus with
partial cortical
destruction (yellow
arrow). (3B) The
corresponding
antero-posterior
radiograph shows
the partial cortical
destruction and
a big soft tissue
component
(orange arrows),
best demonstrated
in MRI (orange
arrows). (3C)
Sagittal T1w, (3D)
sagittal PDw with
fat saturation.
3A 3B
3C 3D
4A 4B
16-year-old male patient with osteoidblastoma (OB) of the right fibula. (4A) lateral radiograph shows well the cortical swelling
of the fibula in the patient with OB but without lucency. The reason for that is best seen in 4B, C (axial CT in different positions)
and 4D (coronal CT), where the upper part of the nidus is completely calcified whereas the smaller lower part shows only little
calcification (*). (4E) shows a coronal CT with the ablation cannula (*) in the upper calcified part of the nidus during CT-guided
radiofrequency ablation and the second ablation channel thereunder (#). T2w STIR images (4F axial and 4G coronal) as well
as T1w axial images (4H, I) demonstrate the vascularisation of the lower and the calcification of the upper nidus part with low
T2w signal.
4
4C
4F
4D
4G
4E
4H
4I
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 73
Clinical Orthopedic Imaging
72 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-37-2048.jpg)
![5A 5B 5C
5D 5E 5F
17-year-old male patient with articular osteoid osteoma (OO) of the left knee joint. (5A–C) Axial, sagittal and coronal CT with the
articular position of the OO show no sclerosis of the nidus-margin (orange circle). In CT you see well that the nidus shows some
central ossifications. The axial T2w MRI with fat saturation (5D) demonstrates the joint effusion and synovitis (yellow arrow).
You can also see well that because of the central nidus calcifications the OO has only isointense to less hyperintense signal in T2w
(orange circle) instead of the typical strong hyperintense signal. (5E) Sagittal PDw MRI also shows the calcification. (5F) Axial
contrast-enhanced T1w MRI with fat saturation demonstrates the enhancing nidus (orange circle).
Osteosarcoma
The patients are usually younger
than 20 years. A 2nd peak exists in the
5th decade and these cases are mostly
secondary in Paget‘s disease and after
irradiation. Osteosarcoma has a predilection
for sites of rapid bone growth,
usually the metaphyseal region.
Typical symptoms are pain and local
swelling. This entity shows typically
destructive periosteal reactions as
mentioned above (Fig. 6). Their X-ray
morphology is very variable: Osteosarcomas
may be osteogenic (i.e. the
tumor induces new bone formation),
lytic or mixed, which is the common
manifestation form (Fig. 7) [12]. If
such a lesion is lytic, consider also
teleangiectatic osteosarcoma! From
origin, sclerosis grade and soft tissue
component, osteosarcomas are separated
into a central, parosteal (originates
from the periosteum) and a
periosteal variant, which is very rare
(1% of osteosarcomas). In periosteal
osteosarcomas the process starts
either in the periosteum or adjacent
soft tissue. Typical – in contrast to
parosteal osteosarcoma – the periosteal
osteogenic sarcoma does not
have large amounts of calcification in
the soft tissue (Fig. 8) [11]. Osteosarcomas
may produce osteoblastic lung
metastases (Fig. 9).
Keys to diagnosis are to detect criteria
of malignancy in X-ray and further
imaging modalities: CT is the best for
identifying periosteal reaction versus
tumor matrix because you can already
see faint mineralization in CT. In MRI
the signal depends on the degree of
matrix mineralization. But MRI is important
for assessing the tumor extent and
for staging purposes, i.e. to identify
skip-lesions, to assess the soft tissue,
nerve and vessel involvement, and a
potential joint infiltration.
5
6A 6B 6C
21-year-old man with a central high-grade osteosarcoma in the distal left femur. Conventional osteosarcomas are the central
osteosarcomas placed in the center of the metaphysis. Figure (6A) shows the antero-posterior and (6B) the lateral radiograph.
In this case you can see in addition to periosteal reactions (orange arrows in 6A) the channel-shaped lucency in the radiograph
correlating with the biopsy channel (yellow arrow in 6B) within a disorganization of the bone pattern and osteoid formation
(orange circle in 6B). You also see the biopsy channel in the coronal T1-weighted MRI (orange arrow in 6C). Figure 6D demonstrates
the heterogeneity of the tumor mass (axial T2w MRI with fat saturation). Performing MRI is important for preoperative
local staging, e.g. in this case the vessel infiltration (orange arrows in 6E) is visible in the axial post-contrast T1-weighted MRI.
6
6D 6E
7A 7B 7C
These images demonstrate well the difference between osteogenic versus lytic osteosarcoma. Figures 7A (antero-posterior
radiograph
of the lower leg) and 7B (sagittal contrast-enhanced T1w MRI with fat saturation of the lower leg) show an osteogenic
parosteal osteosarcoma of the left tibia, a bone forming tumor with fluffy, amorphous, cloudlike mineralization (orange arrow
in 7A) beside sunburst periosteal reaction as a criterion of malignancy (* in 7A). This tumor has a big soft tissue component
(yellow arrow in 7B) with large amounts of calcification (yellow arrow in 7A). Figures 7C (antero-posterior radiograph of the
pelvis) and 7D (axial contrast-enhanced T1w MRI with fat saturation) show a more lytic osteosarcoma of a 61-year-old female
patient in the right os ileum with no mineralization (orange arrow).
7
7D
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 75
Clinical Orthopedic Imaging
74 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-38-2048.jpg)
![8A 8B
8D
57-year-old female patient with a periosteal osteosarcoma (G3) of the right lower leg. Figure 8A, antero-posterior radiograph
of the right knee shows saucerization of the tibial metaphysic (orange circle) and also a bone prominence (yellow arrow).
8B (coronal T1w MRI of the lower right leg) and 8C (sagittal T1w MRI of the lower right leg) show the big inhomogeneous
tumor with a large soft tissue component. Keep in mind that the periosteal osteogenic sarcoma does not have large amounts
of calcification in soft tissue as shown in 8A–C (orange circle). Figures 8D (axial T2w MRI with fat saturation) and 8E (axial
contrast-enhanced T1w MRI with fat saturation) clearly demonstrate that the tumor inexplicably will not invade the medullary
space of the tibia (orange circle) and that the fibula is not involved (yellow arrow).
63-year-old female with osteoblastic lung metastases one year after resection of an osteosarcoma of the left thigh. Axial CT
images (9A–C) show several osteoblastic lung metastases of a bone producing primary tumor: an osteosarcoma. Therefore the
lung metastases may be also sclerotic.
8
9
Chondroid type
Enchondroma
Enchondroma is a benign lytic lesion
typically placed in the hand and chiefly
centrally located, often with endosteal
scalloping. It must have calcification
except in the phalanges (Fig. 10).
A typical size of enchondroma is around
1–2 cm; low grade chondrosarcoma is
larger than 4–5 cm. The enchondroma
shows no periosteal reaction. An
important differential diagnosis is the
bone infarction (Fig. 12).
Keys to diagnosis are: In T1-weighted
MR imaging the lesion has a low
signal.
The T2-weighted signal
depends on the degree of calcification.
After contrast-enhancement
the tumor shows in T1-weighting
MR imaging a lobulated appearance
with septa (Figs. 10 and 11).
Suspicious of malignancy in chondroid
tumors are pain, a size larger than
5 cm, the presence of a soft tissue mass
and a growing surrounding edema
on T2-weighted images.
Multiple enchondromas occur on
occasion, a condition called Ollier‘s
disease. This is not hereditary and
with no increased rate of malignant
degeneration.
By contrast, Maffucci‘s syndrome is
a condition with multiple enchondromas
associated with soft tissue hemangiomas.
Maffucci‘s syndrome is
likewise not hereditary, but is characterized
by an increased incidence
of malignant degeneration of the
enchondromas [11].
9A 9B
8C
8E
9C
10 A 10 B
29-year-old female with an enchondroma of the phalanx D1. (10A) Lateral radiograph of the right D1 shows the lytic lesion in
the proximal metacarpus of D1 which is hardly to identify and without sclerotic rim, according to a Lodwick IB lesion (orange
circle) and without calcifications. (10B-F) show the typical signal characteristics of an enchondroma in MRI (orange circles) and
that the lesion is smaller than 2 cm: low signal in T1-weighted imaging (10B, coronal), high signal in T2-weighted imaging
because of absent calcification as shown in 10A (10E, axial T2w MRI with fat saturation). Coronal contrast-enhanced T1w MRI
(10C), coronal T1w MRI subtraction (10D), and axial contrast-enhanced T1w MRI with fat saturation (10F), show that the tumor
has a lobulated appearance with septa.
10
10 E
10 C
10 F
10 D
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 77
Clinical Orthopedic Imaging
76 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-39-2048.jpg)
![Chondroid type
Osteochondroma (Fig. 13)
A synonym for osteochondroma is
cartilaginous
exostosis. It is a common
benign tumor of the extremities
(10%–15% of all bone tumors) and
is located in 50% of cases in the lower
extremities, in 10–20% in the humerus,
but rarely in the spine. For the diagnosis
it is important to identify the continuation
13 B
20-year-old male patient with an osteochondroma of the proximal right humerus and a loose body within joint space. (13A)
Axial radiograph of the right shoulder shows the sharp-bordered tumor of the humerus, (13B) coronal CT shows the osteochondroma
of the right shoulder, (13C) sagittal T2w MRI shows another part of the osteochondroma with its cartilage cap, where
the cap seems to be much larger than 8 mm, see also 13F, (13D) axial T2w MRI with fat saturation shows the measurement of
the T2w hyperintense cartilage cap with a distance of 8 mm, (13E) coronal T1w MRI shows the cartilage cap that is hypointense
in T1w, (13F) axial contrast-enhanced T1w MRI with fat saturation show the same part of figure 13C with a chondroid lobated
pattern after contrast-enhancement, (13G) 9 MHz ultrasound shows the structure seen in 13C and G as a round non-cystic
structure. This lesion is with all imaging modalities suspicious of a low grade chondrosarcoma. After surgery and histologic
examination revealed it to be no more than an osteochondroma and a neighboring loose body within joint space with caplike
borders and nodose lobulated chondroid tissue with kept structure of the lobules.
13
of bone marrow and trabecular
bone structures into the exostosis
as well as the cartilage cap. The malignant
degeneration occurs mainly in
tumors near the trunk.
Key to diagnosis is a mushroom-like
tumor. The thickness of the cartilage
cap is 8 mm or more (threshold in our
institution, see also explanation in the
next chapter) (Fig. 13) [13]. Contrast
media is not needed to determine the
thickness of the cartilage cap, because
it is clearly visible on T2-weighted
images.
Osteochondroma
vs. chondrosarcoma
A malignant transformation is more
likely if the cartilage cap thickness is
8 mm or more, which is the threshold
of our clinic. Further publicized
threshold values are 1.5 cm according
to Murphey et al. [14] and 2.0 cm
according to Bernard et al. [15].
Proximity to trunk (location in the
pelvis with highest malignant
transformation rate!) and hereditary
multiple exostoses (autosomal
dominant inheritance) (Fig. 14) are
correlated with a higher risk of
malignant transformation (3–5% of
tumors develop into chondrosarcomas).
A further criterion
is a cartilage
cap growth, especially beyond
age 20. Note: It is extremely difficult
for either a radiologist or a pathologist
to differentiate a low-grade
chondrosarcoma from enchondroma
(Fig. 15).
12-year-old male patient with hereditary multiple exostoses. (14A) Axial radiograph of the left shoulder, (14B) antero-posterior
radiograph of the left shoulder, (14C) axial T2w MRI with fat saturation of the left humerus, (14D) sagittal T2w MRI with fat
saturation of the left humerus, (14E) lateral radiograph of the left knee, (14F) antero-posterior radiograph of the left knee, (14G)
coronal STIR MRI of the upper extremities. In 14A, B, E and F the continuation of bone marrow and trabecular bone structures
into the exostosis are clearly depicted. The cartilage cap can be well evaluated in T2-weighted images as seen in 14C, D and
also in 14G.
14
13 A
13 D 13 E 13 F
< 8 mm
13 C 13 G 14 A 14 E
14 C
14 B 14 F
14 D 14 G
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 81
Clinical Orthopedic Imaging
80 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-41-2048.jpg)
![Chondrosarcoma (Figs. 15–17)
Patients are mostly older than 40 years
and experience pain. Tumors are near
the trunk and have a chondroid
matrix. Chondrosarcomas are characterized
by slow growth. Primary
chondrosarcomas are lytic, per-meative
and destructive lesions
with calcification in 50%. Secondary
chondrosarcomas have a cartilage
cap‘s thickness larger than 8 mm as a
sign of malignant transformation of
an osteochondroma (see also the
comments to threshold value in the
last chapter) [13-15].
Keys to diagnosis are lytic, destructive
lesion with flocculent, snowflake or
popcorn calcification in patients older
than 40 years. MRI: soft tissue mass
or edema. The following criteria are in
favor of a chondrosarcoma as opposed
to an enchondroma: Pain, tracer
uptake in bone scan, growth, cortical
bone penetration.
15 B 15 C
31-year-old male patient with grade 1 chondrosarcoma of the right os ilium. (15A) The antero- posterior radiograph of the
right hip joint shows a geographic well-defined lytic lesion in the right acetabulum with a sharp border but without sclerotic rim
according to a Lodwick IB lesion (orange arrow), in the center there is some flocculent calcification. (15B) Coronal CT and (15C)
axial CT of the right hip show the lytic lesion with sharp border, thinned cortex and central punctuate calcification without
cortical destruction. (15D, E) Contrast-enhanced T1w MRI with fat saturation (coronal in 15D and axial in 15E) show a central
contrast-enhancement. Figure 15F shows a coronal STIR MRI with a high signal in the border area of the tumor.
15
Orthopedic Imaging Clinical
16 B 16 C 16 D
16 E 16 F 16G
33-year-old female with grade 2 chondrosarcoma of the left olecranon. 16A shows an antero-posterior radiograph of the left
olecranon with a Lodwick type IC lesion: geographic but blurred border (orange circle). Figure 16B shows the lateral radiograph of
the left olecranon and reveals a cortical destruction (orange arrow). (16C) Coronal T1-weighted MR image of the olecranon clearly
shows the intraosseous borders of the tumor (orange arrows). (16D) Coronal STIR MR image clearly shows the muscle edema
(orange arrow). (16E) Axial T2w MRI with fat saturation shows the chondroid matrix of the tumor (orange circle). (16F) Sagittal
T2w MRI shows the hypointense calcification (orange arrow) in the center of the chondroid tumor. (16G) Sagittal contrast-enhanced
T1w MRI with fat saturation shows the infiltration of the surrounding soft tissue (orange arrow).
16
58-year-old male patient with grade 3 chondrosarcoma of the right humerus. (17A) The antero-posterior radiograph of the right
humerus shows in addition to a patchy lysis pattern (Lodwick II) the cortex destruction (orange circle). (17B) Coronal STIR MRI
clearly shows the extension of this large amorphous lesion (size of 10 cm, long yellow arrow) and the soft tissue infiltration (small
yellow arrow). The tumor has a predominant chondroid matrix with low signal in T1w (17C coronal T1w MRI) and an inhomogeneous
high signal in T2w (17D sagittal T2w MRI) as a further hint of a high-grade chondrosarcoma. (17D) Also shows well the
cortex destruction and soft tissue infiltration (orange circle). (17E) Coronal contrast-enhanced MRI shows necrotic tumor areas
within the tumor (yellow arrows). Also areas without chondroid matrix are a hint of a high-grade chondrosarcoma.
17
15 A
15 D 15 E 15 F
16 A
17 A 17 B 17 C 17 D 17 E
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 83
Clinical Orthopedic Imaging
82 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-42-2048.jpg)
![Chondroblastoma (Fig. 18)
Patients are usually younger than
20 years (i. e. skeletally immature
patients). The lesion must be located
epiphyseally and is rare in metaphysis.
This entity also occurs in carpal
and tarsal bones and rarely in the
patella (which with regard to the
18 C 18 D
12-year-old female patient with a chondroblastoma of the right lateral tibia epiphysis. (18A) Antero-posterior radiograph of the
right knee shows the epiphyseal located lytic lesion of the tibia with discreet sclerotic rim and some central located calcification
(orange circle). (18B) Axial CT shows the puncture needle in that lytic lesion having a sharp border (orange arrow). (18C) Coronal
and (18D) sagittal T1-weighted MR images show the bordered lesion having a discreet hypointense sclerotic rim and a central
hypointense punctual calcification (orange circle). (18E) Axial PD-weighted MR image shows a chondroid component with high
signal and central calcification with low signal (orange circle). (18F) Coronal STIR MR image also shows a chondroid component
with high signal and central calcification with low signal (orange circle) and a bone marrow edema in the circumference (orange
arrows). Therefore biopsy was performed.
18
differential
diagnosis of lytic lesions
behaves like an epiphysis) [11].
Usually
it appears in long bones and
shows in 40–60% calcification.
Differential
diagnoses are the sequestrum
(osteomyelitis) and the eosinophilic
granuloma.
Key to diagnosis: Chondroblastomas
are lytic epiphyseal lesions with sclerotic
rim. In MRI it shows a chondroid component
with high signal in T2-weighted
imaging and calcification with low
signal in T2-weighted imaging [4].
Fibrous type
Non-ossifying fibroma – NOF
(Figs. 19 and 20)
Patients are usually younger than
20 years and have no pain or periosteal
reaction. This lesion is located in the
metaphysis of long bone in eccentric
position and emanates from the cortex,
so that the cortex will be replaced with
benign fibrous tissue. Non-ossifying
fibromas ‘heal’ with sclerosis and disappear
in the following years. Lesions
smaller than 3 cm in length are called
fibrous cortical defect and lesions
larger than 3 cm in length are called
non-ossifying fibroma.
Key to diagnosis: A lytic lesion with
expansive growth and scalloped, well-defined
sclerotic border. The MRI
appearance of an NOF is somewhat
variable. Although they are essentially
always low signal on T1-weighted
MR imaging, they can have high or
low signal on T2-weighted imaging.
NOF has partly homogeneous or
partly non-homogeneous contrast-media
enhancement. During the
‘healing period’ the non-ossifying
fibroma can be hot on radionuclide
bone scans indicating the osteoblastic
activity.
19
14-year-old male patient with
a non-ossifying fibroma of
the left distal tibia. (19A)
Antero-posterior radiograph
shows a classic example of
a non-ossifying fibroma that
is slightly expansile and lytic
and has a scalloped, well-defined
sclerotic border
(Lodwick IA, orange circle).
(19B) Coronal T1-weighted
MR image shows the typical
low signal of the lesion
(orange circle). (19C) Axial
T2-weighted MR image shows
in this case a low signal in
T2-weighting what is variable
(orange arrow). (19D) Axial
contrast-enhanced
T1-weighted MR image with
fat saturation shows that in
this case the NOF has a partly
inhomogeneous contrast-media
enhancement (orange
arrow).
18 A
18 B 18 E 18 F
19 A
19 B
19 C 19 D
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 85
Clinical Orthopedic Imaging
84 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-43-2048.jpg)
![Fibrous dysplasia (Figs. 21–23)
Patients have usually no pain or periosteal
reaction. Fibrous dysplasia can
be either monostotic (most commonly)
or polyostotic (McCune-Albright syndrome)
and has a predilection for the
pelvis, the proximal femur, the ribs
and skull. In its classic description,
fibrous dysplasia has a, ‘ground-glass
appearance’ or ‘smoky appearing’ in
X-ray and/or CT (Fig. 21), but the
ground glass appearance is not always
present. Lesions may be mixed lytic
and sclerotic [11] and bone may be
deformed.
Keys to diagnosis are: No periosteal
reaction. Fibrous dysplasia shows lytic
lesions, as the matrix calcifies it has a
hazy, smoky and ground-glass look to
the point of sclerotic lesion. The signal
alterations of fibrous dysplasia in MRI
follow the uniform pattern of all tumors
(low signal in T1-weighted and intermediate
to high signal in T2-weighted
images). The fibrous tissue enhances
contrast media. If the lesion is located
in the tibia, consider also adamantinoma,
which has malignant potential,
i.e. a mixed lytic and sclerotic lesion in
anterior cortex of tibia that resembles
the fibrous dysplasia.
20 A 20 C
20 D 20 E
Image gallery of the non-ossifying fibroma.
(20A) Antero-posterior radiograph of the right knee of a 17-year-old male patient clearly shows the various
appearance of a NOF. Two healing periods can be seen: Lateral, a lesion with proceeding sclerosis indicating that
the lesion is in a progressed healing stadium (orange circle) and medial, a classical scalloped lesion with
well-
defined sclerotic border (orange arrows).
(20B) Antero-posterior radiograph of the left knee of an 18-year-old male patient shows the typical appearance
of a NOF: Scalloped lesion with well-defined sclerotic border.
(20C) Antero-posterior radiograph of the knee of a 12-year-old male patient shows a lytic lesion with sclerotic rim
(orange arrow) and below an exostosis (yellow arrow). (20D) Coronal T1-weighted MR image shows the typical
low signal of a NOF (orange circle) and (20E) a coronal STIR MR image shows in this case also a low signal
compatible to the diagnosis of a NOF.
20
21 A
21 E 21 F
21 G
21 H
21 B 21 C
21 D
20 B
21
44-year-old male patient with fibrous dysplasia
of the left femur.
(21A) Antero-posterior radiograph of the left femur
shows well the ground glass appearance of a sclerotic
lesion in the proximal diaphysis (orange circle). (21B)
Coronal CT, (21C) axial CT and (21D) 3D figure also
clearly show the ground glass appearance of that
lesion (orange circle). (21E) Coronal T1-weighted MR
image shows low signal of the lesion. (21F) Axial
T2-weighted MR image with fat saturation shows that
the lesion contains only point-shaped lipoid and
calcified parts (orange arrow). (21G) Sagittal
T2-weighted MR image shows in this case a homogenous
low signal, (21H) axial contrast-enhanced
T1-weighted MR image with fat saturation shows
a relatively homogenous contrast enhancing of the
lesion. An inhomogeneous contrast-enhancement
occurs in lesions with bigger parts of blood, fat and
calcifications leading to signal alterations.
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 87
Clinical Orthopedic Imaging
86 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-44-2048.jpg)
![22
Image gallery of fibrous
dysplasia: “Fibrous
dysplasia…can look like
almost anything!” [11],
as is clearly visible when
you compare the following
three cases.
(22A) Antero-posterior
radiograph of the pelvis
of a 36-year-old male
patient clearly shows that
the ipsilateral proximal
femur is always affected
when the pelvis is
involved with fibrous
dysplasia (orange circles).
The lesion in the pelvis
is more lytic than the
lesion in the femur which
is more sclerotic. (22B)
Antero-posterior radiograph
of the right knee
of a 33-year-old female
patient shows a circumscripted
lytic lesion of the
distal femur with smoky
parts (orange arrow).
(22C) Lateral radiograph
of the left lower leg of
a 22-year-old male patient
with a fibrous dysplasia
of the tibia shows a lytic
lesion in the tibia with
cortical destruction
(orange arrows). MRI and
biopsy were needed to
confirm the diagnosis.
Figures 22D–F show the
corresponding MRI
images to this case: (22D)
Coronal T1-weighted MR
image shows the classically
low signal of lesion.
(22E) Sagittal T2-weighted
MR image and (22F) axial
T2-weighted MR image
with fat saturation show
that the lesion is
inhomogeneous.
22 A
22 D 22 E
22 F
22 B 22 C 23 A 23 B 23 C
23 F
23 D 23 E
34-year-old male patient with a polyostotic fibrous dysplasia in pelvis and proximal femur (Albright-syndrome).
(23A) Antero-posterior radiograph of the pelvis, (23B) antero-posterior radiograph of the left femur and (23C) lateral radiograph
of the left femur show lots of lesions with smoky appearance in the right os ileum and the left femur (orange arrow).
(23D) Coronal T1-weighted MR image of the left femur shows a low signal of the lesions (orange arrow). (23E) Coronal STIR MR
image of the left femur shows an intermediate to high signal of the lesions (orange arrow) and (23F) axial contrast-enhanced
T1-weighted MR image shows an inhomogeneous contrast-enhancement of the fibrous tissue (orange arrow).
23
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 89
Clinical Orthopedic Imaging
88 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-45-2048.jpg)
![Lipoid/fatty type
Calcaneus lipoma (Fig. 24)
A common location is the calcaneus.
It is a rare entity and a so-called
‘leave-me-alone lesion’. Key to diagnosis:
Fat signal in all MRI sequences.
Other types
Solitary bone cyst (Figs. 25, 26)
Patients are usually younger than
20 years. Common location: calcaneus,
proximal humerus and femur with central
location of the lesion. Patients have
no pain or periosteal reaction unless
they suffer a fracture through this lesion.
The fracture often produces fragments
24 C
49-year-old male patient with an intra-osseous lipoma of the calcaneus as typical location. (24A) Lateral
radiograph of the calcaneus shows the geographic lesion with sclerotic rim, Lodwick IA (orange arrow). (24B)
Axial contrast-enhanced T1w MRI with fat saturation, (24C) coronal T2w MRI, (24D) sagittal T1w MRI and
(24E) coronal contrast-enhanced T1w MRI. MR images show well that the lesion contains fat, especially
seen in 24B and E (orange circle). Notice the synovial cyst between calcaneus and talus in 24C as auxiliary
diagnosis (orange arrow).
24
24 A
24 B
24 D 24 E
that sink to the bottom of the lesion,
well known as the ‘fallen fragment
sign’ visible on radiographs.
Key to diagnosis: Lytic centrally
located lesion, well-defined with
sclerotic rim (Lodwick type IA). The
MRI shows non-enhancing pure fluid
(in contrary to aneurysmal bone cyst).
Orthopedic Imaging Clinical
25-year-old male patient with a solitary bone cyst of the calcaneus. (25A) Lateral and (25B) antero-posterior radiographs
of
the calcaneus show both the geographic lesion with sclerotic rim, Lodwick IA (orange arrows). Typically for the location in the
anterior to the midportion of the calcaneus and on the inferior border is: only in this position the solitary bone cyst has
a characteristic triangular appearance.
25
25 B
If the lesion is located in the calcaneus
think about the differential
diagnosis of an intra-osseous lipoma.
A differentiation by X-ray is then only
possible if the lipoma has a central
calcification. But this differentiation is
not relevant, because both lesions are
‘leave-me-alone lesions’ [11].
25 A
Save the Date
3rd Heidelberg Summer
School
Musculoskeletal Cross Sectional Imaging 2014
July 25/26, 2014
Heidelberg, Germany
Please visit:
www.heidelbergsummerschool.de
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 91
Clinical Orthopedic Imaging
90 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-46-2048.jpg)
![Giant cell tumor (Fig. 28)
A precondition is that the epiphysis
is closed. This tumour often abuts
the articular surface and most often
has an eccentric localization. This
is often a well defined lesion with
a non-sclerotic margin (Lodwick IB).
Local aggressive growth and lung
metastasis in 5–10% occur.
Key to diagnosis: Osteolytic eccentric,
epiphyseal lesion without
matrix calcification and extensive
thinning of the cortex. The tumor
shows low signal in T1-weighting,
inhomogeneous or low signal in
T2-weighting and contrast-enhancement.
If the tumor contains necrosis
and hemosiderin, this results in an
inhomogeneous contrast-enhancement
pattern.
Ewing‘s sarcoma (Figs. 29, 30)
The classic Ewing‘s sarcoma is a, ‘permeative
lesion in the diaphysis of long
bone in a child’, [11], with osteodestruction
in CT and a very high signal
in T2-weighted imaging indicating
infiltration of bone marrow. The
location of Ewing‘s sarcoma tends to
follow the distribution of red marrow.
In histology small round blue cells
are visible. A large soft tissue mass
is possible. Important differential
diagnoses are osteomyelitis and
eosinophilic granuloma, which have
a benign periosteal reaction and
sometimes a sequestrum.
Keys to diagnosis are: A permeative
lesion or lesion with sclerotic and
29 D 29 E
29 F
16-year-old male patient with Ewing‘s sarcoma of the proximal left forearm. (29A) Bone scan shows an overview of the involvement
of the proximal radius, the ulna and parts of the distal humerus. (29B) Lateral radiograph of the forearm shows the onion-skinned,
multilamellated periosteal reaction of the proximal radius (yellow arrows). (29C) Three-phase radionuclide bone scan
with Tc-99m MDP shows the tracer uptake in the big tumor mass. (29D, E) Contrast-enhanced T1-weighted MRI with fat
saturation axial (D) and coronal (E) and (29F) coronal STIR MRI shows the big tumor involving radius and ulna.
29
3-year-old male patient with Ewing’s sarcoma of the left distal femur. (30A) Antero-posterior radiograph of the femur,
(30B) lateral radiograph of the femur, (30C) coronal contrast-enhanced T1-weighted MRI with fat saturation,
(30D) coronal T1-weighted MRI, (30E) coronal STIR MRI. Figure 30A clearly shows the Codman-triangle, whereby the elevated
periosteum forms an angle with the cortex, (orange circle) and 30B the onion-skinned periosteal reaction (yellow arrow).
Figures 30C–E show the T1w hypo-, T2w hyperintense signal character of the tumor with contrast-enhancement, the large
soft tissue component (yellow arrows) and also the Codman triangle.
30
patchy appearance and periosteal
reaction which can be onion-skinned
(multilamellated), sunburst or amorphous.
Low signal in T1-weighted
MR images, high signal in T2-weighted
MR imaging with strong contrast-enhancement.
More than 50% are
osteolytic lesions. Edema and large
soft tissue mass often occur.
29 A
28 A 28 B
29 B
29 C
30 A 30 B 30 C 30 D 30 E
28
34-year-old female patient with giant cell
tumor of the distal right femur. (28A)
Antero-posterior radiograph of the knee,
(28B) coronal CT of the knee, (28C) coronal
T1w MRI, (28D) T2w axial MRI, (28E)
coronal contrast-enhanced T1w MRI. Figure
28A shows the well-defined lesion with an
incomplete sclerotic rim (Lodwick IB) in the
distal femur (yellow arrow). 28B shows the
extensive thinning of the cortex in lateral
direction with a little cortex destruction
below (yellow arrow). 28C clearly shows the
low signal in T1w, 28D the low signal in T2w
(orange circle) and 28E an inhomogeneous
contrast-enhancement pattern because
of the necrosis area below (yellow arrow).
28 C
28 E
28 D
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 95
Clinical Orthopedic Imaging
94 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-48-2048.jpg)
![Multiple myeloma (Fig. 31)
In multiple myeloma, a proliferation
of monoclonal plasma cells within
the bone marrow occurs. The vertebral
column is mostly affected and
70% of patients are older than
60 years. Multiple lytic lesions in an
adult older than 40 years almost
always suggest metastases or multiple
myeloma. Bone sarcomas are
rare, and the most common cause of
a solitary destructive lesion in an
Metastases (Fig. 32)
40% of all metastases are located
in the vertebral column. The most
frequent
primary tumors are lung,
breast, prostate, renal cell, gastrointes-tinal
and thyroid carcinomas. Bone
marrow infiltration happens
before
osseous destruction. It is important to
pay attention to fractures,
spinal canal
invasion and myelon compression.
Key to diagnosis: For the diagnosis
of bone metastases a low signal in
T1-weighted MR images is more sen-sitive
than osteolysis in CT [24].
Osteolytic metastases have a high signal
in T2-weighted images, whereas
osteoblastic metastases have a low
to isointense signal in T2-weighted
images. Take into account these
factors in older patients and consider
several osteoblastic and/or osteolytic
lesions.
adult is a metastasis. Low-dose CT
is important for proving osteolytic
lesions and MRI [22] for proving bone
marrow affection: Decrease of
T1-weighted signal
in bone marrow
infiltration compared to the disks,
and a signal increase in the STIR
images compared to muscle tissue.
Whole-body MRI is suitable for demonstration
of the tumor burden. It is
important to think of patient’s age
when interpreting T1-weighted MR
imaging, because young patients
still have a cell-rich red bone marrow
and therefore also a low T1 signal.
We differentiate three patterns of bone
marrow infiltration: diffuse, multi-focal,
or ‘salt-and-pepper’ pattern.
Salt-and-pepper pattern indicates a
low grade disease stadium. A single
lesion is called plasmacytoma [22, 23].
65-year-old female patient with multiple myeloma and pain of the
backside. (31A) Antero-posterior radiograph shows osteopenic bone
pattern with several punched out lytic lesions. (31B) Coronal T1-weighted
MR image of the pelvis and (31C) coronal STIR MR image of the pelvis show
a solitary lesion with low signal in T1w and high signal in STIR suitable to a
focal lesion of the left os sacrum (yellow arrow).
31
32
55-year-old man with
osteoblastic metastases
of vertebral column and
pelvis in prostate cancer.
(32A) Lateral radiograph
of the vertebral column,
(32B) antero-posterior
radiograph of the lower
vertebral column and os
sacrum show several
osteoblastic metastases
of the vertebral bodies
and the os sacrum.
(32C) Scintigraphic bone
scan shows more skeletal
metastases.
31 A 31 B
32 C
32 A 32 B
31 C
Head R lat Head L lat
Thorax R V L Thorax L D R
Pelvis R V L Pelvis L D R
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 97
Clinical Orthopedic Imaging
96 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-49-2048.jpg)
![Summary
Role of X-ray
In addition to patient history and
clinical findings, a radiograph in two
orthogonal planes is still of great
importance for determining the
Lodwick classification and the tumor
matrix, whereas the bone matrix is
only poorly visualized in X-ray: You
cannot differentiate between lesions
containing fluid and solid lesions without
mineralized matrix. In general,
conventional X-ray radiography is the
starting point and CT and MR images
should only be interpreted with
concurrent radiographic correlation.
Role of CT
CT is superior to MRI for the assessment
of mineralized structures especially
cortical integrity, matrix mineralization,
and periosteal reactions [21].
Small lucency of the cortex, localized
involvement of the soft tissues, and
thin peripheral periosteal reaction can
Contact
Katharina Grünberg, M.D.
Section Musculoskeletal Radiology
Diagnostic and Interventional Radiology
University Hospital Heidelberg
Schlierbacher Landstraße 200a
69118 Heidelberg
Germany
Katharina.Gruenberg@med.uni-heidelberg.de
References
1 WHO classification of bone tumours 2006.
2 Vanel D, Ruggieri P, Ferrari S, Picci P,
Gambarotti M, Staals E, Alberghini M.
The incidental skeletal lesion: ignore or
explore? Cancer Imaging. 2009 Oct 2;
9 Spec No A: S38-43.
3 Lodwick GS, Wilson AJ, Farrell C, Virtama P,
Dittrich F. Determining growth rates of
focal lesions of bone from radiographs.
Radiology 1980; 134: 577-583.
4 Erlemann R. Basic diagnostics of bone
tumors. Radiologe 2009; 49: 257–267.
5 Oudenhoven LF, Dhondt E, Kahn S, Nieborg
A, Kroon HM, Hogendoorn PC, Gielen JL,
Bloem JL, De Schepper A. Accuracy of
radiography in grading and tissue-specific
diagnosis-a study of 200 consecutive bone
tumors of the hand. Skeletal Radiol. 2006;
35: 78-87.
6 Miller T. Bone tumors and tumorlike conditions:
analysis with conventional radiography.
Radiology 2008; 246: 662-674.
7 Lucas DR, Unni KK, McLeod RA, O’Connor
MI, Sim FH. Osteoblastoma: clinicopathologic
study of 306 cases. Hum Pathol.
1994 Feb;25(2):117-34.
8 Rehnitz C, Sprengel SD, Lehner B,
Ludwig K, Omlor G, Merle C, Kauczor HU,
Ewerbeck V, Weber MA. CT-guided radiofrequency
ablation of osteoid osteoma and
osteoblastoma: clinical success and long-term
follow up in 77 patients. Eur J Radiol
2012 Nov;81(11):3426-34. doi: 10.1016/j.
ejrad.2012.04.037. Epub 2012 Jul 6.
9 Rehnitz C, Sprengel SD, Lehner B,
Ludwig K, Omlor G, Merle C, Kauczor HU,
Ewerbeck V, Weber MA. CT-guided radiofrequency
ablation of osteoid osteoma:
correlation of clinical outcome and imaging
features. Diagn Interv Radiol. 2013
Jul-Aug;19(4):330-9. doi: 10.5152/
dir.2013.096.
be best seen with CT [16]. CT is the
examination of choice in the diagnosis
of the nidus of osteoid osteoma in
dense bone [17]. CT is valuable in the
diagnosis of tumors of the axial
skeleton such as spinal metastasis as
well as in systemic staging.
Role of MRI [18-21]
Without any radiation MRI can be helpful
while evaluating lesions that represent
a differential diagnosis dilemma
between benign and malignant lesions
before a biopsy. For example in aneurysmal
bone cysts MRI can display
fluid levels in blood filled cavities better
than CT. Another example, MRI before
biopsy for staging all suspected sarcomas
of bone could help identifying
extraosseous sarcoma better. MRI plays
an important role in planning limb salvage
surgeries because of its superior
role for soft tissue evaluation including
the presence or absence of neurovascular
invasion [21]. MRI helps by identifying
skip lesions and helps measure the thickness
of cartilage cap. The cap is thin in
benign lesions and thicker in chondrosarcomas
[14, 15]. This aids evaluation of
the entire compartment of long bones in
acceptable time. (Important here is a
large field-of- view of the MR sequence,
see Fig. 33.) This in turn helps to improve
the quality of life by reducing morbidity
without affecting survival. MRI is most
useful in evaluation of spine metastasis
differentiating osteoporotic and metastatic
compression fractures. In Multiple
Myeloma cases whole-body MRI scans
are suitable for demonstration of the
tumor burden. Though not yet in clinical
routine, newer techniques such as diffusion-
weighted imaging and DCE-MRI may
support assessment of tumor response.
More studies are being conducted.
10 Omlor GW, Lehner B, Wiedenhöfer B,
Deininger C, Weber MA, Rehnitz C.
[Radiofrequency ablation in spinal
osteoid osteoma. Options and limits].
[Article in German]. Orthopade. 2012
Aug;41(8):618-22. doi: 10.1007/
s00132-012-1907-x.
11 Clyde A. Helms, Fundamentals of Skeletal
Radiology (2005),3. Edition, Elsevier inc.
12 Murphey MD, Robbin MR, McRae GA,
Flemming DJ, Temple HT, Kransdorf MJ.
The many faces of osteosarcoma.
Radiographics
1997; 17: 1205-1231.
13 Kloth JK, Wolf M, Rehnitz C, Lehner B,
Wiedenhöfer B, Weber MA. [Radiological
diagnostics of spinal tumors. Part 1:
general tumor diagnostics and special
diagnostics of extradural tumors].
Orthopade. 2012 Aug;41(8):595-607.
doi: 10.1007/s00132-012-1978-8.
[Article in German].
14 Murphey MD, Choi JJ, Kransdorf MJ,
Flemming DJ, Gannon FH. Imaging of
osteochondroma: variants and complications
with radiologic-pathologic correlation.
Radiographics. 2000
Sep-Oct;20(5):1407-34.
15 Bernard SA, Murphey MD, Flemming DJ,
Kransdorf MJ. Improved differentiation
of benign osteochondromas from
secondary condrosarcomas with
standardized measurement of cartilage
cap at CT and MR imaging. Radiology
2010 Jun;255(3):857-65.
16 Brown KT, Kattapuram SV, Rosenthal DI.
Computed tomography analysis of bone
tumors: patterns of cortical destruction
and soft tissue extension. Skeletal Radiol
1986; 15: 448-451).
17 Glass RB, Poznanski AK, Fisher MR,
Shkolnik A, Dias L. MR imaging of
osteoid osteoma. J Comput Assist
Tomogr 1986; 10: 1065-1067.
Proposed tumor MRI protocol
Sequences unenhanced
coronal STIR with a large
field-of-view
coronal T1-weighted TSE
(turbo spin echo)
axial T2-weighted TSE with fat
saturation
sagittal T2-weighted TSE
Contrast-agent
(0.1 mmol/kg body weight)
axial contrast-enhanced T1-weighted
TSE with fat saturation
coronal contrast-enhanced
T1-weigthed TSE + subtraction
(contrast-enhanced minus native
T1-weighted MRI scan)
The contrast-enhanced sequences
are important in biopsy planning for
identifying necrotic and viable tumor
tissue. The biopsy should be targeted
to the viable tumor area.
Optional
MR-angiography
Dynamic T1-weighted contrast-enhanced
MRI (DCE-MRI)
Dynamic sequences are important for
biopsy planning to identify vital tumor
tissue [19-21], to which the biopsy
should be guided.
13-year-old female patient with Ewing’s sarcoma. (33A, B) Coronal STIR MR
images show the tumor in the left femur diaphysis (orange circle) with a large
soft tissue mass surrounded by a soft tissue edema (yellow arrows) and a skip
lesion in the femoral neck (orange arrow).
33
18 Anderson SE, Steinbach LS, Schlicht S,
Powell G, Davies M, Choong P. Magnetic
resonance imaging of bone tumors
and joints. Top Magn Reson Imaging.
2007 Dec; 18(6):457-65.
19 Fayad LM, Jacobs MA, Wang X, Carrini JA,
Bluemke DA. Musculoskeletal tumours:
how to use anatomic, functional, and
metabolic MR techniques. Radiology
2012; 265: 340-356.
20 Alyas F, James SL, Davies AM, Saifuddin
A. The role of MR imaging in the
diagnostic characterisation of appendicular
bone tumours and tumour-like
conditions. Eur Radiol 2007; 17:
2675-2686.
21 Roberts CC, Liu PT, Wenger DE. Musculoskeletal
tumor imaging, biopsy, and
therapies: self-assessment module. AJR
Am J Roentgenol. 2009; 193(6 Suppl):
S74-78.
22 Fechtner K, Hillengass J, Delorme S, Heiss
C, Neben K, Goldschmidt H, Kauczor HU,
Weber MA. Staging monoclonal plasma
cell disease: comparison of the Durie-
Salmon and the Durie-Salmon PLUS
staging systems. Radiology. 2010 Oct;
257(1):195-204. doi: 10.1148/
radiol.10091809.
23 Bäuerle T, Hillengass J, Fechtner K,
Zechmann CM, Grenacher L, Moehler
TM, Christiane H, Wagner-Gund B,
Neben K, Kauczor HU, Goldschmidt H,
Delorme S. Multiple myeloma and
monoclonal gammopathy of undetermined
significance: importance of
whole-body versus spinal MR imaging.
Radiology. 2009 Aug; 252(2):477-85.
doi: 10.1148/radiol.2522081756.
24 Bohndorf K, et al. Radiologische
Diagnostik der Knochen und Gelenke.
(2006) 2nd edition, Thieme.
33 A 33 B
Orthopedic Imaging Clinical
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 99
Clinical Orthopedic Imaging
98 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-50-2048.jpg)
![Combined 18F-FDG PET and MRI Evaluation
of a case of Hypertrophic Cardiomyopathy
Using Simultaneous MR-PET
Ihn-ho Cho, M.D.; Eun-jung Kong, M.D.
Department of Nuclear Medicine, Yeungnam University Hospital, Daegu, South Korea
Introduction
Hypertrophic cardiomyopathy (HCM)
is a common condition causing left
ventricular outflow obstruction, as
well as cardiac arrhythmias. Cardiac
MRI is a key modality for evaluation
of HCM. Apart from estimating left
ventricular (LV) wall thickness, LV
function and aortic flow, MRI is capable
of estimating the late gadolinium
enhancement in affected myocardium,
which has been shown to have a
direct correlation with incidence and
Patient history
A 25-year-old man presented to the
cardiology department with incidental
ECG abnormality after fractures to
his left 2nd and 4th fingers. Although
he had not consulted a doctor, he had
been suffering from mild dyspnea
with chest discomfort at rest and
exacerbation at exercise since May
2012. Echocardiography revealed
non-obstructive hypertrophic cardiomyopathy
(Maron III) with trivial MR.
The patient was referred for a simultaneous
MR-PET study for 18F-FDG
PET and cardiac MRI with Gadolinium
(Gd) contrast for evaluation of the
morphological and metabolic status
of the hypertrophic myocardium.
The patient was injected with 10 mCi
18F- FDG following glucose loading.
Simultaneous MR-PET study performed
on a Biograph mMR was
started one hour following tracer
injection. Following standard Dixon
sequence acquisition for attenuation
correction, the comprehensive cardiac
MRI sequences were acquired
including MR perfusion after Gd contrast
infusion, as well as post contrast
late Gd enhancement studies. Static
18F-FDG PET was acquired simultaneously
during the MRI acquisition.
Cardiovascular Imaging Clinical
2A 2
Discussion
The late Gd enhancement within
the hypertrophic septum along with
the non-uniform glucose metabolism
demonstrated by the patchy 18F-FDG
uptake within the hypertrophic septum
exactly corresponding to the area of
Gd enhancement reflect myocardial
fibrosis within the asymmetric septal
hypertrophy. Myocardial fibrosis and
the presence of late Gd enhancement
on MRI has been shown to be associated
with increased risk of cardiac
arrhythmia [1] as evident from the
symptoms of this patient.
severity of arrhythmias in HCM [1].
In patients with HCM, late gadolinium
enhancement (LGE) on CE-MRI is presumed
to represent intramyocardial
fibrosis. PET myocardial perfusion
studies have shown slight impairment
of myocardial blood flow with pharmacological
stress in hypertrophic
myocardium in HCM, presumably
related to microvascular disease [2].
18F-FDG PET has been sporadically
studied in HCM, mostly for evaluation
of the metabolic status of the
hypertrophic myocardial segment, especially
after interventions such as transcoronary
ablation of septal hypertrophy
(TASH) [3] or to demonstrate
partial myocardial fibrosis [4]. This
clinical example illustrates the value of
integrated simultaneous 18F-FDG PET
and MRI acquisition performed on the
Biograph
mMR system.
1A 1B
Short-axis views of end diastole and end systole at 3 different sections in the left ventricle obtained from gated TrueFISP cine
MRI acquisitions performed on Biograph mMR. Note the thick hypertrophic septum (white arrow), which demonstrates the
degree of asymmetric septal hypertrophy.
1
1D
1F
1C
1E
1G
1
1
End Diastole End Systole
2
3
2
3
Simultaneous MR-PET acquisition
provides combined acquisition of
both modalities, thereby ensuring
accurate fusion between morphological
and functional images due to
simultaneous PET acquisition for every
MR sequence. The exact coregistration
of the patchy 18F-FDG uptake
in the area of Gd enhancement within
the hypertrophic upper septum
reflects the advantage of simultaneous
acquisition.
End diastolic
and end
systolic views
of 2-chamber
and 4-chamber
views obtained
from gated cine
TrueFISP acquisitions
showing
thickness of
the asymmetric
septal hypertrophy
(white
arrow).
2C
2B
2D
End Diastole
4-chamber view 2-chamber view
End Systole
3
3
Static 18F-FDG
PET images
in short-axis,
horizontal
long-axis and
vertical long-axis
views
demonstrating
normal uptake
in the LV
myocardium
except the
non-uniform
uptake pattern
in the hypertrophied
septum
(white arrows).
LV cavity size
appears normal.
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 101
Clinical Cardiovascular Imaging
100 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-51-2048.jpg)
![Clinical Cardiology Cardiology Clinical
New Generation Cardiac Parametric Mapping:
the Clinical Role of T1 and T2 Mapping
T1 mapping
Initial T1 measurement methods were
multi-breath-hold. These were time
consuming and clunky, but were able
to measure well diffuse myocardial
fibrosis, a fundamental myocardial
property with high potential clinical
significance [1]. Healthy volunteers
and those with disease had different
extents of diffuse fibrosis [2], and
these were shown to be clinically
significant in a number of diseases.
T1 mapping methods based on the
MOLLI* approach with modifications
for shorter breath-holds, better heart
rate independence and better image
registration for cleaner maps, however,
transformed the field – albeit still
with a variety of potential sequences
in use [3-5]. There are two key ways
of using T1 mapping: Without (or
Viviana Maestrini; Amna Abdel-Gadir; Anna S. Herrey; James C. Moon
The Heart Hospital Imaging Centre, University College London Hospitals, London, UK
designed to optimize contrast
between ‘normal’ and abnormal –
a dichotomy of health and disease. As
a result, global myocardial pathologies
such as diffuse infiltration (fibrosis,
amyloid, iron, fat, pan-inflammation)
are missed.
Recently, rapid technical innovations
have generated new ‘mapping’ techniques.
Rather than being ‘weighted’,
these create a pixel map where each
pixel value is the T1 or T2 (or T2*),
displayed in color. These new
sequences are single breath-hold,
increasingly robust and now widely
available. With T1 mapping, clever
contrast agent use also permits the
measurement of the extracellular
volume
(ECV), quantifying the interstitium
(odema, fibrosis or amyloid),
also as a map. Early results with these
methodologies are exciting – potentially
representing a new era of CMR.
Introduction
Cardiovascular magnetic resonance
(CMR) is an essential tool in cardiology
and excellent for cardiac function
and perfusion. However, a key, unique
advantage is its ability to directly
scrutinize the fundamental material
properties of myocardium – ‘myocardial
tissue characterization’.
Between 2001 and 2011, the key
methods for tissue characterization
have been sequences ‘weighted’ to
a magnetic property – T1-weighted
imaging for scar (LGE) and T2-weighted
for edema (area at risk, myocarditis).
These, particularly LGE imaging, have
changed our understanding and clinical
practice in cardiology.
However, there are limitations to
these approaches: Both are difficult
to quantify – the LGE technique in
particular is very robust in infarction,
but harder to quantify in non-ischemic
cardiomyopathy. A more fundamental
difference is that sequences are
before) contrast – Native T1 mapping;
and with contrast, typically by subtracting
the pre and post maps with
hematocrit correction to generate
the ECV [6].
Native T1
Native T1 mapping (pre-contrast T1)
can demonstrate intrinsic myocardial
contrast (Fig. 1). T1, measured in milliseconds,
is higher where the extracellular
compartment is increased.
Fibrosis (focal, as in infarction, or diffuse)
[7-8], odema [9-10] and amyloid
[11], are examples. T1 is lower in
lipid (Anderson Fabry disease, AFD)
[12], and iron [13] accumulation.
These changes are large in some rare
disease. Global myocardial changes
are robustly detectable without contrast,
even in early disease. In iron, AFD
and amyloid, changes appear before
any other abnormality – there may be
no left ventricular hypertrophy, a normal
electrocardiogram, and normal
conventional CMR, for example – genuinely
new information. In established
disease, low T1 values in AFD appear
to absolutely distinguish it from other
causes of left ventricular hypertrophy
[12] whilst in established amyloid T1
elevation tracks known markers of
cardiac severity [11].
A note of caution, however. Native
T1, although stable between healthy
volunteers to 1 part in 30, is dependent
on platform (magnet manufacturer,
sequence and sequence variant,
field strength) [14]. Normal reference
ranges for your setup are needed.
Lowest ECV Tertile
Middle ECV Tertile
Highest ECV Tertile
p < 0.001 fortrend
p < 0.015 for Middle Tertile
compared to others
2 ECV in non scar areas (LGE excluded) is associated with all-cause mortality [21].
The signal acquired is also a composite
signal – generated by both interstitium
and myocytes. The use of
an extracellular contrast agent adds
another dimension to T1 mapping
and the ability to characterize the
extracellular compartment specifically.
Extracellular volume (ECV)
Initially, post-contrast T1 was measured,
but this is confounded by renal
clearance, gadolinium dose, body
composition, acquisition time post
bolus, and hematocrit. Better is measuring
the ECV. The ratio of change
of T1 between blood and myocardium
after contrast, at sufficient equilibrium
(e.g. after 15 minutes post-bolus – no
infusion generally needed) [15, 16],
represents the contrast agent partition
coefficient [17], and if corrected
for the hematocrit, the myocardial
extracellular space – ECV [1]. The ECV
is specific for extracellular expansion,
and well validated. Clinically this
occurs in fibrosis, amyloid and
odema. To distinguish, the degree of
ECV change and the clinical context
is important. A multiparametric
approach (e. g. T2 mapping or
T2-weighted imaging in addition)
may therefore be useful. Amyloid can
have far higher ECVs than any other
disease [18] whereas ageing has small
changes – near the detection limits,
but of high potential clinical importance
[19, 20]. For low ECV expansion
diseases, biases from blood pool
partial volume errors need to be meticulously
addressed. Nevertheless, even
modest ECV changes appear prognostic.
In 793 consecutive patients
(all-comers but excluding amyloid and
HCM, measuring outside LGE areas)
followed over 1 year, global ECV predicted
short term-mortality (Fig. 2)
* The product is currently under development;
is not for sale in the U.S. and other
countries, and its future availability cannot
be ensured.
1B 1C 3A 3B 3C
1
Native T1 maps of (1A) healthy
volunteer (author VM): the
myocardium appears homogenously
green and the blood is red; (1B)
cardiac amyloid: the myocardium
has a higher T1 (red); (1C)
Anderson Fabry disease: the
myocardium has a lower T1 (blue)
from lipid – except the inferolateral
wall where there is red from
fibrosis; (1D) myocarditis, the
myocardium has a higher T1 (red)
from edema, which is regional; (1E)
iron overload: the myocardium has
a lower T1 (blue) from iron.
1A
2
A patient with myocarditis. On the left side a native T1 map showing the higher T1 value in the inferolateral wall (1115 ms);
in the centre, a post-contrast T1 map showing the shortened T1 value after contrast administration (594 ms); on the right side
the derived ECV map showing higher value of ECV (58%) compared to remote myocardium.
3
1D 1E
100%
0%
Proportion Surviving
Years of Follow-up
1.0
0.9
0.8
0.7
0.6
0.5
0 0.5 1.0 1.5 2.0
104 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 105](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-53-2048.jpg)
![Clinical Cardiology Cardiology Clinical
Progress is rapid; challenges remain.
Delivery across sites and standardization
is now beginning with new
draft guidelines for T1 mapping
in preparation. Watch this space.
References
1 Flett AS, Hayward MP, Ashworth MT,
Hansen MS, Taylor AM, Elliott PM,
McGregor C, Moon JC. Equilibrium
Contrast Cardiovascular Magnetic
Resonance for the measurement of
diffuse myocardial fibrosis: preliminary
validation in humans. Circulation
2010;122:138-144.
2 Sado DM, Flett AS, Banypersad SM, White
SK, Maestrini V, Quarta G, Lachmann RH,
Murphy E, Mehta A, Hughes DA, McKenna
WJ, Taylor AM, Hausenloy DJ, Hawkins
PN, Elliott PM, Moon JC. Cardiovascular
magnetic resonance measurement of
myocardial extracellular volume in health
and disease. Heart 2012;98:1436-1441.
3 Piechnik SK, Ferreira VM, Dall’Armellina E,
Cochlin LE, Greiser A, Neubauer S,
Robson MD. Shortened Modified Look-
Locker Inversion recovery (ShMOLLI) for
clinical myocardial T1-mapping at 1.5
and 3 T within a 9 heartbeat breathhold.
J Cardiovasc Magn Reson 2010;12:69.
4 Messroghli DR, Greiser A, Fröhlich M,
Dietz R, Schulz-Menger J. Optimization
and validation of a fully-integrated pulse
sequence for modified look-locker
inversion-recovery (MOLLI) T1 mapping
of the heart. J Magn Reson Imaging
2007;26:1081–1086.
5 Fontana M, White SK, Banypersad SM,
Sado DM, Maestrini V, Flett AS, Piechnik
SK, Neubauer S, Roberts N, Moon JC.
Comparison of T1 mapping techniques for
ECV quantification. Histological validation
and reproducibility of ShMOLLI versus
multibreath-hold T1 quantification
equilibrium contrast CMR. J Cardiovasc
Magn Reson 201;14:88.
6 Kellman P, Wilson JR, Xue H, Ugander M,
Arai AE. Extracellular volume fraction
mapping in the myocardium, part 1:
evaluation of an automated method.
J Cardiovasc Magn Reson 2012;14:63.
7 Dass S, Suttie JJ, Piechnik SK, Ferreira VM,
Holloway CJ, Banerjee R, Mahmod M,
Cochlin L, Karamitsos TD, Robson MD,
Watkins H, Neubauer S. Myocardial tissue
characterization using magnetic resonance
non contrast T1 mapping in hypertrophic
and dilated cardiomyopathy. Circ
Cardiovasc Imaging. 2012; 6:726-33.
8 Puntmann VO, Voigt T, Chen Z, Mayr M,
Karim R, Rhode K, Pastor A, Carr-White G,
Razavi R, Schaeffter T, Nagel E. Native T1
mapping in differentiation of normal
myocardium from diffuse disease in
hypertrophic and dilated cardiomyopathy.
J Am Coll Cardiovasc Imgaging
2013;6:475–84.
Contact
Dr. James C. Moon
The Heart Hospital Imaging Centre
University College London Hospitals
16–18 Westmoreland Street
London W1G 8PH
UK
Phone: +44 (20) 34563081
Fax: +44 (20) 34563086
james.moon@uclh.nhs.uk
4A 4B
4C 4D
(4A) T2 mapping in a normal volunteer (author VM). (4B) High T2 value in patient
with myocarditis – here epicardial edema. (4C) Edema in acute myocardial
infarction – here patchy due to microvascular obstruction – see LGE, (4D).
4
[21]. The same group also found
(n ~1000) higher ECVs in diabetics.
Those on renin-angiotensin-aldosterone
system blockade had lower ECVs.
ECV also predicted mortality and/or
incident hospitalization for heart
failure
in diabetics [22].
The use and capability of ECV quantification
is growing. T1 mapping is
getting better and inline ECV maps
are now possible where each pixel
carries directly the ECV value (Fig. 3)
– a more biologically relevant figure
than T1 [6].
T2 mapping
T2-weighted CMR identifies myocardial
odema both in inflammatory
pathologies and acute ischemia, delineating
the area at risk. However, these
imaging techniques (e. g. STIR) are
fragile in the heart and can be challenging,
both to acquire and to interpret.
Preliminary advances were made
with T2-weighted SSFP sequences,
which reduce false negatives and
positives [23, 24]. T2 mapping seems
a further increment [25] (Fig. 4). As
with T1 mapping, global diseases such
as pan-myocarditis may now be identified
by T2 mapping, and preliminary
results are showing this in several
rheumatologic diseases (lupus, systemic
capillary leak syndrome) and
transplant rejection, detecting early
rejection missed by other modalities
[26, 27].
Conclusion
Mapping – T1, T2, ECV mapping of
myocardium is an emerging topic with
the potential to be a powerful tool in
the identification and quantification
of diffuse myocardial processes without
biopsy. Early evidence suggests
that this technique detects early stage
disease missed by other imaging
methods and has potential as a prognosticator,
as a surrogate endpoint
in trials, and to monitor therapy.
9 Ferreira VM, Piechnik SK, Dall’Armellina E,
Karamitsos TD, Francis JM, Choudhury
RP, Friedrich MG, Robson MD, Neubauer
S. Non-contrast T1-mapping detects acute
myocardial edema with high diagnostic
accuracy: a comparison to T2-weighted
cardiovascular magnetic resonance. J
Cardiovasc Magn Reson 2012; 14:42.
10 Dall’Armellina E, Piechnik SK, Ferreira VM,
Si Ql, Robson MD, Francis JM, Cuculi F,
Kharbanda RK, Banning AP, Choudhury
RP, Karamitsos TD, Neubauer S. Cardiovascular
magnetic resonance by non
contrast T1-mapping allows assessment
of severity of injury in acute myocardial
infarction. J Cardiovasc Magn Reson
2012;14:15.
11 Karamitsos TD, Piechnik SK, Banypersad
SM, Fontana M, MD, Ntusi NB, Ferreira
VM, Whelan CJ, Myerson SG, Robson MD,
Hawkins PN, Neubauer S, Moon JC.
Non-contrast T1 Mapping for the
Diagnosis of Cardiac Amyloidosis.
J Am Coll Cardiol Img 2013;6:488–97.
12 Sado DM, White SK, Piechnik SK,
Banypersad SM, Treibel T, Captur G,
Fontana M, Maestrini V, Flett AS, Robson
MD, Lachmann RH, Murphy E, Mehta A,
Hughes D, Neubauer S, Elliott PM,
Moon JC. Identification and assessment
of Anderson-Fabry Disease by Cardiovascular
Magnetic Resonance Non-contrast
myocardial T1 Mapping clinical
perspective. Circ Cardiovasc Imaging
2013;6:392-398.
13 Pedersen SF, Thrys SA, Robich MP, Paaske
WP, Ringgaard S, Bøtker HE, Hansen ESS,
Kim WY. Assessment of intramyocardial
hemorrhage by T1-weighted cardiovascular
magnetic resonance in reperfused
acute myocardial infarction. J Cardiovasc
Magn Reson 2012; 14:59.
14 Raman FS, Kawel-Boehm N, Gai N, Freed
M, Han J, Liu CY, Lima JAC, Bluemke DA,
Liu S. Modified look-locker inversion
recovery T1 mapping indices: assessment
of accuracy and reproducibility between
magnetic resonance scanners.
J Cardiovasc Magn Reson 2013; 15:64.
15 White SK, Sado DM, Fontana M, Banypersad
SM, Maestrini V, Flett AS, Piechnik SK,
Robson MD, Hausenloy DJ, Sheikh AM,
Hawkins PN, Moon JC. T1 Mapping for
Myocardial Extracellular Volume
measurement by CMR: Bolus Only Versus
Primed Infusion Technique, 2013 Apr 5
[Epub ahead of print].
16 Schelbert EB, Testa SM, Meier CG,
Ceyrolles WJ, Levenson JE, Blair AJ,
Kellman P, Jones BL, Ludwig DR,
Schwartzman D, Shroff SG, Wong TC.
Myocardial extravascular extracellular
volume fraction measurement by
gadolinium cardiovascular magnetic
resonance in humans: slow infusion
versus bolus. J Cardiovasc Magn Reson
2011, Mar 4;13-16.
17 Flacke SJ, Fischer SE, Lorenz CH.
Measurement of the gadopentetate
dimeglumine partition coefficient in
human myocardium in vivo: normal
distribution and elevation in acute and
chronic infarction. Radiology
2001;218:703-10.
18 Banypersad SM, Sado DM, Flett AS,
Gibbs SDG, Pinney JH, Maestrini V,
Cox AT, Fontana M, Whelan CJ,
Wechalekar AD, Hawkins PN, Moon JC.
Quantification of myocardial extracellular
volume fraction in systemic AL amyloidosis:
An Equilibrium Contrast Cardio-vascular
Magnetic Resonance Study.
Circ Cardiovasc Imaging 2013;6:34-39.
19 Ugander M, Oki AJ, Hsu LY, Kellman P,
Greiser A, Aletras AH, Sibley CT, Chen MY,
Bandettini WP, Arai AE. Extracellular
volume imaging by magnetic resonance
imaging provides insights into overt
and sub-clinical myocardial pathology.
Eur Heart J 2012; 33: 1268–1278.
20 Liu CY, Chang Liu Y, Wu C, Armstrong A,
Volpe GJ, van der Geest RJ, Liu Y, Hundley
WG, Gomes AS, Liu S, Nacif M, Bluemke
DA, Lima JAC. Evaluation of age related
interstitial myocardial fibrosis with Cardiac
Magnetic Resonance Contrast-Enhanced
T1 Mapping in the Multi-ethnic Study of
Atherosclerosis (MESA). J Am Coll Cardiol
2013 Jul 3 [Epub ahead of print].
21 Wong TC, Piehler K, Meier CG, Testa SM,
Klock AM, Aneizi AA, Shakesprere J,
Kellman P, Shroff SG, Schwartzman DS,
Mulukutla SR, Simon MA, Schelbert EB.
Association between extracellular matrix
expansion quantified by cardiovascular
magnetic resonance and short-term
mortality. Circulation 2012 Sep
4;126(10):1206-16.
22 Wong TC, Piehler KM, Kang IA, Kadakkal
A, Kellman P, Schwartzman DS, Mulukutla
SR, Simon MA, Shroff SG, Kuller LH,
Schelbert EB. Myocardial extracellular
volume fraction quantified by cardiovascular
magnetic resonance is increased
in diabetes and associated with mortality
and incident heart failure admission. Eur
Heart J 2013 Jun 11 [Epub ahead of print].
23 Giri S, Chung YC, Merchant A, Mihai G,
Rajagopalan S, Raman SV, Simonetti OP.
T2 quantification for improved detection
of myocardial edema. J Cardiovasc Magn
Reson 2009; 11:56.
24 Verhaert D, Thavendiranathan P, Giri S,
Mihai G, Rajagopalan S, Simonetti OP,
Raman SV. Direct T2 Quantification of
Myocardial Edema in Acute Ischemic
Injury. J Am Coll Cardiol Img 2011;4:
269-78.
25 Ugander M, Bagi PS, Oki AB, Chen B,
Hsu LY, Aletras AH, Shah S, Greiser A,
Kellman P, Arai AE. Myocardial oedema
as detected by Pre-contrast T1 and T2
CMR delineates area at risk associated
with acute myocardial infarction.
J Am Coll Cardiol Img 2012;5:596–603.
26 ThavendiranathanP, Walls M, Giri S,
Verhaert D, Rajagopalan S, Moore S,
Simonetti OP, Raman SV. Improved
detection of myocardial involvement in
acute inflammatory cardiomyopathies
using T2 Mapping. Circ Cardiovasc
Imaging 2012;5:102-110.
27 Usman AA, Taimen K, Wasielewski M,
McDonald J, Shah S, Shivraman G,
Cotts W, McGee E, Gordon R, Collins JD,
Markl M, Carr JC. Cardiac Magnetic
Resonance T2 Mapping in the monitoring
and follow-up of acute cardiac transplant
rejection: A Pilot Study. Circ Cardiovasc
Imaging. 2012; 6:782-90.
120
ms
0
ms
106 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 107](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-54-2048.jpg)
![Clinical Cardiovascular Imaging Cardiovascular Imaging Clinical
Preliminary Experiences with Compressed
Sensing Multi-Slice Cine Acquisitions for
the Assessment of Left Ventricular Function:
CV_sparse WIP
G. Vincenti, M.D.1; D. Piccini2,4; P. Monney, M.D.1; J. Chaptinel3; T. Rutz, M.D.1; S. Coppo3; M. O. Zenge, Ph.D.4;
M. Schmidt4; M. S. Nadar5; Q. Wang5; P. Chevre1, 6; M.; Stuber, Ph.D.3; J. Schwitter, M.D.1
1 Division of Cardiology and Cardiac MR Center, University Hospital of Lausanne (CHUV), Lausanne, Switzerland
2 Advanced Clinical Imaging Technology, Siemens Healthcare IM BM PI, Lausanne, Switzerland
3 Department of Radiology, University Hospital (CHUV) and University of Lausanne (UNIL) / Center for Biomedical Imaging
(CIBM), Lausanne, Switzerland
4 MR Applications and Workflow Development, Healthcare Sector, Siemens AG, Erlangen, Germany
5 Siemens Corportate Technology, Princeton, USA
6 Department of Radiology, University Hospital Lausanne, Switzerland
ize pathological myocardial tissue was
the basis to assign a class 1 indication
for patients with known or suspected
heart failure to undergo CMR in the
new Heart Failure Guidelines of the
European Society of Cardiology [3].
decision making [3] e.g. to start [4]
or stop [5] specific drug treatments
or to implant devices [6]. CMR is
generally accepted as the gold standard
method to yield most accurate
measures of LV ejection fraction and
LV volumes. This capability and the
additional value of CMR to characterIntroduction
Left ventricular (LV) ejection fraction
is one of the most important measures
in cardiology and part of every cardiac
imaging evaluation as it is recognized
as one of the strongest predictors
of outcome [1]. It allows to assess
the effect of established or novel
treatments [2], and it is crucial for
The evaluation of LV volumes and LV
ejection fraction are based on well-defined
protocols [7] and it involves
the acquisition of a stack of LV short
axis cine images from which volumes
are calculated by applying Simpson’s
rule. These stacks are typically acquired
in multiple breath-holds. Quality criteria
[8] for these functional images are
available and are implemented e.g.
for the quality assessment within the
European CMR registry which currently
holds approximately 33,000 patients
and connects 59 centers [9].
Recently, compressed sensing (CS)
techniques emerged as a means to
considerably accelerate data acquisition
without compromising significantly
image quality. CS has three
requirements:
1) transform sparsity,
2) incoherence of undersampling
artifacts,
and
3) nonlinear reconstruction (for
details, see below).
Based on these prerequisites, a CS
approach for the acquisition of cardiac
cine images was developed and
tested*. In particular, the potential to
acquire several slices covering the
heart in different orientations within
a single breath-hold would allow to
apply model-based analysis tools
which theoretically could improve the
motion assessment at the base of the
heart, where considerable through-plane
motion on short-axis slices can
introduce substantial errors in LV
volume and LV ejection fraction calculations.
Conversely, with a multi-breath-
hold approach, there are
typically small differences in breath-hold
positions which can introduce
errors in volume and function calculations.
The pulse sequence tested
here allows for the acquisition of 7
cine slices within 14 heartbeats with
an excellent temporal and spatial
resolution.
Such a pulse sequence would also
offer the advantage to obtain functional
information in at least a single
plane in patients unable to hold their
breath for several heartbeats or in
patients with frequent extrasystoles
or atrial fibrillation. However, it should
be mentioned that accurate quantitative
measures of LV volumes and
function cannot be obtained in highly
arrhythmic hearts or in atrial fibrillation,
as under such conditions volumes
and ejection fraction change
from beat to beat due to variable filling
conditions. Nevertheless, rough
estimates of LV volumes and function
would still be desirable in arrhythmic
patients.
In a group of healthy volunteers and
patients with different LV pathologies,
the novel single-breath-hold CS
cine approach was compared with
the standard multi-breath-hold cine
technique with respect to measure
LV volumes and LV ejection fraction.
The CV_sparse
work-in-progress
(WIP)
The CV_sparse WIP package implements
sparse, incoherent sampling
and iterative reconstruction for cardiac
applications. This method in
principle allows for high acceleration
factors which enable triggered 2D
real-time cine CMR while preserving
high spatial and/or temporal resolution
of conventional cine acquisitions.
Compressed sensing methods
exploit the potential of image compression
during the acquisition of
raw input data. Three components
[10] are crucial for the concept of
compressed sensing to work
I. Sparsity: In order to guarantee
compressibility of the input data,
sparsity must be present in a specific
transform domain. Sparsity can be
computed e.g. by calculating differences
between neighboring pixels
or by calculating finite differences in
angiograms which then detect primarily
vessel contours which typically
1
* Work in progress: The product is still under
development and not commercially available
yet. Its future availability cannot be ensured.
1 Display of the represent a few percent of the
planning of the
7 slices (4 short
axis and 3 long
axis slices)
acquired within
a single breath-hold
with the
three localizers.
1
2A 2B
Displays of the data analysis tools for the conventional short axis stack of cine images covering the entire LV (2A) and the 4D
analysis tool (2B), which is model-based and takes long axis shortening of the LV, i.e. mitral annulus motion into account.
Note that with both analysis tools, LV trabeculations are included into the LV volume, particularly in the end-diastolic images
(corresponding images on the left of top row in 2A and 2B).
2
108 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 109](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-55-2048.jpg)
![Clinical Cardiovascular Imaging Cardiovascular Imaging Clinical
entire image data only. Furthermore,
sparsity is not limited to the spatial
domain: the acquisition of cine
images of the heart can be highly
sparsified in the temporal dimension.
II. Incoherent sampling: The aliasing
artifacts due to k-space undersampling
must be incoherent, i.e.
noise-like, in that transform domain.
Here, it is to mention that fully random
k-space sampling is suboptimal
as k-space trajectories should be
smooth for hardware and physiological
considerations. Therefore, incoherent
sampling schemes must be
designed to avoid these concerns
while fulfilling the condition of random,
i.e. incoherent sampling.
III. Reconstruction: A non-linear iterative
optimization corrects for sub-sampling
artifacts during the process
of image reconstruction yielding to a
best solution with a sparse
representation
in a specific transform
domain and which is consistent with
the input data. Such compressed
sensing techniques can also be combined
with parallel imaging techniques
[11].
WIP CV_sparse Sequence
The current CV_sparse sequence [12]
realizes incoherent sampling by
initially
distributing the readouts
pseudo-randomly on the Cartesian
grid in k-space. In addition, for
cine-CMR imaging, a pseudo-random
offset
is applied from frame-to-frame
which results in an incoherent temporal
jitter. Finally, a variable sampling
density in k-space stabilizes
the iterative reconstruction. To avoid
eddy current effects for balanced
steady-state free precession (bSSFP)
acquisitions, pairing [13] can also be
applied. Thus, the tested CV_sparse
sequence is characterized by sparse,
incoherent sampling in space and
time, non-linear iterative reconstruction
integrating SENSE, and L1 wavelet
regularization in the phase encoding
direction and/or the temporal
dimension. With regard to reconstruction,
the ICE program runs a non-linear
iterative reconstruction with
k-t regularization in space and time
specifically modified for compressed
sensing. The algorithm derives from
a parallel imaging type reconstruction
which takes coil sensitivity maps
into account, thus supporting predominantly
high acceleration factors.
For cine CMR, no additional reference
scans are needed because – similar
to TPAT – the coil sensitivity maps are
calculated from the temporal average
of the input data in a central region
of k-space consisting of not more
than 48 reference lines. The extensive
calculations for image reconstruction
typically running 80 iterations
are performed online on all
CPUs on the MARS computer in parallel,
in order to reduce reconstruction
times.
Volunteer and Patient
studies
In order to obtain insight into the
image quality of single-breath-hold
multi-slice cine CMR images acquired
with the compressed sensing (CS)
approach, we studied a group of
healthy volunteers and a patient
group with different pathologies of
the left ventricle. In addition to the
evaluation of image quality, the
robustness and the precision of the
CS approach for LV volumes and LV
ejection fraction was also assessed in
comparison with a standard high-resolution
cine CMR approach. All CMR
examinations were performed on a
1.5T MAGNETOM Aera (Siemens
Healthcare, Erlangen, Germany). The
imaging protocol consisted of a set
of cardiac localizers followed by the
acquisition of a stack of conventional
short-axis SSFP cine images covering
the entire LV with a spatial and temporal
resolution of 1.2 x 1.6 mm2,
and approximately 40 ms, respectively
(slice thickness: 8 mm; gap between
slices: 2 mm). LV 2-chamber, 3-chamber,
and 4-chamber long-axis acquisitions
were obtained for image quality
assessment but were not used for LV
volume quantifications. As a next step,
to test the new CS-based technique,
slice orientations were planned to cover
the LV with 4 short-axis slices distributed
evenly over the LV long axis complemented
by 3 long-axis slices (i.e. a
2-chamber, 3-chamber, and 4-chamber
slice) (Fig. 1). These 7 slices were
then acquired in a single breath-hold
maneuver lasting 14 heart beats (i.e.
2 heart beats per slice) resulting in an
acceleration factor of 11.0 with a temporal
and spatial resolution of 30 ms
and 1.5 x 1.5 mm2, respectively (slice
thickness: 6 mm). As the reconstruction
algorithm is susceptible
to aliasing
in the phase-encoding direction, the
7 slices were first acquired with a non-cine
acquisition to check for correct
phase-encoding directions and, if
needed, to adjust the field-of-view
to avoid fold-over artifacts. After
confirmation
of correct imaging
parameters, the 7-slice single-breath-
hold cine CS-acquisition was
performed. In order to obtain a reference
for the LV volume measurement,
a phase-contrast flow measurement
in the ascending aorta was performed
to be compared with the
LV stroke volumes calculated from
the standard and CS cine data.
The conventional stack of cine SSFP
images was analyzed by the Argus
software (Siemens Argus 4D Ventricular
Function, Fig. 2A). The CS cine
data were analyzed by the 4D-Argus
software (Siemens Argus, Fig. 2B).
Such software is based on an LV
model and, with relatively few operator
interactions, the contours for the
LV endocardium and epicardium are
generated by the analysis tool. Of
note, this 4D analysis tool automatically
tracks the 3-dimensional motion
of the mitral annulus throughout the
cardiac cycle which allows for an
accurate volume calculation particularly
at the base of the heart.
Results and discussion
Image quality – robustness
of the technique
Overall, a very good image quality
of the single-breath-hold multi-slice
CS acquisitions was obtained in the
12 volunteers and 14 patient studies.
All CS data sets were of adequate
quality to undergo 4D analysis. Small
structures such as trabeculations
were visualized in the CS data sets
as shown in Figures 3 and 4. However,
very small structures, detectable by
the conventional cine acquisitions,
were less well discernible by the CS
images. Therefore, it should be mentioned
here, that this accelerated
single-
breath-hold CS approach would
be adequate for functional measurements,
i.e. LV ejection fraction
assessment (see also results below),
whereas assessment of small structures
as present in many cardiomyopathies
is more reliable when performed
on conventional cine images.
Temporal resolution of the new technique
appears adequate to even
detect visually the dyssynchroneous
contraction pattern in left bundle
branch block. Also, the image contrast
between the LV myocardium
and the blood pool was high on the
CS images allowing for an easy
assessment of the LV motion pattern.
As a result, the single-breath-hold
cine approach permits to reconstruct
the LV in 3D space with high temporal
resolution as illustrated in Figure
5. Since these data allow to correctly
include the 3D motion of the base
of the heart during the cardiac cycle,
the LV stroke volume appears to be
measurable by the CS approach with
higher accuracy than with the conventional
multi-breath-hold approach
(see results below). With an accurate
measurement of the LV stroke volume,
the quantification of a mitral
insufficiency should theoretically benefit
(when calculating mitral regurgitant
volume as ‘LV stroke volume
minus aortic forward-flow volume’).
As a current limitation of the CS
approach, its susceptibility for fold-over
artifacts should be mentioned
(Figs. 6A). Therefore, the field-of-view
must cover the entire anatomy
and thus, some penalty in spatial resStandard
cine
9 heartbeats
CV_SPARSE
3 heartbeats
CV_SPARSE
2 heartbeats
CV_SPARSE
1 heartbeat
Examples of visualization of small trabecular structures in the LV (in the rectangle) with the standard cine SSFP sequence (image
on the left) and the accelerated compressed sensing sequences (images on the right). Despite increasing acceleration most information
on small intraluminal structures remains visible.
3
RCA
Example demonstrating the performance of the compressed sensing
technique visualizing small structures such as the right coronary artery
(RCA) with high temporal and spatial resolution acquired within 2 heartbeats.
Short-axis view of the base of the heart (1 out of 17 frames).
4
3
4
110 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 111](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-56-2048.jpg)
![Clinical Cardiovascular Imaging Cardiovascular Imaging Clinical
olution may occur in relation to the
patient’s anatomy. In addition, the
sparsity in the temporal domain may
be limited in anatomical regions of
very high flow, and therefore, in
some acquisitions, flow-related artifacts
occurred in the phase-encoding
direction during systole (Figs. 6B, C).
Also, in its current version, the
sequence is prospective, thus it does
not cover the very last phases of the
cardiac cycle and the reconstruction
times for the CS images lasted several
minutes precluding an immediate
assessment of the image data quality
or using this image information to
plan next steps of a CMR examination.
Performance of the single-breath-
hold CS approach in
comparison with the stan-dard
multi-breath-hold cine
approach
From a quantitative point-of-view,
the accurate and reliable measurement
of LV volumes and function is
crucial as many therapeutic decisions
directly depend on these measures
[3–6]. In this current relatively small
study group, LV end-diastolic and
end-systolic volumes measured by
the single-breath-hold CS approach
were comparable with those calculated
from the standard multi-breath-hold
cine SSFP approach. LVEDV and
LVESV differed by 10 ml ± 17 ml and
2 ml ± 12 ml, respectively. Most importantly,
LV ejection fraction differed
by only 1.3 ± 4.7% (50.6% vs 49.3%
for multi-breath-hold and single-breath-hold,
respectively, p = 0.17; regression:
r = 0.96, p < 0.0001; y = 0.96x +
0.8 ml). Thus, it can be concluded that
the single-breath-hold CS approach
could potentially replace the multi-breath-
hold standard technique for the
assessment of LV volumes and systolic
function.
What about the accuracy of
the novel single-breath-hold
CS technique?
To assess the accuracy of the LV volume
measurements, LV stroke volume
was compared with the LV output
measured in the ascending aorta with
phase-contrast MR. As the flow measurements
were performed distally to
the coronary arteries, flow in the coronaries
was estimated as the LV mass
multiplied by 0.8 ml/min/g. An excellent
agreement was found with a
mean of 86.8 ml/beat for the aortic
flow measurement and 91.9 ml/beat
for the LV measurements derived
from the single-breath-hold CS data
(r = 0.93, p < 0.0001). By Bland-Altman
analysis, the stroke volume approach
overestimated by 5.2 ml/beat versus
the reference flow measurement. For
the conventional stroke volume measurements,
this difference was 15.6
ml/beat (linear regression analysis vs
aortic
flow: r = 0.69, p < 0.01). More
importantly, the CS LV stroke data were
not only more precise with a smaller
mean difference, the variability of the
CS data vs the reference flow data was
less with a standard deviation as low
as 6.8 ml/beat vs 12.9 ml/beat for the
standard multi-breath-hold approach
(Fig. 7). Several explanations may apply
for the higher accuracy of the single-breath-
hold multi-slice CS approach in
comparison to the conventional multi-breath-
hold approach:
1) With the single-breath-hold
approach, all acquired slices are correctly
co-registered, i.e. they are correctly
aligned in space, a prerequisite
for the 4D-analysis tool to work
properly.
2) This 4D-analysis tool allows for an
accurate tracking of the mitral valve
plane motion during the cardiac cycle
as shown in Figure 5, which is important
as the cross-sectional area of the
heart at its base is large and thus, inaccurate
slice positioning at the base of
Display of the 3D reconstruction derived from the 7 slices acquired within a single breath-hold. Note the long-axis shortening of the
LV during systole allowing for accurate LV volume measurements (5A, 5B, yellow plane). Any orientation of the 3D is available for
inspection of function (5A–D).
5
6A
A typical fold-over
artifact along the
phase-encoding
direction in a short
axis slice, oriented
superior-inferior for
demonstrative
purpose.
6B
No flow-related
artifacts are
visible on the
end-diastolic
phases, while
small artifacts in
phase-encoding
direction (Artif,
arrows) occur in
mid-systole
projecting over
the mitral valve
(6C).
5A 5B
5C 5D
6A
6B
6C
Artif
Artif
112 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 113](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-57-2048.jpg)
![Clinical Cardiovascular Imaging Cardiovascular Imaging Clinical
50 60 70
120 130
130
120
110
100
90
80
70
the heart with conventional short-axis
slices typically translate in
relatively large errors. Nevertheless,
we observed a systematic overestimation
of the stroke volume by the
CS approach of 5.2 ml/beat in comparison
to the flow measurements.
In normal hearts with tricuspid aortic
valves, an underestimation of aortic
flow by the phase-contrast technique
is very unlikely [14]. Thus, overestimation
of stroke volume by the volume
approach is to consider. In the volume
contours, the papillary muscles
are excluded as illustrated in Figure 8.
As these papillary muscles are excluded
in both the diastolic and systolic contours,
this aspect should not affect
net LV stroke volume. However, as
shown in Figure 8, smaller trabeculations
of the LV wall are included into
the LV blood pool contour in the
diastolic
phase, while these trabeculations,
CS technique
Standard technique
when compacted in the
end-
systolic phase, are excluded from
the blood pool resulting in a small
overestimation of the end-diastolic
volume, and thus, LV stroke volume.
This explanation is likely as Van
Rossum
et al. demonstrated a slight
underestimation of the LV mass when
calculated on end-diastolic phases
versus end-systolic phases, as trabeculations
in end-diastole are typically
excluded from the LV walls [15].
In summary, this novel very fast
acquisition strategy based on a CS
technique allows to cover the entire
LV with high temporal and spatial
resolution within a single breath-hold.
The image quality based on these
preliminary results appears adequate
to yield highly accurate measures
of LV volumes, LV stroke volume,
LV mass, and LV ejection fraction.
7
Testing of this very fast multi-slice cine
approach for the atria and the right
ventricle is currently ongoing. Finally,
these preliminary data show that compressed
sensing MR acquisitions in
the heart are feasible in humans and
compressed sensing might be implemented
for other important cardiac
sequences such as fibrosis/viability
imaging, i.e. late gadolinium enhancement,
coronary MR angiography, or
MR first-pass perfusion.
The Cardiac MR Center of the
University Hospital Lausanne
The Cardiac Magnetic Resonance Center
(CRMC) of the University Hospital of
Lausanne (Centre Hospitalier Universitaire
Vaudois; CHUV) was established
in 2009. The CMR center is dedicated
to high-quality clinical work-up of cardiac
patients, to deliver state-of-the-art
training in CMR to cardiologists and
radiologists, and to pursue research.
In the CMR center education is provided
for two specialties while focusing
on one organ system. Traditionally,
radiologists have focussed on using
one technique for different organs,
while cardiologists have concentrated
on one organ and perhaps one technique.
Now in the CMR center the
focus is put on a combination of specialists
with different background on
one organ. Research at the CMR center
is devoted to four major areas: the
study of
1.) cardiac function and tissue characterization,
specifically to better understand
diastolic dysfunction,
2.) the development of MR-compatible
cardiac devices such as pacemakers
and ICDs;
3.) the utilization of hyperpolarized
13C-carbon contrast media to investigate
metabolism in the heart, and
An excellent correlation
is obtained for
the LV stroke volume
calculated from the
compressed sensing
data with the flow
volume in the aorta
measured by phase-contast
technique.
Variability of the
conventional LV
stroke volume data
appears higher than
for the compressed
sensing data.
LV stroke volume: comparison vs aortic forward flow
LV short-axis slice: CV_SPARSE
4.) the development of 19F-fluorine-based
CMR techniques to detect
inflammation and to label and track
cells non-invasively.
For the latter two topics, the CMR
center established tight collaborations
with the Center for Biomedical
Imaging (CIBM), a network around
Lake Geneva that includes the Ecole
Polytechnique Fédérale de Lausanne
(EPFL), and the universities and university
hospitals of Lausanne and
Geneva. In particular, strong collaborative
links are in place with the
CVMR team of Prof. Matthias Stuber,
a part of the CIBM and located at the
University Hospital Lausanne and
with Prof. A. Comment, with whom
we perform the studies on real-time
metabolism based on the 13C-carbon
hyperpolarization (DNP) technique.
In addition, collaborative studies are
ongoing with the Heart Failure and
Cardiac Transplantation Unit led by
Prof. R. Hullin (detection of graft
rejection by tissue characterization)
and the Oncology Department led
by Prof. Coukos (T cell tracking by19F-
MRI in collaboration with Prof.
Stuber, R. van Heeswijk, CIBM, and
Prof. O. Michielin, Oncology). This
structure allows for a direct interdisciplinary
interaction between physicians,
engineers, and basic scientists
on a daily basis with the aim to
enable innovative research and fast
translation of these techniques from
bench to bedside.
The CMRC is also the center of competence
for the quality assessment of
the European CMR registry which
holds currently approximately 33,000
patient studies acquired in 59 centers
across Europe.
The members of the CRMC team are:
Prof. J. Schwitter (director of the
center),
PD Dr. X. Jeanrenaud, Dr. D.
Locca, MER, Dr. P. Monney, Dr. T.
Rutz, Dr. C. Sierro, and Dr. S. Koestner
(cardiologists, staff members),
Overestimation of end-diastolic LV volumes by volumetric measurements. In comparison to ejected blood from the LV as measured
with phase-contrast techniques, the volumetric measurements of LV stroke volume overestimated by approximately 5 ml, most
likely by overestimation of LV end-diastolic volume. Small trabculations (yellow contours in 8A) are included into the LV blood
volume (red contour in 8A) in diastole, while these trabeculations (yellow contours in 8B) are typically included in the end-systolic
phase (red contours in 8B). For the same reasons, LV mass (= green contour minus red contour) is often slightly underestimated in
diastole vs systole.
8
7
8A 8B
End-diastolic frame End-systolic frame
ml/beat (aortic forward flow by PC)
ml/beat (LV stroke volume)
80 90 100 110
60
114 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 115](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-58-2048.jpg)
![References
1 Curtis JP, Sokol SI, Wang Y, Rathore SS,
Ko DT, Jadbabaie F, Portnay EL, Marshalko
SJ, Radford MJ, Krumholz HM. The association
of left ventricular ejection fraction,
mortality, and cause of death in stable
outpatients with heart failure. Journal of
the American College of Cardiology.
2003;42(4):736-42.
2 Sürder D, Manka R, Lo Cicero V, Moccetti T,
Rufibach K, Soncin S, Turchetto L,
Radrizzani M, Astori G, Schwitter J, Erne P,
Jamshidi P, Auf Der Maur C, Zuber M,
Windecker S, Moschovitis A, Wahl A,
Bühler I, Wyss C, Landmesser U, Lüscher T,
Corti R. Intracoronary injection of bone
marrow derived mononuclear cells, early
or late after acute myocardial infarction:
Effects on global LV-function: 4 months
results of the SWISS-AMI trial. Circulation.
2013;127:1968-79.
11 Liang D, Liu B, Wang J, Ying L. Accelerating
SENSE using compressed sensing. Magnetic
Resonance in Medicine. 2009;62(6):
1574-84.
12 Liu J. Dynamic cardiac MRI reconstruction
with weighted redundant Haar wavelets.
Magn Reson Med. 2012;Proc. ISMRM
2012,abstract.
13 Bieri O, Markl M, Scheffler K. Analysis
and compensation of eddy currents in
balanced SSFP. Magn Reson Med.
2005;54:129-37.
14 Muzzarelli S, Monney P, O’Brien K, Faletra F,
Moccetti T, Vogt P, Schwitter J. Quantification
of aortic valve regurgitation by
phase-contrast magnetic resonance in
patients with bicuspid aortic valve: where
to measure the flow? . Eur Heart J - CV
Imaging. 2013:in press.
15 Papavassiliu T, Kühl HP, Schröder M,
Süselbeck T, Bondarenko O, Böhm CK,
Beek A, Hofman MMB, van Rossum AC.
Effect of Endocardial Trabeculae on
Left Ventricular Measurements and
Measurement Reproducibility at Cardiovascular
MR Imaging1. Radiology. 2005
July 1, 2005;236(1):57-64.
Contact
Professor Juerg Schwitter
Médecin Chef Cardiologie
Directeur du Centre de la RM
Cardiaque du CHUV
Centre Hospitalier
Universitaire
Vaudois – CHUV
Rue du Bugnon 46
1011 Lausanne
Suisse
Phone: +41 21 314 0012
jurg.schwitter@chuv.ch
www.cardiologie.chuv.ch
Dr. G. Vincenti (cardiologist) and
Dr. N. Barras (cardiologist in training,
rotation), PD. Dr. S. Muzzarellli (affiliated
cardiologist), Prof. C. Beigelman
and Dr. X. Boulanger (radiologists,
staff members), Dr. G.L. Fetz (radiologist
in training, rotation), C. Gonzales,
PhD (19F-fluorine project leader),
H. Yoshihara, PhD (13C-carbon project
leader), V. Klinke (medical student,
doctoral thesis), C. Bongard (medical
student, master thesis), P. Chevre
(chief CMR technician), and F. Recordon
and N. Lauriers (research
nurses).
Acknowledgements
The authors would like to thank all
the members of the team of MR technologists
at the CHUV for their highly
valuable participation, helpfulness
and support during the daily clinical
CMR examinations and with the
research protocols. Finally, a very
important acknowledgment goes to
Dr. Michael Zenge, Ms. Michaela
Schmidt, and the whole Siemens MR
Cardio team of Edgar Müller in
Erlangen.
3 McMurray JJV, Adamopoulos S, Anker SD,
Auricchio A, Böhm M, Dickstein K, Falk V,
Filippatos G, Fonseca C, Sanchez MAG,
Jaarsma T, Køber L, Lip GYH, Maggioni
AP, Parkhomenko A, Pieske BM, Popescu
BA, Rønnevik PK, Rutten FH, Schwitter J,
Seferovic P, Stepinska J, Trindade PT,
Voors AA, Zannad F, Zeiher A. ESC Guidelines
for the diagnosis and treatment of
acute and chronic heart failure 2012.
European Heart Journal. 2012 May 19,
2012(33):1787–847.
4 Zannad F, McMurray JJV, Krum H,
van Veldhuisen DJ, Swedberg K, Shi H,
Vincent J, Pocock SJ, Pitt B. Eplerenone
in Patients with Systolic Heart Failure
and Mild Symptoms. New England
Journal of Medicine. 2011;364(1):11-21.
5 Gharib MI, Burnett AK. Chemotherapy-induced
cardiotoxicity: current practice
and prospects of prophylaxis. European
Journal of Heart Failure. 2002 June 1,
2002;4(3):235-42.
6 Bardy GH, Lee KL, Mark DB, Poole JE,
Packer DL, Boineau R, Domanski M,
Troutman C, Anderson J, Johnson G,
McNulty SE, Clapp-Channing N, Davidson-
Ray LD, Fraulo ES, Fishbein DP, Luceri
RM, Ip JH. Amiodarone or an Implantable
Cardioverter–Defibrillator for Congestive
Heart Failure. New England Journal of
Medicine. 2005;352(3):225-37.
7 Schwitter J. CMR-Update. 2. Edition ed.
Lausanne, Switzerland. www.herz-mri.ch.
8 Klinke V, Muzzarelli S, Lauriers N,
Locca D, Vincenti G, Monney P, Lu C,
Nothnagel D, Pilz G, Lombardi M,
van Rossum A, Wagner A, Bruder O,
Mahrholdt H, Schwitter J. Quality
assessment of cardiovascular magnetic
resonance in the setting of the European
CMR registry: description and validation
of standardized criteria. Journal of
Cardiovascular Magnetic Resonance.
2013;15(1):55.
9 Bruder O, Wagner A, Lombardi M,
Schwitter J, van Rossum A, Pilz G,
Nothnagel D, Steen H, Petersen S,
Nagel E, Prasad S, Schumm J, Greulich S,
Cagnolo A, Monney P, Deluigi C, Dill T,
Frank H, Sabin G, Schneider S,
Mahrholdt H. European Cardiovascular
Magnetic Resonance (EuroCMR)
registry-multi national results from 57
centers in 15 countries. J Cardiovasc
Magn Reson. 2013;15:1-9.
10 Lustig M, Donoho D, Pauly JM. Sparse
MRI: The application of compressed
sensing for rapid MR imaging. Magnetic
Resonance in Medicine.
2007;58(6):1182-95.
Accelerated Segmented Cine TrueFISP
of the Heart on a 1.5T MAGNETOM Aera
Using k-t-sparse SENSE
Maria Carr1; Bruce Spottiswoode2; Bradley Allen1; Michaela Schmidt2; Mariappan Nadar4; Qiu Wang4;
Jeremy Collins1; James Carr1; Michael Zenge2
1 Northwestern University, Feinberg School of Medicine, Chicago, IL, USA
2 Siemens Healthcare
3 Siemens Corporate Technology, Princeton, United States
Introduction
Cine MRI of the heart is widely
regarded as the gold standard for
assessment of left ventricular volume
and myocardial mass and is increasingly
utilized for assessment of cardiac
anatomy and pathology as part
of clinical routine. Conventional cine
imaging approaches typically require
1 slice per breath-hold, resulting in
lengthy protocols for complete cardiac
coverage. Parallel imaging allows
some shortening of the acquisition
time, such that 2–3 slices can be
acquired in a single breath-hold. In
cardiac cine imaging artifacts become
more prevalent with increasing acceleration
factor. This will negatively
impact the diagnostic utility of the
images and may reduce accuracy of
quantitative measurements. However,
regularized iterative reconstruction
techniques can be used to considerably
improve the images obtained
from highly undersampled data. In
this work, L1-regularized iterative
SENSE as proposed in [1] was applied
to reconstruct under-sampled k-space
data. This technique* takes advantage
of the de-noising characteristics
of Wavelet regularization and promises
to very effectively suppress sub-sampling
artifacts. This may allow for
high acceleration factors to be used,
while diagnostic image quality is
preserved.
The purpose of this study was to
compare segmented cine TrueFISP
images from a group of volunteers
and patients using three acceleration
and reconstruction approaches: iPAT
factor 2 with conventional reconstruction;
T-PAT factor 4 with convenTable
tional reconstruction; and T-PAT factor
4 with iterative k-t-sparse SENSE
reconstruction.
Technique
Cardiac MRI seems to be particularly
well suited to benefit from a group of
novel image reconstruction methods
known as compressed sensing [2]
which promise to significantly speed
up data acquisition. Compressed
sensing methods were introduced to
MR imaging [3, 4] just a few years
ago and have since been successfully
combined with parallel imaging [5,
6]. Such methods try to utilize the
* Work in progress: The product is still
under
development and not commercially
available yet. Its future availability cannot
be ensured.
1: MRI conventional and iterative imaging parameters
Parameters Conventional iPAT 2 Conventional T-PAT 4 Iterative T-PAT 4
Iterative recon No No Yes
Parallel imaging iPAT2 (GRAPPA) TPAT4 TPAT4
TR/TE (ms) 3.2 / 1.6 3.2 / 1.6 3.2 / 1.6
Flip angle (degrees) 70 70 70
Pixel size (mm2) 1.9 × 1.9 1.9 × 1.9 1.9 × 1.9
Slice thickness (mm) 8 8 8
Temp. res. (msec) 38 38 38
Acq. time (sec) 7 3.2 3.2
Clinical Cardiovascular Imaging
116 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
Cardiovascular Imaging Technology
MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 117](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-59-2048.jpg)
![As outlined by Liu et al. in [1], the
image reconstruction can be formulated
as an unconstrained optimization
problem. In the current implementation,
this optimization is solved using
a Nesterov-type algorithm [7]. The
L1-regularization with a redundant
Haar transform is efficiently solved
using a Dykstra-type algorithm [8].
This allowed a smooth integration into
the current MAGNETOM platform and,
therefore, facilitates a broad clinical
evaluation.
Materials and methods
Nine healthy human volunteers
(57.4 male/56.7 female) and
20 patients (54.4 male/40.0 female)
with suspected cardiac disease were
scanned on a 1.5T MAGNETOM Aera
system under an approved institutional
review board protocol. All nine volunteers
and 16 patients were imaged
using segmented cine TrueFISP
sequences with conventional GRAPPA
factor 2 acceleration (conventional
iPAT 2) T-PAT factor 4 acceleration
(conventional T-PAT 4), and T-PAT factor
4 acceleration with iterative k-t-sparse
SENSE reconstruction (iterative
T-PAT 4). The remaining 4 patients
were scanned using only conventional
iPAT 2 and iterative T-PAT 4 techniques.
Note that the iterative technique is
fully integrated into the standard
reconstruction environment.
The imaging parameters for each
imaging sequence are provided in
Table 1. All three sequences were run
in 3 chamber and 4 chamber views,
as well as a stack of short axis slices.
Quantitative analysis was performed
on all volunteer data sets at a syngo
MultiModality Workplace (Leonardo)
using Argus post-processing software
(Siemens Healthcare, Erlangen,
Germany)
by an experienced cardiovascular
MRI technician. Ejection fraction,
end-diastolic volume, end-systolic
volume, stroke volume, cardiac
output,
and myocardial mass were calculated.
In all volunteers and patients,
5
4
3
2
1
blinded qualitative scoring was performed
by a radiologist using a 5 point
Likert scale to assess overall image
quality (1 – non diagnostic; 2 – poor;
3 – fair; 4 – good; 5 – excellent).
Images were also scored for artifact
and noise (1 – severe; 2 – moderate;
3 – mild; 4 – trace; 5 – none).
All continuous variables were compared
between groups using an
unpaired t-test, while ordinal qualitative
variables were compared using
a Wilcoxon signed-rank test.
Results
All images were acquired successfully
and image quality was of diagnostic
quality in all cases. The average scan
time per slice for conventional iPAT 2,
conventional T-PAT 4 and iterative
T-PAT 4 were for patients 7.7 ± 1.5 sec,
5.6 ± 1.5 sec and 2.9 ± 1.5 sec and
for the volunteers 9.8 ± 1.5 sec, 3.2 ±
1.5 sec and 3.0 ± 1.5 sec, respectively.
The results in scan time are illustrated
in Figure 1. In both patients and volunteers,
conventional iPAT 2 were significantly
longer than both conventional
T-PAT 4 and iterative T-PAT 4 techniques
(p < 0.001 for each group).
The results for ejection fraction (EF)
for all three imaging techniques are
provided in Figure 2. The average EF
for conventional T-PAT 4 was slightly
lower than that measured for conventional
iPAT 2 and iterative T-PAT,
but the group size is relatively small
(9 subjects) and this difference was
not significant (p = 0.34 and p = 0.22
respectively).There was no statistically
significant difference in ejection
fraction between the conventional
iPAT 2 and the iterative T-PAT 4
sequences (p = 0.48).
The results for image quality, noise
and artifact are provided in Figure 3.
The iterative T-PAT 4 images had comparable
image quality, noise and artifact
scores compared to the conventional
iPAT 2 images. The conventional
T-PAT 4 images had lower image quality,
more artifacts and higher noise
compared to the other techniques.
Figures 4 and 5 show an example of
4-chamber and mid-short axis images
from all three techniques in a patient
with basal septal hypertrophy. In both
series’, the conventional iPAT 2 and
iterative T-PAT 4 images are comparable
in quality, while the conventional
T-PAT 4 image is visibly noisier.
Cardiovascular Imaging Technology
Discussion
This study compares a novel accelerated
segmented cine TrueFISP technique
to conventional iPAT 2 cine
TrueFISP and T-PAT 4 cine TrueFISP in
a cohort of normal subjects and
patients. The iterative reconstruction
technique provided comparable measurements
of ejection fraction to the
clinical gold standard (conventional
iPAT 2). The accelerated segmented
cine TrueFISP with T-PAT 4, which
was used as comparison technique,
produced slightly lower EF values
compared to the other techniques,
although this was not found to be
statistically significant. The iterative
reconstruction produced comparable
image quality, noise and artifact
scores to the conventional reconstruction
using iPAT 2. The conventional
T-PAT 4 technique had lower image
quality and higher noise scores compared
to the other two techniques.
The iterative T-PAT 4 segmented cine
technique allows for greater than
50% reduction in acquisition time for
comparable image quality and spatial
resolution as the clinically used iPAT 2
cine TrueFISP technique. This iterative
technique could be extended to
permit complete heart coverage in
a single breath-hold thus greatly simplifying
and shortening routine clinical
cardiac MRI protocols, which has
been one of the biggest obstacles to
wide acceptance of cardiac MRI. With
a shorter cine acquisition, additional
advanced imaging techniques, such
as perfusion and flow, can be more
readily added to patient scans within
a reasonable protocol length.
Technology Cardiovascular Imaging
12
10
8
6
4
Single slice scan time in patients and volunteers. There was a statistically
significant reduction in scan time compared to the standard iPAT2 for both
TPAT4 acceleration and iterative reconstruction TPAT4 acceleration.
1
Qualitative scores in patients and volunteers. Image quality was highest and
noise and artifact were lowest with iterative T-PAT 4 and conventional iPAT 2
compared to conventional T-PAT 4.
3
full potential of image compression
during the acquisition of raw input
data. In the case of highly subsampled
input data, a non-linear iterative
optimization avoids sub-sampling
artifacts during the process of image
reconstruction. The resulting images
represent the best solution consistent
with the input data, which have
a sparse representation in a specific
transform domain. In the most favorable
case, residual artifacts are not
visibly perceptible or are diagnostically
irrelevant.
0
Conventional iPAT 2
Scan Time (sec)
Conventional T-PAT 4 Iterative T-PAT 4
2
Standard iPAT 2 T-PAT 4 Acceleration Iterative Reconstruction T-PAT 4 Accel.
0
Quality Noise Artifact
1 3
65,00
60,00
55,00
50,00
45,00
40,00
Ejection fraction in volunteers. Quantitatively measured ejection fractions
were comparable across all three techniques.
2
Conventional iPAT 2
Ejection Fraction (%)
Conventional T-PAT 4 Iterative T-PAT 4
2
118 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 119](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-60-2048.jpg)
![Technology Cardiovascular Imaging Cardiovascular Imaging Technology
There are currently some limitations
to the technique. Firstly, the use of
SENSE implies that aliasing artifacts
can occur if the field-of-view is
smaller than the subject, which is
sometimes difficult to avoid in the
short axis orientation. But a solution
to this is promised to be part of a
future release of the current proto-type.
Secondly, the image reconstruc-tion
times of the current implemen-tation
seems to be prohibitive for
routine clinical use. However,
we anticipate future algorithmic
6A 6B
6 Real-time cine TrueFISP T-PAT 6 images reconstructed using (6A) conventional, and (6B) iterative techniques.
Contact
Maria Carr, RT (CT)(MR)
CV Research Technologist
Department of Radiology
Northwestern University
Feinberg School of Medicine
737 N. Michigan Ave.
Suite 1600
Chicago, IL 60611
USA
Phone: +1 312-926-5292
m-carr@northwestern.edu
References
1 Liu J, Rapin J, Chang TC, Lefebvre A,
Zenge M, Mueller E, Nadar MS. Dynamic
cardiac MRI reconstruction with
weighted redundant Haar wavelets. In
Proceedings of the 20th Annual Meeting
of ISMRM, Melbourne, Australia, 2002.
p 4249.
2 Candes EJ, Wakin MB. An Introduction
to compressive sampling. IEEE Signal
Processing Magazine 2008. 25(2):21-30.
doi: 10.1109/MSP.2007.914731.
improvements with increased compu-tational
power to reduce the recon-struction
time to clinically acceptable
values.
Of course, iterative reconstruction
techniques are not just limited to
cine imaging of the heart. Future
work may see this technique applied
to time intense techniques such as
4D flow phase contrast MRI and 3D
coronary MR angiography, making
them more clinically applicable.
Furthermore, higher acceleration
rates might be achieved by using an
incoherent sampling pattern [9].
With sufficiently high acceleration, the
technique can also be used effectively
for real time cine cardiac imaging in
patients with breath-holding difficul-ties
or arrhythmia. Figure 6 shows that
real-time acquisition with T-PAT 6 and
k-t iterative reconstruction still results
in excellent image quality.
In conclusion, cine TrueFISP of the
heart with inline k-t-sparse iterative
reconstruction is a promising tech-nique
for obtaining high quality cine
images at a fraction of the scan time
compared to conventional techniques.
Acknowledgement
The authors would like to thank Judy
Wood, Manger of the MRI Department
at Northwestern Memorial Hospital,
for her continued support and collabo-ration
with our ongoing research
through the years. Secondly, we would
like to thank the magnificent Cardio-vascular
Technologist’s Cheryl Jarvis,
Tinu John, Paul Magarity, Scott Luster
for their patience and dedication to
research. Finally, the Resource Coordi-nators
that help us make this possible
Irene Lekkas, Melissa Niemczura and
Paulino San Pedro.
3 Block KT, Uecker M, Frahm J. Unders-ampled
Radial MRI with Multiple Coils.
Iterative Image Reconstruction Using a
Total Variation Constraint. Magn Reson
Med 2007. 57(6):1086-98.
4 Lustig M, Donoho D, Pauly JM. Sparse
MRI: The application of compressed
sensing for rapid MR imaging. Magn
Reson Med 2007. 58(6):1182-95.
5 Liang D, Liu B, Wang J, Ying L. Acceler-ating
SENSE using compressed sensing.
Magn Reson Med 2009. 62(6):154-84.
doi: 10.1002/mrm.22161.
6 Lustig M, Pauly, JM. SPIRiT: Iterative
self-
consistent parallel imaging
reconstruction from arbitrary k-space.
Magn Reson Med 2010. 64(2):457-71.
doi: 10.1002/mrm.22428.
7 Beck A, Teboulle M. A fast iterative
shrinkage-thresholding algorithm for
linear inverse problems. SIAM J Imaging
Sciences 2009. 2(1): 183-202.
8 Dykstra RL. An algorithm for restricted
least squares regression. J Amer Stat
Assoc 1983 78(384):837-842.
9 Schmidt M, Ekinci O, Liu J, Lefebvre A,
Nadar MS, Mueller E, Zenge MO. Novel
highly accelerated real-time CINE-MRI
featuring compressed sensing with k-t
regularization in comparison to TSENSE
segmented and real-time Cine imaging.
J Cardiovasc Magn Reson 2013.
15(Suppl 1):P36.
4A 4B 4C
Four chamber cine TrueFISP from a normal volunteer. (4A) Conventional iPAT 2, acquisition time 8 s. (4B) Conventional
T-PAT 4, acquisition time 3 seconds. (4C) Iterative T-PAT 4, acquisition time 3 seconds.
4
5A 5B 5C
End-systolic short axis cine TrueFISP images from a patient with a history of myocardial infarction. A metal artifact from a previous
sternotomy is noted in the sternum. There is wall thinning in the inferolateral wall with akinesia on cine views, consistent with
an old infarct in the circumflex territory. (5A) Conventional iPAT 2, (5B) conventional T-PAT 4, (5C) iterative T-PAT 4.
5
120 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 121](https://image.slidesharecdn.com/magnetomflash55rsnaissue-01177877-140821155554-phpapp02/75/Magnetom-flash-55-61-2048.jpg)
Uploaded byJhon Arriaga Cordova
Magnetom flash 55
The editorial discusses how collaborations between academic medical centers and industry have advanced MRI through various examples in the magazine issue. It highlights how such partnerships have leveraged MRI's contrast mechanisms and improved image quality, examination speed, and patient comfort. The growing healthcare needs require continued diagnostic imaging advances. The issue features articles demonstrating innovations in motion-robust imaging, rapid protocols, quantitative mapping, and noise reduction.
More Related Content
Magnetom flash 55
- 1. MAGNETOM Flash TheMagazine of MRI 55 Clinical T1w MRI using Radial VIBE Page 6 Acute MR Stroke Protocol in 6 Minutes Page 44 New Generation Cardiac Parametric Mapping: The Clinical Role of T1 and T2 Mapping Page 104 Issue Number 5/2013 | RSNA Edition Not for distribution in the US
- 2. Editorial The themeof the 99th Annual Meeting of the Radiological Society of North America is “The Power of Partnership”. Nowhere is this concept better exemplified than in the cooperation between academic medical centers and industry partners in the development and improvement of diagnostic imaging. This issue of MAGNETOM Flash contains a wealth of examples of how such collaborations have advanced the discipline of MRI. As the world population’s healthcare needs grow, so must diagnosis and disease management continue to advance. Diagnostic imaging plays an increasingly central role in detecting and characterizing disease, and guiding therapy. In particular, MRI remains a cornerstone of neurologic, orthopedic, oncologic, and cardiovascular imaging. MRI has long had advantages in leveraging useful contrast mechanisms for visualization of anatomy and pathology. This is well-demonstrated in articles describing visualization of diffusion- weighted imaging data [Doring et al. page 12], spectroscopic imaging of prostate cancer [Scheenen et al. page 16], susceptibility-weighted imaging [Ascencio et al. page 52], Mustafa R. Bashir, M.D., is an Assistant Professor of Radiology at Duke University Medical Center, Durham, USA. He joined the faculty at Duke in 2010, and serves as the Director of MRI and Director of Body MRI. In 2012, he was named the Medical Director of the Center for Advanced Magnetic Resonance Development, Duke Radiology’s MRI research and development facility. He has been awarded several industry research grants as principal investigator and serves as site imaging principal investigator on several NIH-funded grants. Dr. Bashir serves as a working group chair for the Liver Imaging Reporting and Data System (LI-RADS) committee for the American College of Radiology and is a site radiologist for the NIH-funded Non- Alcoholic Steatohepatitis Clinical Research Network. His clinical and research interests include abdominal MRI, liver imaging, quantitative imaging, and novel contrast mechanisms. and quantitative myocardial relaxivity mapping [Moon et al. page 104]. However, gone are the days when lengthy examinations producing inconsistent image quality were considered acceptable. In an atmosphere of rising cost and diminishing resources, all imaging is under pressure to demonstrate consistent examination quality, despite increasing use in challenging populations, such as the obese and those with diminished breath-holding capacity. In their article on the New York University- Langone Medical Center experience using Radial VIBE*, Tobias Block et al. show the power of a motion-robust non-Cartesian strategy to obtain free-breathing, artifact-free, volumetric T1-weighted image sets in the body [page 6]. Such paradigms can be applied to improve image quality in patients unable to hold their breath, and to enhance the MRI experience by providing healthier patients with a more comfortable examination with fewer breath-holds. In addition, colleagues at the University Hospital of Lausanne and Northwestern University demonstrate the feasibility of rapid cardiac acquisition using compressed sensing* methods [page 108 and 117]. Such techniques are shown to produce high-resolution, multiplanar acquisitions in single breath-holds, which can both shorten total examination time and provide comparable or more accurate measure-ments of left ventricular ejection fraction and stroke volume, compared with conventional methods. The broad availability of commercial wide-bore systems with high channel counts makes clinical MR imaging a reality in a larger portion of the population. Particularly in the United States, where over 35% of the population is obese [http://www.cdc.gov/obesity/ data/adult.html], high-quality imaging is now available to more patients than ever, with greater physical comfort. In addition to comfort, turnaround time is also considered an important measure of examination quality, and as a first-line diagnostic modality, MRI must provide rapid, definitive diagnosis in order for appropriate treatment to be rendered in a timely manner. Working Together Editorial “In an atmosphere of rising cost and diminishing resources, all imaging is under pressure to demonstrate consistent examination quality, despite increasing use in challenging populations, such as the obese and those with diminished breath- holding capacity.” Mustafa R. Bashir, M.D. This is exemplified in the article by Kambiz Nael et al., who describe a six-minute comprehensive acute stroke protocol, combining brain structure imaging, functional measures including diffusion- and perfusion-weighted imaging, and MR angiography [page 44]. This ‘one-stop-shop’ approach can facilitate rapid triage of appropriate patients to endovascular management while avoiding unnecessary, and potentially dangerous, delays in diagnosis. Finally, Mark Griswold et al. and Masahiro Ida address another important element of a comfortable MRI experience, as they discuss simple methods for optimizing frequently-used pulse sequences to reduce acoustic noise [page 30 and 35]. In the following pages, fifteen high-quality articles from a diverse group of authors are presented. These highlight important advances that build on the excellent contrast/visualization capabilities of MRI, strengthen image quality and robustness, or that improve the patient experience and throughput. Importantly, they show the success that can be realized by bringing innovators from academia and industry together into cooperative teams. Happy reading, and see you at RSNA! Editorial Board We appreciate your comments. Please contact us at [email protected] Review Board Lisa Chuah, Ph.D. Global Segment Manager Neurology Lars Drüppel, Ph.D. Global Segment Manager Cardiovascular MR Wilhelm Horger Application Development Oncology Michelle Kessler US Installed Base Manager Berthold Kiefer, Ph.D. Oncological and Interventional Applications Sunil Kumar S.L., Ph.D. Senior Manager Applications Reto Merges Head of Outbound Marketing MR Applications Heiko Meyer, Ph.D. Neuro Applications Edgar Müller Cardiovascular Applications Nashiely Sofia Pineda Alonso, Ph.D. Global Segment Manager Men’s and Women’s Health Silke Quick Global Segment Manager Body Imaging Heike Weh Clinical Data Manager Antje Hellwich Associate Editor Sven Zühlsdorff, Ph.D. Clinical Collaboration Manager, Chicago, IL, USA Ralph Strecker MR Collaborations Manager, São Paulo, Brazil Wellesley Were MR Business Development Manager Australia and New Zealand Gary R. McNeal, MS (BME) Advanced Application Specialist, Cardiovascular MR Imaging Hoffman Estates, IL, USA Peter Kreisler, Ph.D. Collaborations & Applications, Erlangen, Germany *WIP, the product is currently under development and is not for sale in the US and other countries. Its future availability cannot be ensured. 2 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 3
- 3. Content Content RCA 24 52 70 108 Positioning and immobilization of SWI with MAGNETOM ESSENZA Pictorial essay: Benign and malignant RT patients bone tumors Compressed sensing* 64 Imaging MS lesions in the cervical spinal cord Cover Growth with Body MRI. New Certainty in Liver MRI. Already today, many important staging and treatment decisions are taken on the basis of Body MRI. By using MRI for your body imaging examinations you can profit from excellent soft tissue contrast, high spatial and temporal resolution, as well as 3D and 4D data acquisition. Siemens Healthcare enables you to expand your Body MRI services. www.siemens.com/ growth-with-BodyMRI Clinical Head-to-Toe Imaging 6 Improving the robustness of clinical T1-weighted MRI using Radial VIBE* Tobias Block, et al. 12 New features of syngo MR D13 for improved whole-body DWI Thomas Doring, et al. Clinical Oncology 16 The metabolite ratio in spectroscopic imaging of prostate cancer Tom Scheenen, et al. 24 Evaluation of the CIVCO Indexed Patient Position System (IPPS) MRI- overlay for positioning and immobilization of radiotherapy patients Thomas Koch, et al. Technology 30 Making MRI quieter: Optimizing TSE with parallel imaging Eric Y. Pierre, et al. 35 Quiet T1-weighted 3D imaging of the central nervous system using PETRA* Masahiro Ida, et al. Clinical Neurology 44 How I do it: Acute stroke protocol in 6 minutes Kambiz Nael, et al. 52 SWI with 1.5T MAGNETOM ESSENZA José L. Ascencio, et al. 58 How I do it: Curve fitting of the lipid-lactate range in an MR Spectrum: Some useful tips Helmut Rumpel, et al. 64 T1w PSIR for imaging multiple sclerosis in the cervical spinal cord Bart Schraa Clinical Orthopedic Imaging 70 Pictorial Essay: Benign and malignant bone tumors Katharina Gruenberg, et al. Clincial Cardiovascular Imaging 100 Combined 18F-FDG PET and MRI evaluation of a case of hypertrophic cardiomyopathy using Biograph mMR Ihn-ho Cho, et al. 104 New generation cardiac parametric mapping: The clinical role of T1 and T2 mapping James C. Moon, et al. 108 Preliminary experiences with compressed sensing* multi-slice cine acquisitions for the assessment of left ventricular function J. Schwitter, et al. 117 Accelerated segmented cine TrueFISP of the heart on a 1.5T MAGNETOM Aera using k-t-sparse SENSE* Maria Carr, et al. The information presented in MAGNETOM Flash is for illustration only and is not intended to be relied upon by the reader for instruction as to the practice of medicine. Any health care practitioner reading this information is reminded that they must use their own learning, training and expertise in dealing with their individual patients. This material does not substitute for that duty and is not intended by Siemens Medical Solutions to be used for any purpose in that regard. The treating physician bears the sole responsibility for the diagnosis and treatment of patients, including drugs and doses prescribed in connection with such use. The Operating Instructions must always be strictly followed when operating the MR System. The source for the technical data is the corresponding data sheets. Content * WIP, the product is currently under development and is not for sale in the US and other countries. Its future availability cannot be ensured. 4 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 5
- 4. Clinical Head-to-Toe ImagingHead-to-Toe Imaging Clinical Improving the Robustness of Clinical T1-Weighted MRI Using Radial VIBE Kai Tobias Block1; Hersh Chandarana1; Girish Fatterpekar1; Mari Hagiwara1; Sarah Milla1; Thomas Mulholland1; Mary Bruno1; Christian Geppert2; Daniel K. Sodickson1 1 Department of Radiology, NYU Langone Medical Center, New York, NY, USA 2 Siemens Medical Solutions, New York, NY, USA Introduction Despite the tremendous developments that MR imaging has made over the last decades, one of the major limitations of conventional MRI is its pronounced sensitivity to motion, which requires strict immobility of the patient during the data acquisition. In clinical practice, however, suppression of motion is often not possible. As a consequence, MR images frequently show motion artifacts that appear as shifted object copies, which are well-known as ‘ghosting’ artifacts and which, depending on the artifact strength, can potentially obscure important diagnostic information. Ghosting artifacts pose a particular problem for abdominopelvic exams that need to be performed during suspended respiration. Because many patients struggle to adequately hold breath during the scan, the number of exams with suboptimal image quality is relatively high. This has impaired the acceptance of MRI as an imaging modality of choice in many abdominopelvic indications. Other widely utilized MRI applications such as head and neck imaging are also often affected by motion-induced ghosting artifacts, e.g., if patients are anxious, or if wthey cannot suppress swallowing or coughing during the exam. Radial k-space acquisition scheme The high sensitivity to motion results from the data-sampling strategy used in conventional MR imaging to spatially resolve the object. Conventional sequences acquire the data space Interestingly, although the advantages for clinical applications seem clear and although the idea of radial sampling has been known since the early days of MRI, the technique has not been widely employed in clinical practice so far. Radial sampling was first described by Lauterbur in his seminal MRI paper from 1973 [1]. However, because practical implementation required coping with a number of technical complexities, it was soon replaced by the Cartesian acquisition scheme which could be more easily and more robustly implemented on early MRI systems. These technical complexities include a more sophisticated image reconstruction, higher required homogeneity of the magnetic field, and the need for much more accurate and precise generation of time-varying gradient fields. Consequently, radial sampling has only been used sporadically in research projects while clinically established techniques are currently almost exclusively based on the Cartesian scheme. Over the last several years, however, it has become possible to resolve the majority of issues that prevented a practical application of radial sampling, in part through improvements of the MR hardware designs and in part through new algorithmic developments. Therefore, it is now for the first time feasible to utilize radial acquisitions routinely on unmodified clinical MRI systems, with sufficient reliability and robustness for clinical applications and with image quality comparable to that of the conventional Cartesian scans. Radial VIBE sequence The Radial VIBE sequence* is the first available works-in-progress sequence for Siemens MR systems that integrates these developments for volumetric acquisitions and provides radial k-space sampling in a fully seamless way, aiming at achieving higher robustness to motion and flow effects in daily practice. It is based on the conventional product VIBE sequence, which is an optimized T1-weighted 3D gradient echo sequence (3D FLASH) with various fat-saturation options. Radial sampling has been implemented using a 3D ‘stack-of-stars’ approach, which acquires the kx-ky plane along radial spokes and the kz dimension (k-space) using a sampling scheme along parallel lines (Fig. 1A), which is usually referred to as ‘Cartesian’ sampling. The acquired parallel lines differ by a fixed difference in the signal phase, which is why the scheme is also called ‘phase encoding’ principle. However, if the object moves during the exam, phase offsets are created that disturb the phase-encoding scheme. In a simplified view, it can be thought of as jittering of the sampled lines, which causes gaps in the k-space coverage and results in aliasing artifacts along the phase-encoding direction from improper data sampling. Hence, the Cartesian geometry is inherently prone to motion-induced phase distortions. Even if navigation or triggering techniques are used to minimize phase inconsistencies within the acquired data, a certain amount of residual ghosting artifacts is almost always present. The situation can be improved when changing the k-space acquisition to a different sampling geometry. One promising alternative is the ‘radial’ sampling scheme, which acquires the data along rotated spokes (Fig. 1B). Due to the overlap of the spokes in the center, gaps in the k-space coverage cannot occur if individual spokes are ‘jittered’ and, therefore, appearance of ghosting artifacts is not possible with this scheme. Furthermore, the overlap has a motion-averaging effect. Data inconsistencies can instead lead to ‘streak’ artifacts. However, in most cases the streaks have only a mild effect on the image quality, and they can easily be identified as artifacts due to their characteristic visual appearance (e.g., Fig. 3B). Because the artifacts appear mainly as ‘texture’ added to the underlying object, the likelihood that lesions get obscured is significantly lower than for the more dominant Cartesian ghosting artifacts. with conventional sampling, resulting in cylindrical k-space coverage (see Fig. 2). This trajectory design enables use of time-efficient fat-saturation methods, such as Quick FatSat or SPAIR, with minimal artifact strength, which is important as radial scans should be performed with fat suppression for most applications. Although Cartesian acquisition steps are employed along the kz dimension, a high degree of motion robustness is achieved due to the use of an incoherent temporal acquisition order. The Radial VIBE sequence can be used on the full range of Siemens MR systems, including systems from the B-line generation (e.g., MAGNETOM Avanto, Trio, Verio) and D-line generation (e.g., MAGNETOM Skyra, Aera), and it can also be used on the Biograph mMR MR-PET system as well as Siemens’ 7T** systems. Because the sequence does not require any 2 modification of the MR hardware or reconstruction system, it can be deployed to installed systems and used clinically for fat-saturated T1-weighted exams as a motion-robust alternative to 3D GRE, VIBE, MPRAGE, or 2D TSE sequences. Clinical applications and results Over the last two years, the sequence has been tested extensively at NYU Langone Medical Center to evaluate the achievable image quality across various MR systems in daily routine applications. Radial VIBE scans were added to clinical protocols under IRB approval in more than 5000 patient exams and compared to established reference protocols. Several clinical studies have been performed or are ongoing that investigate the improvement in diagnostic accuracy resulting from the absence of ghosting artifacts. Free-breathing abdominal imaging A key application of the Radial VIBE sequence is imaging of the abdomen and/or pelvis before and after injection of a contrast medium, which is conventionally performed during suspended respiration. With Radial VIBE, it is possible to acquire the data during continued shallow breathing, which therefore can be the preferred exam strategy for patients who are unable to sustain the normally 1A 1B (1A) Conventional ‘Cartesian’ MRI sampling scheme along parallel lines, and (1B) radial sampling scheme along rotated spokes that overlap in the center of k-space. 1 Stack-of-stars trajectory as implemented by Radial VIBE, which employs radial k-space sampling in the kx-ky plane and Cartesian sampling along kz. 2 kz * Radial VIBE is a prototype for StarVIBE. StarVIBE is now 510k released and is available for 1.5T MAGNETOM Aera and 3T MAGNETOM Skyra. Radial VIBE is work in progress. ** The product is under development and not commercially available yet. Its future availability cannot be ensured. This research system is not cleared, approved or licensed in any jurisdiction for patient examinations. This research system is not labelled according to applicable medical device law and therefore may only be used for volunteer or patient examinations in the context of clinical studies according to applicable law. 6 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 7
- 5. Clinical Head-to-Toe ImagingHead-to-Toe Imaging Clinical 7A 7B required breath-hold time, such as elderly or severely sick patients. A blinded-reader study by Chandarana et al. demonstrated that the average image quality obtained with free-breathing radial acquisition is comparable to conventional breath-hold exams and significantly better than free-breathing exams with Cartesian acquisition [2]. As an example, figure 3 compares a free-breathing Radial VIBE exam to a conventional Cartesian exam of a patient with insufficient breath-hold capability. The Radial VIBE image is affected by a certain amount of streak artifacts but clearly depicts a lesion in the right lobe of the liver, which is fully obscured in the Cartesian scan. High-resolution abdominopelvic imaging The ability to acquire data during continued respiration also has advantages for the examination of patients with proper breath-hold capacity. With conventional Cartesian sequences, the achievable spatial resolution in abdominopelvic exams is limited by the amount of k-space data obtainable within typical breath-hold durations of less than 20 sec. Because Radial VIBE eliminates the need for breath holding, it is possible to sample data over several minutes and, thus, to increase the spatial resolution by a significant factor. Figure 4 demonstrates this possibility for an isotropic 1 mm high-resolution scan of the liver 20 min after injection of Gadoxetate Disodium, which provides clearly sharper visualization of the biliary duct compared to the corresponding Cartesian protocol. In figure 5, the achievable resolution improvement is shown for the case of MR enterography, which is another good candidate for Radial VIBE due to the higher overall robustness to the bowel motion. Pediatric imaging During the clinical evaluation phase, Radial VIBE demonstrated particular value for the application in pediatric* patients. Pediatric exams are often conducted under general anesthesia or deep sedation, which makes active breath holding impossible. Therefore, conventional abdominopelvic scans are in most cases affected by respiration artifacts that impair the achievable effective resolution and diagnostic accuracy. Due to the inherent motion robustness, much sharper and crisper images are obtained with Radial VIBE, as evident from the depiction of small cysts in the kidneys of a patient with Tuberous Sclerosis shown in figure 6. A retrospective blinded-reader study of our case collection revealed that 8% of all lesions were only identified with Radial VIBE but missed in the corresponding Cartesian reference exams [3]. In young neonatal patients, sedation is usually avoided due to the higher risk of potential adverse effects. Imaging these patients is challenging because they often move spontaneously in the scanner. Also in this patient cohort Radial VIBE provides improved image quality and reliability, which is demonstrated in figure 7 for a brain exam of a 4-day-old patient, in this case compared to a Cartesian MPRAGE protocol. *MR scanning has not been established as safe for imaging fetuses and infants less than two years of age. The responsible physician must evaluate the benefits of the MR examination compared to those of other imaging procedures. 4 5 6 Abdominopelvic exam of a sedated pediatric patient with Tuberous Sclerosis. (6A) Because suspending respiration is not possible under deep sedation, conventional exams are affected by respiration artifacts. (6B) Radial VIBE provides significantly sharper images with improved spatial resolution, as visible from the small cysts in the kidneys. 6A 6B Brain exam of a 4-day-old* patient using (7A) conventional MPRAGE and (7B) Radial VIBE sequence. Due to vigorous patient activity, the MPRAGE scan is affected by strong ghosting artifacts, while Radial VIBE provides diagnostic image quality. *MR scanning has not been established as safe for imaging fetuses and infants less than two years of age. The responsible physician must evaluate the benefits of the MR examination compared to those of other imaging procedures. 7 (3A) Conventional VIBE exam of a patient failing to hold breath during the acquisition and (3B) Radial VIBE acquisition during free breathing. Radial VIBE provides significantly higher image quality and reveals a lesion in the liver (arrow) not seen on the conventional scan 3 3A 3B (4A) Conventional breath-hold VIBE exam of a patient 20 min after injection of Gadoxetate Disodium. (4B) Because Radial VIBE exams can be performed during continued respiration, data can be acquired over longer time, resulting in clearly improved resolution (here 1.0 mm isotropic). 4A 4B MR enterography using (5A) conventional VIBE and (5B) Radial VIBE acquisition. The higher motion robustness achieved with Radial VIBE leads to sharper images and improved resolution. 5A 5B 8 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 9
- 6. Clinical Head-to-Toe ImagingHead-to-Toe Imaging Clinical 8A 8B 9B 9C Examination of the neck and upper chest using (8A) a conventional 2D TSE sequence and (8B) Radial VIBE. Because of the respiration and strong blood flow, the TSE scan shows drastic artifacts. Two suspicious lesions are more clearly visible on the Radial VIBE exam (arrows). Imaging of the neck and upper chest Although imaging of the head and neck region appears less critical at first glance, severe motion-related artifacts occur quite often in routine exams. Conventional neck protocols usually include slice-selective T1-weighted TSE sequences, which are especially sensitive to motion and flow. If patients are unable to suppress swallowing or coughing during the acquisition, images are rendered non-diagnostic. Furthermore, adequate examination of the upper chest region is often not possible because of drastic artifacts from respiration and strong blood flow in the proximity of the heart. Radial VIBE exams are a promising alternative for this application and are largely unaffected by swallowing, minor head movements, or flow, which is illustrated in figure 8. The sequence also maintains a convincing sensitivity to chest lesions in the presence of respiratory motion [4]. Because Radial VIBE scans are immune to ghosting artifacts, exams can be performed with high isotropic spatial resolution, which allows for retrospective reconstruction in multiple planes (MPRs). In this way, it is possible to substitute multiple conventional slice-selective protocols in varying orientation with a single Radial VIBE high-resolution scan. A representative example is shown in figure 9. Imaging of the orbits, inner auditory canal, and full brain Finally, the sequence also offers improved sharpness and clarity for the examination of the orbits. When patients move the eyes or change the position of the eyelids during the exam, conventional protocols show a band of strong ghosting artifacts along the phase-encoding direction, which can make identifying pathologies a difficult task. Radial VIBE provides cleaner depiction of the optic nerves and improved suppression of intra- and extraconal fat [5]. Flow effects from surrounding larger blood vessels can lead to mild streak patterns but are less prominent than for most Cartesian protocols and can be additionally attenuated with the use of parallel saturation bands. The possibility to create high-resolution MPRs is another advantage of using Radial VIBE for this application, which is demonstrated in figure 10 for a patient with optic nerve sheath meningioma. In a similar way, the sequence can be applied for examinations of the inner auditory canal (IAC) or the full brain, in which a particularly high sharpness of vessel structures is achieved. Conclusion The large number of successful patient exams of various body parts conducted with Radial VIBE over the last two years demonstrates that radial sampling is now robust and reliable for routine use on standard clinical MR systems. Due to the higher resistance to patient motion and the absence of ghosting artifacts, improved image quality can be obtained in applications where motion-induced image artifacts are a common problem. In particular, the Radial VIBE sequence enables exams of the abdomen and upper chest during continued shallow respiration, which can be a significant advantage for patients that struggle to adequately hold breath. Furthermore, the sequence enables reconfiguring exam protocols towards higher spatial resolution and allows consolidating redundant acquisitions into MPR-capable isotropic scans. Because the sequence works robustly on existing MRI hardware, Radial VIBE has the potential to find broad application as motion-robust T1-weighted sequence alternative and will complement the spectrum of clinically established imaging protocols. 8 Sag Cor Tra Neck exam using transversal Radial VIBE acquisition with 1 mm isotropic resolution. Due to the robustness to swallowing and minor head motion, high quality 3D scans are possible that can be reconstructed in multiple planes (MPR). This enables consolidating redundant 2D protocols with varying scan orientation. 9 9A References 1 Lauterbur PD. Image formation by induced local interactions: Examples employing nuclear magnetic resonance. Nature 242:190–191, 1973. 2 Chandarana H, Block KT, Rosenkrantz AB, Lim RP, Kim D, Mossa DJ, Babb JS, Kiefer B, Lee VS. Free-breathing radial 3D fat-suppressed T1-weighted gradient echo sequence: a viable alternative for contrast-enhanced liver imaging in patients unable to suspend respiration. Invest Radiology 46(10):648-53, 2011. 3 Chandarana H, Block KT, Winfeld JM, Lala SV, Mazori D, Giuffrida E, Babb JS, Milla S. Free-breathing contrast-enhanced T1-weighted gradient-echo imaging with radial k-space sampling for paediatric abdominopelvic MRI. European Radiology, September 2013. 4 Chandarana H, Heacock L, Rakheja R, Demello LR, Bonavita J, Block KT, Geppert C, Babb JS, Friedman KP. Pulmonary Nodules in Patients with Primary Malignancy: Comparison of Hybrid PET/MR and PET/CT Imaging. Radiology 268(3):874-81, 2013. 5 Bangiyev L, Raz E, Block KT, Hagiwara M, Yu E, Fatterpekar GM. Contrast-enhanced radial 3D fat-suppressed T1-weighted gradient echo (Radial-VIBE) 10 Multiplanar reconstructions of a transversal Radial VIBE exam with 0.8 mm resolution. The sequence achieves good fat suppression and provides sharp depiction of the optic nerves without artifacts from eye motion. An abnormal contrast enhancement of the left optic nerve is clearly visible (arrows), which is indicative of an optic nerve sheath meningioma. sequence: A viable and potentially superior alternative to conventional T1-MPRAGE with water excitation and fat-suppressed contrast-enhanced T1W sequence for evalu-ation of the orbit. ASNR Annual Meeting 2013: O-413. Contact Tobias Block, Ph.D. Assistant Professor of Radiology New York University School of Medicine Center for Biomedical Imaging New York, NY 10016 USA [email protected] 10A 10B 10C 10 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 11
- 7. 3 Same patientas in figure 3 with curved reconstructed MPRs in the coronal plane. (4A) Clearly visible the discontinuity in the spinal cord due to composing errors. (4B) Significantly better registration of the stations. 4 4B New Features of syngo MR D13 for Improved Whole-Body Diffusion-Weighted MRI Thomas Doring1; Ralph Strecker2; Michael da Silva1; Wilhelm Horger3; Roberto Domingues1; Leonardo Kayat Bittencourt1; Romeu Domingues1 1 CDPI and Multi-imagem (DASA), Rio de Janeiro, Brazil 2 Siemens Ltda, São Paulo, Brazil 3 Siemens Healthcare, Erlangen, Germany Backround Whole-body diffusion-weighted imaging (WB-DWI) is gaining in clinical importance for oncological imaging. It has been shown to be a promising tool principally for tumor detection, tumor characterization, and therapy monitoring of bone metastases [1, 2]. The clinical implementation of WB-DWI aims for standardization of the acquisition protocol. For this reason, further improvements of data acquisition, analysis and display of the results are requested. The recently launched new software version syngo MR D13 provides several new features for WB-DWI such as variable averaging of the b-values, inline composing, and a Bias Field Correction (BiFiC) filter in order to overcome previously existing limitations such as long acquisition times, mis-registration between, and intensity inhomogeneities across image stations. New features in syngo MR D13 b-value specific averaging One limitation of the broader clinical usability is the long acquisition time of over 25 minutes for head-to-pelvis WB-DWI MRI. For more efficient scanning the new feature b-value specific averaging was developed: This allows us to set the number of averages (NEX) for each b-value individually. The current product protocol uses a b-value of 50 with NEX 2, and a b-value of 800 with NEX 5, resulting in a reduction of scan time of 30%, when compared to the previously used NEX 5 for all b-values (Fig. 1). A similar image quality can be achieved for lower b-value images with lower NEX with almost no impact on the signal-to-noise ratio (SNR) for the calculated ADC. New composing mode diffusion An inline composing filter can be activated within the diffusion sequence of syngo MR D13 on the diffusion taskcard of the sequence (Fig. 2). The composing itself is a fully automatic process and creates for each b-value a continuous stack of composed images as an individual new series. Similarly a new series for the optional calculated b-value images is generated. Depending on the local shim situation the frequency differences between neighbored stations can lead to discontinuities of anatomical structures like the ‘broken-spine’ artifact. During the composing step a correction is applied showing a much smoother transition of local anatomy (Fig. 2). In older software versions (syngo MR D11) composing had to be done manually within the syngo.3D taskcard by dragging and dropping all trace-weighted series at once with no possibility to correct for any discontinuities in the anatomy. The new inline composing feature significantly improves the acquisition workflow of the technologist as it allows to load the single composed series to syngo.3D for the generation of the 3D reformatted maximum intensity projection (MIP) images. Bias Field Correction (BiFiC) filter The BiFiC filter as a homomorphic filter aims to normalize inhomogeneities in image intensities from multi-station measurements such as whole-spine imaging. After completing the inline composing step the filter is automatically applied to the composed 3D continuous image stack and saved as the new composed series. The strength of the filter (weak, medium, strong) can be set within the diffusion task card (Fig. 2, arrow). In the Diffusion taskcard it is now possible to select the number of averages for each b-value individually (Here: b50 NEX 2, b800 NEX 5, red arrows). 1 1 4A 64-year-old female patient, after surgery, with endometrial stromal sarcoma, that has evolved with bony metastases. Acquisition parameters: 1.5T MAGNETOM Aera, echoplanar imaging diffusion sequence with fat suppression (STIR), TR 14100 ms, TE 79 ms, TI 180 ms, 4 stations, 50 slices with 5 mm slice thickness, no gap, voxel size 1.7 × 1.7 × 5 mm3, b50 with 2 and b800 with 5 averages. (3A) b800 manually composed images within the syngo.3D tool shows the artifact in the neck shoulder transition where the stations are joined (arrows). (3B) The new inline composing feature demonstrates significantly better composing of the neck and shoulder transition. 3A 3B By checking the Inline Composing box in the diffusion taskcard of the sequence the automatic composing modus is activated. The strength of the BiFiC filter can be set to weak, medium or strong (red arrows). 2 2 Head-to-Toe Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 13 Clinical Head-to-Toe Imaging 12 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 8. References 1 PadhaniAR, Koh DM, Collins DJ. Whole-Body Diffusion-weighted MR Imaging in Cancer: Current Status and Research Directions. Radiology. 2011 Dec; 261(3): 700-718. Contact Thomas Doring, Ph.D. Clínica de Diagnóstico por Imagem Rio de Janeiro Brazil [email protected] Clinical Head-to-Toe Imaging Conclusion The new features within the product sequence of syngo MR D13 improve the image quality of whole-body DWI when compared to older software versions. The new inline composing filter, in particular, shows good results in the neck-shoulder transition compared to the previously manual technique in syngo MR D11 that was not able to recover discontinuities in the spine. Improvements in the clinical workflow are also addressed. 5 56-year-old male patient underwent WB-DWI. Coronal (5A) and Sagital (5B) MIP from b800. It can be seen that the BiFiC Filter works well in the body although the signal in the head/neck is cancelled out by susceptibility artifacts. Patient had a PSA of 2580 ng/ml, with three prior negative biopsies and one negative transurethral prostate resection. Digital rectal examination was normal. Multiparametric prostate MR revealed a highly suspicious lesion on the right anterior the peripheral zone, along with massively enlarged iliac and periaortic lymph nodes seen on WB-DWI. 5A 5B 2 Initial Experience with Whole-Body Diffusion-Weighted Imaging in Oncological and Non-Oncological Patients. Marcos Vieira Godinho, Romulo Varella de Oliveira, Clarissa Canella, Flavia Costa, Thomas Doring, Ralph Strecker, Romeu Cortes Domingues, Leonardo Kayat Bittencourt. MAGNETOM Flash 2/2013: 94-102. 14 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world Growth with Body MRI. New Certainty in Liver MRI. Are you striving to always catch the correct timepoint of lesion enhancement throughout the arterial phase? With TWIST-VIBE you can! Do you have patients who cannot hold their breath throughout an MRI scan? With StarVIBE they don’t need to! Are you facing challenges in Body MRI? With us you will be able to solve them! Find out how we do that and see what other users say: www.siemens.com/growth-with-bodyMRI
- 9. Clinical Oncology OncologyClinical The Metabolite Ratio in Spectroscopic Imaging of Prostate Cancer Alan J. Wright; Thiele Kobus; Arend Heerschap; Tom W. J. Scheenen Radboud University Nijmegen Medical Centre, Radiology Department, Nijmegen, The Netherlands Introduction Prostate cancer is the second leading cause of cancer related death in Western countries [1]. The prevalence of the disease is very high, but many men diagnosed with the disease will die from unrelated causes. This is because prostate cancer very often is a disease of old age that grows slowly. Common treatment for prostate cancer in clinical practice involves radical resection of the entire gland or radiotherapy with a dose distributed over the whole organ. Provided that the cancer has not metastasized, these therapies are curative, though concern over their side effects has led to patients and their doctors delaying this treatment and, instead, entering into active surveillance or watchful waiting programs. In order for patients to safely forgo curative treatment, it is essential to characterize their disease: to determine that it is sufficiently benign that growth will be slow and metastasis improbable. Selecting these patients, with low risk disease, that are appropriate for active surveillance requires accurate diagnosis of not just the presence of tumor, but how aggressive it is: i.e. how fast it is growing and how likely it is to metastasise to the lymphatic system. Magnetic resonance imaging (MRI) is an emerging technique for making this patient selection. It can diagnose the presence of tumor, localize it in the organ and provide information as to how aggressive it is. The MRI exams employed for this purpose usually involve multiple imaging sequences including a T2-weighted sequence, diffusion-weighted imaging (DWI) and one or more further techniques such as dynamic contrast enhanced MRI (DCE-MRI) or Proton Magnetic Resonance Spectroscopic Imaging (1H MRSI) [2]. Radiologists can read the different imaging modalities to decide the location, size and potential malignancy of the tumor which are all indicators of its metastatic potential. Acquiring and reporting imaging data in this way is known as multiparametric (mp) MRI. MRSI is the only mpMRI methodology that acquires data from molecules other than water [19]. A three dimensional (3D) 1H MRSI data set consists of a grid of spatial locations throughout the prostate (see Fig. 1) called voxels. For each voxel a spectrum is available. Each spectrum consists of a number of peaks on a frequency axis, corresponding to resonances from protons with a certain chemical shift in different molecules. The size of a peak at a certain frequency (chemical shift) corresponds to the amount of the metabolite present in the voxel. In this way MRSI measures the bio-chemicals in regions of tissue in vivo without the need for any external contrast agent or invasive procedures. Examples of spectra from two voxels, acquired at a magnetic field strength of 3 Tesla (3T), are given in figure 1B, C, which clearly shows the differing profiles that are characteristic of benign prostate tissue and its tumors. Important metabolites in prostate MRSI The initial papers on in vivo prostate MRSI were performed at a magnetic field strength of 1.5T [3-5], and three assignments were provided for the observed resonances: choline, creatine and citrate (Fig. 2). The small number of these assignments reflected the simplicity of the spectrum, which contained two groups of resonances: one in the region of 3.3 to 3 ppm, which will be referred to as the choline-creatine region, and another at 2.55–2.75 ppm, which shall be called the citrate group. These assignments related to what were believed to be the strongest metabolite resonances. People should be aware however, that the assignments are representative of multiple similar molecules. The choline assignment reflects the methyl resonances from multiple compounds containing a choline group (Fig. 4): choline, phosphocholine and glycerophosphocholine. Similarly, creatine refers to both creatine and phosphocreatine. In between the choline and creatine signals another group of resonances are present: the polyamines (mainly spermine and spermidine). The citrate resonances are from citrate only but can have a complicated shape, although in vivo at 1.5T they give the appearance of a single peak. Nowadays a magnetic field strength of 3T is used more and more for prostate spectroscopic imaging, which gives opportunities to better resolve the choline, polyamines, creatine resonances, but also changes the shape of the citrate signal. Larger choline signals are associated with tumor in nearly all cancers [6]. High choline signals are interpreted as being evidence of rapid proliferative growth and, more directly, the increased membrane turnover required for cell division. Membranes contain phospholipids: phosphatidyl choline and phosphatidyl ethanolamine, which are synthesised by a metabolic pathway involving choline- containing metabolites known as the Kennedy pathway. It is in the synthesis and catabolism of these products, upregulated in proliferative tumor growth, that causes the increase in these signals. The large amplitude of citrate resonances observed in prostate tissue is due to an altered metabolism particular to this gland. Prostate tissue accumulates high concentrations of zinc ions which inhibit mitochondrial aconitase, leading to a build up of citrate in the prostate’s epithelial cells [7]. This citrate is further secreted into the ductal spaces of the prostate as (1A) T2-weighted MR image of a transverse section through a prostate with an overlaid grid of MRSI spectra from voxels within the prostate. 1 (1B) One example spectrum shown on a ppm scale from a region of benign prostate tissue. (1C) A spectrum from another voxel that, in this case, co-localises to a region of tumor. 1 Benign tissue Choline-Creatine Citrate 3.2 2.9 2.6 ppm Tumortissue Choline 3.2 2.9 2.6 ppm Examples of 1H MRSI spectra acquired from benign prostate tissue and tumor at 1.5T. 2 1A 1B 1C 2A 2B 16 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 17
- 10. Clinical Oncology OncologyClinical T2-weighted MRI image of a transverse section through a prostate as shown in figure 1 with an overlaid grid of MRSI-voxel data displayed as fitted spectra (5A). 5 1.5T Cho Ci ppm Choline Phosphocholine Glycero-phospholine CH3 N CH3 CH3 OH CH3 N CH3 CH3 OPO3H2 CH3 N CH3 CH3 Creatine Phospho-creatine Citrate CH3 N NH2 NH O OH CH3 N NH PO3H2 OH NH part of prostatic fluid, which has a high concentration of this metabolite. Prostate carcinomas do not accumulate zinc ions, so they do not have this high citrate concentration. The increased presence of tumor cells within a 1H MRSI voxel can, therefore, have two diminishing effects on the observed citrate signals: epithelial cells that accumulate citrate can transform into, or be replaced by, tumor which has low citrate, or the lesion can grow through the ductal spaces, 3T Cho Ci ppm OPO2HO OH OH O O OH O OH OH HH HH OH O thus displacing the prostatic fluid. The relative contribution of each of these two physiological changes, whether we are observing tumor formation and malignant progression or a histological change in tumor invasion of ductal structure, is not yet known. It is, however, clear that there is an inverse correlation of the levels of citrate metabolite and tumor cell density with some evidence to support a similar correlation with the aggressiveness of the tumor as well [8]. The introduction of a metabolite ratio To transform the described changes in choline and citrate signals between benign (high citrate) and tumorous tissue (low citrate, high choline) into a marker for prostate cancer, the metabolite ratio was introduced [3-5]. The signal intensities of the different spectral peaks were quantified by simple integration of the two groups of resonances (the choline-creatine region and the citrate group), and the results were expressed as a ratio of the two. This gave the choline plus creatine over citrate ratio (abbreviated to CC/C [4]) or its inverse (with citrate as the numerator, [3, 5]). With choline in the numerator and citrate in the denominator, it became a positive biomarker for the presence of cancer. Acquiring the MRSI data sets As the prostate is embedded in lipid tissue, and lipids can cause very strong unwanted resonance artefacts in prostate spectra, the pulse sequence to acquire proton spectra is equipped with five properties to keep lipid signals out and retain optimal signals-of-interest in the whole prostate [9]. 1. Localization of the signal with slice-selective pulses. The point resolved spectroscopy sequence (PRESS) is a combination of one slice selective excitation pulse and two slice selective refocusing pulses leading to an echo at the desired echo time. The three slices are orthogonal, producing an echo of the volume-of- interest (crossing of three slices) only. 2. Weighted acquisition and filtering. Proton MRSI data sets are acquired using a phase encoding technique where the gradients across spatial dimensions are varied with each repeat of the pulse sequence. By using weighted averaging of these phase encoding steps (smaller gradient steps are averaged more often than larger gradient steps) and adjusted filtering of the noise in these weighted steps, the resulting shape of a voxel after the mathematical translation of the signal into an image (Fourier Transform) is a sphere. Contrary to conventional acquisition without filtering, the spherical voxels after filtering are not contaminated with signals from non-neighboring voxels. 3. Frequency-selective water and lipid suppression. The pulse sequence has two additional refocusing pulses Structural formulas of the key small molecule metabolites observed in the spectra of prostate tissue. For each group, cholines and creatines, the common moiety is highlighted in red. The protons that give the MR spectral resonances present in the choline-creatine region are indicated in green. Choline-containing metabolites have 9 co-resonant protons in the region 3.2–3.25 ppm. Creatines have three co-resonant protons at 3.05 ppm. Citrate has four protons that resonate at two chemical shifts (2.6 and 2.7 ppm), one for each proton in a pair bonded to the same carbon. This pair also has a coupling between them and the symmetry of the whole molecule ensures that two protons co-resonate at each frequency. 4 Simulated spectral shape of Cho and Ci for typical echo times (120 ms at 1.5T, 145 ms at 3T). Identical concentrations, i.e. scale factors, are applied, but different line broadening of the signals (4 Hz at 1.5T; 8 Hz at 3T). Note the difference in the spectral shape of Ci and the different peak amplitude ratios for Cho/Ci. 3 3A 4 3B that only touch upon water and lipid signals. Together with strong crushing gradients, signals from water and lipids are suppressed. 4. Outer volume suppression. Around the prostate, slice-selective pulses can be positioned to suppress all signals in the selected slabs. These slices can be positioned quite close to the prostate, even inside the PRESS-selected volume-of-interest. 5. Long echo time. To accommodate all localization and frequency selective pulses, the echo time of 1H MRSI of the prostate is around 120 ms at 1.5T and 145 ms at 3T. At longer echo times, lipid signals decay due to their short T2 relaxation time. The prostate is small enough (< 75 cubic centimetres) to allow a 3D 1H MRSI data set to be acquired, with complete organ coverage, within 10 minutes of acquisition time. The nominal voxel size is usually around 6 × 6 × 6 mm, which after filtering as described above results in a true voxel size of 0.63 cm3. Spectral patterns Due to multiple different protons in the molecule, a single metabolite can have multiple resonances. If interactions exist between protons within a metabolite, the shape of a spectral peak can be complicated. A resonance group of protons that has a mixture of positive and negative parts is said to have a dispersion component in its shape; a symmetrical-positive peak is referred to as an absorption shape. Choline and creatine resonances appear as simple peaks (singlets), 5A 5B 5C 5 Two spectra (shown previously in figure 1) with their fitted model metabolite signals (5B, C). 18 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 19
- 11. Clinical Oncology OncologyClinical Table 1: Typical integral values of the CC/C ratio in prostate tissue at 1.5T [17] and pseudo integral values of CC/C at 3T [18]: Tissue 1.5 Tesla* 3 Tesla** Non-cancer peripheral zone 0.28 (0.21– 0.37) 0.22 (0.12) Non-cancer central gland*** 0.36 (0.28–0.44) 0.34 (0.14) Cancer 0.68 (0.43–1.35) 1.3 (3.7) *median and 25th and 75th percentile **mean and standard deviation ***combined transition zone and central zone T2-weighted MR image of a transverse section through a prostate as shown in figure 1 with an overlaid grid of MRSI-voxel data displayed as spectra (6A) and as the pseudo integral CC/C ratio (6B) from metabolite fitting of the spectra. The ratio data points are interpolated and shown on a color scale of values from 0 (blue) to 3 (red). The tumor containing region of the prostate corresponds to the higher ratio values: the cyan-red area in the image. although they very often cannot be separated from each other as they overlap within in vivo spectral linewidths. The structure of citrate, given in figure 4, results in protons at two different chemical shifts, with coupling between each proton and one other (a strongly coupled spin system). The spectral shape of these protons depends on their exact chemical shift, the coupling constant between them, the pulse sequence timing and the main magnetic field strength. At an echo time of 120 ms at 1.5T (and a very short delay between excitation and first 180 degree refocusing pulse), the spectral peaks of citrate are close to a positive absorption mode. The spectral shape consists mainly of an inner doublet with small side lobes on the outer wings. Together with line broadening the citrate protons quite closely resemble a single, somewhat broadened peak. The small side lobes around this peak are hardly detectable over the spectral noise in vivo. At 3T with an echo time of 145 ms (examples given in Fig. 1), the negative dispersion components of the citrate shape cannot be ignored. Its side lobes are substantially larger and reveal also some negative components [10]. Therefore the area under the curve, the integral, is substantially smaller at 3T than at 1.5T. Because of its complicated shape, it is essential at 3T to incorporate this shape in quantification of the signal. Signal quantification by integration or metabolite fitting The size of the peaks of the individual resonances represent the amount of the metabolite present in the voxel. Integration provides a simple method to quantify the spectra, as long as all signals have an absorption shape. Although it cannot discriminate between overlapping resonances, as long as overlapping signals (choline and creatine) are summed in a ratio this does not matter. With clear separation between citrate resonances and the choline-creatine region, the CC/C ratio can be calculated. However, as pointed out earlier, the spectral shape of citrate is not straightforward, and ignoring the small satellites at 1.5T, or simply integrating the large dispersion parts of the signal at 3T, would inevitably lead to underestimation of the total citrate signal intensity. An alternative is to fit the spectra with models of the citrate resonances with their expected shape. The shape can either be measured, using a solution of citrate placed in the MRI system and a spectrum acquired with the same sequence as the in vivo data, or it can be calculated using a quantum mechanical simulation (Fig. 3). By this process of spectral fitting, models of each metabolite’s spectral peaks are fit to the total spectrum and the intensities of each fitted model are calculated. A linear combination of the metabolite models is found by the fitting routine such that Data = C1 *choline model + C2 * creatine model + C3 *citrate model + baseline Eqn1. The coefficients C1-4 give the relative concentrations of the individual metabolites. When fitting with syngo.via, the result of a fit to a spectral peak can be expressed in two ways: as an integral value, which describes the area under the fitted spectral peak, or as a relative concentration (incorporating the number of protons in the corresponding peak) of the metabolite, called the scale factor (SF) of the metabolite. As noted earlier, the integral value of citrate is different for 1.5 vs. 3 Tesla due to the different spectral patterns and would also change if pulse sequence timing other than standard 6 would be used. If the scale factor is multiplied with the number of resonating protons (#H), it represents the intensity of a signal, in relation to the integral value of a pure singlet of one resonating proton in absorption mode. We call this entity pseudo integral, which is calculated as A. pseudo integral (Metabolite) = #H * SF (Metabolite). For citrate this pseudo integral is perhaps best described as the numerical integral of the magnitude (all negative intensity turned positive) of the citrate spectral shape, ignoring signal cancellations of absorption and dispersion parts of the shape. The spectral fits are shown for the two spectra in figure 5 with model spectra of the three metabolites choline, creatine and citrate. It can be seen from these spectra that the relative amplitudes of the metabolites vary between the benign and the tumor spectrum. As expected, the benign spectrum has a higher citrate amplitude while the tumor has a greater choline amplitude, relative to the other metabolites. Combined in the CC/C ratio, the positive biomarker for the presence of tumor in the prostate is calculated. Depending on the used quantification (spectral integration without fitting (a), fitted relative concentrations (b) or pseudo integrals (c)) the CC/C can be calculated by: (a) {Integral(Choline) + Integral ( Creatine)} / Integral(Citrate) (b) {SF(Choline) + SF(Creatine) } / SF(Citrate) (c) {9*SF(Choline) + 3*SF(Creatine)} / 4*SF(Citrate), respectively. The numbers in the last equation correspond to the number of protons of the different signals. Generally, use of the pseudo integral ratio is strongly preferred, as it is least sensitive to large variations in individual metabolite fits in overlapping signals (choline and creatine). Note (again) that this pseudo integral ratio does not aim to provide a ratio of absolute metabolite concentrations, as this is very difficult with overlapping metabolite signals, partially saturated metabolite signals due to short TR (T1 effects), and variation in signal attenuation due to the use of a long echo time (T2 effects). Now, what could be the effect on the ratio if further metabolites are included in the fitting? Could even polyamines be incorporated in the analysis [11]? After separate fitting, the main focus of the analysis could just be on choline and citrate, which have opposite changes in intensity with cancer, to make a simpler and potentially more sensitive choline/ citrate ratio. Various metabolite ratios have been proposed [12, 13], and there is certainly value in using choline over creatine as a secondary marker of tumor malignancy that can give complementary information to the CC/C ratio [14-16]. However, any of these interpretations are limited by how well the individual metabolite resonances can be resolved. At 3T the choline, polyamines and creatine resonances all overlap (Figs. 1 and 5). In practice this lack of resolution in the spectrum translates to errors in the model fitting where one metabolite can be overestimated at the expense of another. For example a choline over citrate ratio could be underestimated if the polyamines fit was overestimated and accounted for some of the true choline signal. While acquisition and fitting methods are being actively researched to improve the individual quantification of these metabolites, it is more reliable to stick to the pseudo-integral CC/C ratio. Once reliably calculated, the CC/C ratio combines the essence of the observable spectroscopic data into a single quantity that can be displayed on an image (Fig. 6), combining the key information into a simple to read form for radiological reporting. Published values of the ratios for tumor and benign tissue, which are calculated in a similar way to the syngo.via fitting, are listed in table 1. Future perspective of MRSI for prostate cancer The CC/C ratio is the most used method for interpreting 1H MRSI data of prostate and prostate cancer. It remains, essentially, the integral of the choline-creatine region divided by the citrate region, a simple combination of the metabolite information in a single-value marker that is sensitive to the presence of tumor. The use of areas under the resonances in the ratio has the implication that the absolute value of this biomarker is largely dependent on the acquisition sequence used. Any change in field strength, the pulses or pulse timings will change resonance amplitude and shape due to T1 and T2 relaxations and the scalar couplings of especially citrate. Values of the ratio quoted in the literature for tumor or benign tissues depend strongly on how the 6A 6B 20 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 21
- 12. Clinical Oncology PediatricImagHinogw C-lIi-ndioc-aitl References 1 Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012; 62(1):10-29. 2 Hoeks CMA, Barentsz JO, Hambrock T, Yakar D, Somford DM, Heijmink SWTPJ, Scheenen TWJ, Vos PC, Huisman H, van Oort IM, Witjes JA, Heerschap A, Fütterer JJ. Prostate Cancer: Multiparametric MR Imaging for Detection, Localization, and Staging. Radiology 2011; 261: 46 –66. 3 Heerschap A, Jager GJ, van der Graaf M, Barentsz JO, Ruijs SH. Proton MR spectroscopy of the normal human prostate with an endorectal coil and a double spin-echo pulse sequence. Magn Reson Med 1997;37(2):204-213. 4 Kurhanewicz J, Vigneron DB, Hricak H, Parivar F, Nelson SJ, Shinohara K, Carroll PR. Prostate cancer: metabolic response to cryosurgery as detected with 3D H-1 MR spectroscopic imaging. Radiology 1996;200(2):489-496. 5 Kurhanewicz J, Vigneron DB, Nelson SJ, Hricak H, MacDonald JM, Konety B, Narayan P. Citrate as an in vivo marker to discriminate prostate cancer from benign prostatic hyperplasia and normal prostate peripheral zone: detection via localized proton spectroscopy. Urology 1995;45(3):459-466. 6 Glunde K, Bhujwalla ZM, Ronen SM. Choline metabolism in malignant transformation. Nat Rev Cancer;11(12): 835-848. 7 Costello LC, Franklin RB. Novel role of zinc in the regulation of prostate citrate metabolism and its implications in prostate cancer. The Prostate 1998; 35(4):285-296. 8 Giskeodegard GF, Bertilsson H, Selnaes KM, Wright AJ, Bathen TF, Viset T, Halgunset J, Angelsen A, Gribbestad IS, Tessem MB. Spermine and citrate as metabolic biomarkers for assessing prostate cancer aggressiveness. PloS one;8(4):e62375. 9 Scheenen TW, Klomp DW, Roll SA, Futterer JJ, Barentsz JO, Heerschap A. Fast acquisition-weighted three-dimensional proton MR spectroscopic imaging of the human prostate. Magn Reson Med 2004;52(1):80-88. 10 Scheenen TW, Gambarota G, Weiland E, Klomp DW, Futterer JJ, Barentsz JO, Heerschap A. Optimal timing for in vivo 1H-MR spectroscopic imaging of the human prostate at 3T. Magn Reson Med 2005;53(6):1268-1274. 11 Shukla-Dave A, Hricak H, Moskowitz C, Ishill N, Akin O, Kuroiwa K, Spector J, Kumar M, Reuter VE, Koutcher JA, Zakian KL. Detection of prostate cancer with MR spectroscopic imaging: an expanded paradigm incorporating polyamines. Radiology 2007;245(2):499-506. 12 Garcia-Martin ML, Adrados M, Ortega MP, Fernandez Gonzalez I, Lopez-Larrubia P, Viano J, Garcia-Segura JM. Quantitative (1) H MR spectroscopic imaging of the prostate gland using LCModel and a dedicated basis-set: correlation with histologic findings. Magn Reson Med; 65(2):329-339. 13 Heerschap A, Jager GJ, van der Graaf M, Barentsz JO, de la Rosette JJ, Oosterhof GO, Ruijter ET, Ruijs SH. In vivo proton MR spectroscopy reveals altered metabolite content in malignant prostate tissue. Anticancer research 1997; 17(3A): 1455-1460. 14 Jung JA, Coakley FV, Vigneron DB, Swanson MG, Qayyum A, Weinberg V, Jones KD, Carroll PR, Kurhanewicz J. Prostate depiction at endorectal MR spectroscopic imaging: investigation of a standardized evaluation system. Radiology 2004;233(3):701-708. 15 Futterer JJ, Scheenen TW, Heijmink SW, Huisman HJ, Hulsbergen-Van de Kaa CA, Witjes JA, Heerschap A, Barentsz JO. Standardized threshold approach using three-dimensional proton magnetic resonance spectroscopic imaging in prostate cancer localization of the entire prostate. Investigative radiology 2007;42(2):116-122. 16 Kobus T, Hambrock T, Hulsbergen-van de Kaa CA, Wright AJ, Barentsz JO, Heerschap A, Scheenen TW. In vivo assessment of prostate cancer aggressiveness using magnetic resonance spectroscopic imaging at 3 T with an endorectal coil. European urology;60(5):1074-1080. 17 Scheenen TWJ, Fütterer J, Weiland E and others. Discriminating cancer from noncancer tissue in the prostate by 3-dimensional proton magnetic resonance spectroscopic imaging: A prospective multicenter validation study. Invest Radiol 2011;46(1):25-33. 18 Scheenen TW, Heijmink SW, Roell SA, Hulsbergen-Van de Kaa CA, Knipscheer BC, Witjes JA, Barentsz JO, Heerschap A. Three-dimensional proton MR spectroscopy of human prostate at 3 T without endorectal coil: feasibility. Radiology 2007;245(2):507-516. 19 Kobus T, Wright AJ, Scheenen TW, Heerschap A. Mapping of prostate cancer by 1H MRSI. NMR Biomed. 2013 Jun 13. doi: 10.1002/nbm.2973. [Epub ahead of print] PMID:23761200. ratio is actually calculated and are, therefore, often not directly comparable. However, using the Siemens-supplied default protocols for acquisition and syngo.via postprocessing enables one to make use of published values as given in table 1, and incorporate 1H MRSI of the prostate into their clinical routine. Contact Tom Scheenen, Ph.D. Radboud University Nijmegen Medical Centre Radiology Department P.O. Box 9102 6500 HC Nijmegen The Netherlands [email protected] Alan Wright Arend Heerschap Thiele Kobus Tom Scheenen 22 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world Get your free copy of the PI-RADS Scoring Image Atlas Visit us at www.siemens.com/ magnetom-world Go to > Publications > Subscriptions > MRI Poster Clinical Men’s Health PI-RADS Clinical Men’s Health Classification: Structured Reporting for MRI of the Prostate PI-RADS Clinical Men’s Health Classification: Structured Reporting for MRI of the Prostate PI-RADS Classification: Structured Reporting for MRI of the Prostate M. Röthke1; D. Blondin2; H.-P. Schlemmer1; T. Franiel3 1 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany 2 Department M. Röthke1; of Diagnostic D. and Blondin2; Interventional H.-P. Schlemmer1; Radiology, University T. Franiel3 Hospital Düsseldorf, Germany 3 Department of Radiology, Charité Campus Mitte, Medical University Berlin, Germany 1 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany 2 Department M. Röthke1; of Diagnostic D. and Blondin2; Interventional H.-P. Schlemmer1; Radiology, University T. Franiel3 Hospital Düsseldorf, Germany 3 Department of Radiology, Charité Campus Mitte, Medical University Berlin, Germany 1 Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany 2 Department of Diagnostic and Interventional Radiology, University Hospital Düsseldorf, Germany 3 Department of Radiology, Charité Campus Mitte, Medical University Berlin, Germany Introduction Prostate MRI has become an increas-ingly common adjunctive procedure in a structured reporting scheme (PI-RADS) based on the BI-RADS classi-fication Introduction Prostate MRI has become an increas-ingly the detection of prostate cancer. In Germany, it is mainly used in patients with prior negative biopsies and/or abnormal or increasing PSA levels. The procedure of choice is multipara-metric based on a Likert scale with scores ranging from 1 to 5. However, it lacks illustration of the individual manifes-tations common adjunctive procedure in the detection of prostate cancer. In Germany, it is mainly used in patients with prior negative biopsies and/or abnormal or increasing PSA levels. The procedure of choice is multipara-metric MRI, a combination of high-resolution T2-weighted (T2w) mor-phological sequences and the uniform instructions for aggregated scoring of the individual submodali-ties. MRI, a combination of high-resolution multiparametric techniques of diffu-sion- classification in daily routine difficult, especially for radiologists who are less experienced in prostate MRI. It is therefore the aim of this paper to concretize the PI-RADS model for the detection of prostate cancer using representative images for the relevant scores, and to add a scoring table that combines the aggregated multipara-metric T2-weighted (T2w) mor-phological sequences and the weighted MRI (DWI), dynamic contrast-enhanced MRI (DCE-MRI), and proton MR spectroscopy (1H-MRS) [1, 2]. Previously, there were no uni-form multiparametric techniques of diffu-sion- weighted MRI (DWI), dynamic contrast-enhanced MRI (DCE-MRI), and proton MR spectroscopy (1H-MRS) [1, 2]. Previously, there were no uni-form recommendations in the form of guidelines for the implementation and standardized communication of findings. To improve the quality of the procedure and reporting, a group of experts of the European Society of Urogenital Radiology (ESUR) has recently published a guideline for MRI of the prostate [3]. In addition to pro-viding recommendations in the form This makes use of the PI-RADS of guidelines for the implementation and standardized communication of findings. To improve the quality of the procedure and reporting, a group of experts of the European Society of Urogenital Radiology (ESUR) has recently published a guideline for MRI of the prostate [3]. In addition to pro-viding recommendations relating to indications and minimum standards for MR protocols, the guideline describes according to the Likert scale. In addi-tion, reporting scheme is presented, which enables accurate communication of the findings to the urologist. Further-more, techniques are described and critically recommendations relating to indications and minimum standards for MR protocols, the guideline describes for breast imaging. This is assessed in terms of their advantages and disadvantages. a structured reporting scheme (PI-RADS) based on the BI-RADS classi-fication Materials and methods The fundamentals of technical imple-mentation for breast imaging. This is based on a Likert scale with scores ranging from 1 to 5. However, it lacks illustration of the individual manifes-tations and their criteria as well as selected by the authors by consensus on the basis of representative image findings from the 3 institutions. The scoring intervals for the aggregated PI-RADS score were also determined by consensus. The individual imaging aspects were described and evaluated with reference to current literature by one author in each case (T2w: M.R., DCE-MRI: T.F., DWI: D.B., MRS: H.S.). Furthermore, a graphic reporting scheme that allows the findings to be documented in terms of localization and classification was developed, taking into account the consensus paper on MRI of the prostate published in 2011 [4]. and their criteria as well as uniform instructions for aggregated scoring of the individual submodali-ties. This makes use of the PI-RADS classification in daily routine difficult, especially for radiologists who are less experienced in prostate MRI. It is therefore the aim of this paper to concretize the PI-RADS model for the detection of prostate cancer using representative images for the relevant scores, and to add a scoring table that combines the aggregated multipara-metric scores to a total PI-RADS score a standardized graphic prostate scores to a total PI-RADS score according to the Likert scale. In addi-tion, a standardized graphic prostate reporting scheme is presented, which enables accurate communication of the findings to the urologist. Further-more, the individual multiparametric the individual multiparametric techniques are described and critically assessed in terms of their advantages and disadvantages. were determined by con-sensus. Materials and methods The fundamentals of technical imple-mentation The sample images were were determined by con-sensus. The sample images were selected by the authors by consensus on the basis of representative image findings from the 3 institutions. The scoring intervals for the aggregated PI-RADS score were also determined by consensus. The individual imaging aspects were described and evaluated with reference to current literature by one author in each case (T2w: M.R., DCE-MRI: T.F., DWI: D.B., MRS: H.S.). Furthermore, a graphic reporting scheme that allows the findings to be documented in terms of localization and classification was developed, taking into account the consensus paper on MRI of the prostate published in 2011 [4]. 1 I: Normal PZ in T2w hyperintense II: Hypointense discrete focal lesion (wedge or band-shaped, I: Normal PZ in T2w hyperintense ill-defined) III: Changes not falling into categories 1+2 & 4+5 II: Hypointense discrete focal lesion (wedge or band-shaped, ill-defined) IV: Severely hypo-intense round-shaped, well-defined III: Changes not falling into categories 1+2 & 4+5 focal lesion, without extra-capsular extension IV: Severely hypo-intense focal lesion, round-shaped, well-defined V: Hypointense mass, round and bulging, with capsular involvement or seminal vesicle invasion without extra-capsular extension V: Hypointense mass, round and bulging, with capsular involvement or seminal vesicle invasion 1 PI-RADS classification of T2w: peripheral glandular sections. 30 MAGNETOM Flash | 4/2013 | www.siemens.com/magnetom-world 1604_MAGNETOM_Flash_54_ASTRO_Inhalt.indd 30 09.09.13 16:20 Read the comprehensive article “PI-RADS Classification: Structured Reporting for MRI of the Prostate” by Matthias Röthke et al. in MAGNETOM Flash 4/2013 page 30-38. Available for download at www.siemens.com/ magnetom-world I Choline Citrate II Choline Citrate III Choline Citrate IV Choline Citrate V Choline Citrate www.siemens.com/magnetom-world PI-RADS SCORING Image Atlas Prostate MRI Answers for life. DCE type 1 curve = 1 point I: Cho << Citrate II: Hypointense dull focal lesion (wedge or band-shaped, ill-defined) III: Changes not falling into categories 1+2 or 4+5 IV: Hypointense focal lesion without extracapsular extension or bulging V: Hypointense mass with extracapsular extension or bulging I: Normal, hyperintense PZ in T2w, peripheral glandular lesions II: Hypointense lesion with well-defined capsule; band-shaped hypointense regions III: Changes not falling into categories 1+2 or 4+5 IV: Hypointense lesion without capsular involvement with ill-defined margins, “erased charcoal sign” V: Oval-shaped mass with capsular involvement; infiltrating mass with invasion into anterior structures I: Stromal & glandular hyperplasia without focal hypointense lesions T2w, central glandular lesions II: Diffuse hyperintensity on DWI b ≥ 800 image, no focal ADC reduction III: Changes not falling into categories 1+2 or 4+5 IV: Focal area with reduced ADC but isointense SI on DWI b ≥ 800 image V: Focal area with reduced ADC and hyperintense SI on DWI b ≥ 800 image I: No reduction in ADC and no increase in SI on DWI b ≥ 800 images 2 points = probably benign 3 points = indeterminate 4 points = probably malignant 5 points = highly suspicious of 1 point = most probably benign malignancy DCE type 2 curve = 2 points DCE type 3 curve = 3 points DCE-MRI – asymmetric, non-focal: + 1 point DCE-MRI – asymmetric, unusual location: + 2 points DCE-MRI – asymmetric, focal location: + 2 points DCE-MRI – symmetric, non-focal: + 0 points II: Cho << Citrate III: Cho = Citrate IV: Cho > Citrate V: Cho >> Citrate For details please refer to: M. Röthke, D. Blondin, H.-P. Schlemmer, T. Franiel: “PI-RADS Classification: Structured Reporting for MRI of the Prostate”, MAGNETOM Flash issue 4/2013, ASTRO edition, page 30-38. T2w, peripheral glandular lesions T2w, central glandular lesions DWI b ≥ 800 ADC DCE time curve / parametric color map 1H-MRS Introduction Prostate MRI has become an increas-ingly common adjunctive procedure in the detection of prostate cancer. In Germany, it is mainly used in patients with prior negative biopsies and/or abnormal or increasing PSA levels. The procedure of choice is multipara-metric MRI, a combination of high-resolution T2-weighted (T2w) mor-phological sequences and the multiparametric techniques of diffu-sion- weighted MRI (DWI), dynamic contrast-enhanced MRI (DCE-MRI), and proton MR spectroscopy (1H-MRS) [1, 2]. Previously, there were no uni-form recommendations in the form of guidelines for the implementation and standardized communication of findings. To improve the quality of the procedure and reporting, a group of experts of the European Society of Urogenital Radiology (ESUR) has recently published a guideline for MRI of the prostate [3]. In addition to pro-viding recommendations relating to indications and minimum standards for MR protocols, the guideline describes a structured reporting scheme (PI-RADS) based on the BI-RADS classi-fication for breast imaging. This is based on a Likert scale with scores ranging from 1 to 5. However, it lacks illustration of the individual manifes-tations and their criteria as well as uniform instructions for aggregated scoring of the individual submodali-ties. This makes use of the PI-RADS classification in daily routine difficult, especially for radiologists who are less experienced in prostate MRI. It is therefore the aim of this paper to concretize the PI-RADS model for the detection of prostate cancer using representative images for the relevant scores, and to add a scoring table that combines the aggregated multipara-metric scores to a total PI-RADS score according to the Likert scale. In addi-tion, a standardized graphic prostate reporting scheme is presented, which enables accurate communication of the findings to the urologist. Further-more, the individual multiparametric techniques are described and critically assessed in terms of their advantages and disadvantages. Materials and methods The fundamentals of technical imple-mentation were determined by con-sensus. The sample images were selected by the authors by consensus on the basis of representative image findings from the 3 institutions. The scoring intervals for the aggregated PI-RADS score were also determined by consensus. The individual imaging aspects were described and evaluated with reference to current literature by one author in each case (T2w: M.R., DCE-MRI: T.F., DWI: D.B., MRS: H.S.). Furthermore, a graphic reporting scheme that allows the findings to be documented in terms of localization and classification was developed, taking into account the consensus paper on MRI of the prostate published in 2011 [4]. I: Normal PZ in T2w hyperintense II: Hypointense discrete focal lesion (wedge or band-shaped, ill-defined) III: Changes not falling into categories 1+2 & 4+5 IV: Severely hypo-intense focal lesion, round-shaped, well-defined without extra-capsular extension V: Hypointense mass, round and bulging, with capsular involvement or seminal vesicle invasion 1 1 PI-RADS classification of T2w: peripheral glandular sections. 30 MAGNETOM Flash | 4/2013 | www.siemens.com/magnetom-world 1604_MAGNETOM_Flash_54_ASTRO_Inhalt.indd 30 09.09.13 16:20 1 1 PI-RADS classification of T2w: peripheral glandular sections. 30 MAGNETOM Flash | 4/2013 | www.siemens.com/magnetom-world 1604_MAGNETOM_Flash_54_ASTRO_Inhalt.indd 30 09.09.13 16:20
- 13. Evaluation of theCIVCO Indexed Patient Position System (IPPS) MRI-Overlay for Positioning and Immobilization of Radiotherapy Patients Th. Koch1; K. Freundl1; M. Lenhart2; G. Klautke1; H.-J. Thiel1 1 Klinik und Praxis für Strahlentherapie und Radioonkologie, Sozialstiftung Bamberg, Germany 2 Klinik für Diagnostische Radiologie, Interventionelle Radiologie und Neuroradiologie, Bamberg, Germany Abstract The emerging development in modern radiotherapy planning (RTP) requires sophisticated imaging modalities. RTP for high precision requires exact delineation of the tumor, but this is currently the weakest link in the whole RTP process [1]. Therefore Magnetic resonance imaging (MRI) is of increasing interest in radiotherapy treatment planning because it has a superior soft tissue contrast, making it possible to define tumors and surrounding healthy organs with greater accuracy. The way to use MRI in radiotherapy can be different. The MRI datasets can be used as secondary images to support the tumor delineation. This is routinely in use in many radiotherapy departments. Two other methods of MRI guidance in the RTP process are until now only research 3 One index bar is latched to the accuracy strongly depends on the MRI scan position. Hanvey et al. [3] and Brunt et al. [4] have shown that it is indispensable for the MRI dataset to be created in the treatment position which is primarily defined by the CT scan. The MRI dataset can also feasibily be used as the only dataset. Because of the lack of electron density information, which is required for dosimetric calculations, bulk densities have to be applied to the MRI images. For this purpose the different anatomic regions like bone, lung, air cavities and soft tissue have to be overwritten with the physical densities. With this method it is possible to achieve dose calculation results quite similar to the calculation in the CT dataset in the head and neck region [5, 6] as well as in the pelvic region [7]. The advantage of this method is that by avoiding the CT scan you save some time and money. In this case it is necessary for the treatment position to be determined during the MRI scan, hence the MRI scanner has to be equipped with the same positioning and immobilization tools as the Linac. Further problems to overcome are the evaluation and correction of possible image distortions and the determination of accurate bulk densities. After the RTP process there are a lot of remaining uncertainties such as set-up errors, motion of the target structures and during the treatment changes of the tumor volume and shrinking. This problem can be overcome with the so-called image-guided radiotherapy (IGRT). IGRT involves a periodical verification (weekly or more frequent) of tumor position and size with appropriate imaging systems. It is evident that IGRT is only as good as the accuracy with which the target structures can be defined. For this reason some groups try to develop hybrid systems, where a Linac or a cobalt treatment unit is combined with an MRI scanner for a so-called MR-guided radiotherapy [8-10]. Again: MR-guided radiotherapy can only be successful when the reference MRI dataset has been created in the treatment position. In any of the above three cases, where MRI can be helpful to improve the accuracy of radiotherapy, it is strongly advised that one has a robust and reproducible patient positioning and immobilization system, mainly at the MRI scanner, which is used for MR-guided RTP. Siemens provides with the CIVCO IPPS MRI-Overlay a suitable solution. In our clinic we have introduced and tested this MRI- overlay, especially for patients with tumors in the pelvis and for brain tumors and metastasis. Method Our 1.5T MAGNETOM Aera system (Siemens Healthcare, Erlangen, Germany) is located in the radiology department and can temporarily be used by the staff of the radiotherapy department. For the purpose of MR-guided RTP we have equipped the MAGNETOM Aera with the CIVCO IPPS MRI-Overlay. This overlay enables the fixation of positioning and immobilization tools necessary for radioprojects, but interest in them is increasing. The first method is to use MRI data as the primary and only image dataset and the second is the application of the MRI data as reference dataset for a so-called ’MRI-guided radiotherapy in hybrid systems’ (Linear Accelerator (Linac) or Cobalt RT units combined with MRI). For all cases it is essential to create the MRI datasets in the radiotherapy treatment position. For this reason the CIVCO Indexed Patient Positioning System (IPPS) MRI-Overlay was introduced and tested with our Siemens MAGNETOM Aera MRI Scanner. Introduction Although computed tomography (CT) images are the current gold standard in radiotherapy planning, MRI becomes more and more interesting. Whilst CT has limitations in accuracy concerning the visualization of boundaries between tumor and surrounding healthy organs, MRI can overcome these problems by yielding superior soft tissue contrast. Currently there are three different possible strategies by which MRI can help to improve radiotherapy treatment planning: The MRI datasets can be used as secondary images for treatment planning. These MR images can be used to delineate the tumor and the surrounding organs, whilst the CT images – the primary planning data – are necessary to calculate the 3D dose distribution. The two image datasets have to be co-registered thoroughly to ensure that the anatomy correlates (see for example [2]). The registration therapy treatments. For our purpose we have used an MR compatible mask system for head and neck cases and vacuum cushions for patients with diseases in the pelvic region both from Medical Intelligence (Elekta, Schwabmünchen, Germany). These tools can all be fixed with so-called index bars (Figs. 4, 12) at the MRI- Overlay. These index bars are custom designed for our purpose by Innovative Technologies Völp (IT-V, Innsbruck, Austria) for the MRI-Overlay and for use in the high field magnetic environment. For the correct positioning of the patients, the laser system Dorado 3 (LAP, Lüneburg, Germany) was additionally installed in the MRI room. The preliminary modifications and the patient positioning is described in the following for two cases. The first case describes the procedure for a patient with a head tumor. The first step is the removal of the standard cushion of the MRI couch and the mounting of the MRI-Overlay (Figs. 1–3). One index bar is necessary to fix the mask system on the overlay (Figs. 4, 5) to avoid movements and rotations during the scan. Because the standard head coil set cannot be used with the mask system, two flex coils (Flex4 Large) have to be prepared (Figs. 6–8). In figure 8 one can see, that the correct head angle could be adjusted. Now the patient is placed on the overlay and in the mask system. The patient’s head can be immobilized with the real and proper mask made from thermoplastic material called iCAST (Medical Intelligence, Elekta, Schwabmünchen, Germany) 1 After the removal of the standard cushion the CIVCO 1.5T MAGNETOM Aera with the standard cushion on the MRI couch. IPPS MRI-Overlay can be mounted. 2 1 2 3 The lines indicate the position for the index bars. MRI-Overlay. 4 4 The mask system for head and neck fits to the index bar to avoid movement. 5 5 Radiation Oncology Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 25 Clinical Radiation Oncology 24 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 14. Clinical Radiation OncologyRadiation Oncology Clinical 6 7 8 10 11 as can be seen in figure 9. Now the flex coils can be fixed with hook-and-loop tapes and placed very tight to the patient (Figs. 10, 11). Now the MRI scan can be started. The second case describes the preparation before the MRI scan for a patient with a tumor in the pelvic region. The first two steps are identical, the remove of the standard cushion followed by the mount of the overlay (Figs.1, 2). Then a custom-made vacuum cushion for the lower extremities is attached to the overlay with two index bars (Figs. 12, 13). For a robust position of the patients with diseases in the pelvis it is very important to keep the legs in well-defined position – not only during imaging but also throughout the whole treatment course, which spans over seven weeks. Any changes there can result in undesired rotations of the pelvis and in the end the tumor position and shape can also change. In figure 13 a second custom-made vacuum cushion can be seen. The only purpose of this vacuum cushion is to enable a comfortable position of the patient during scan and later during the treatment (Fig. 14). The more comfortably the patient lies on the table the more robust and reproducible is the positioning. Fortunately MAGNETOM Aera has a bore diameter of 70 cm, hence there are almost no limitations concerning patient positioning. Now the accurate position of the patient should be checked with the moveable laser-system (Fig. 15). This is necessary to avoid rotations of the pelvis around the patients longitudinal and lateral axis. For the fixation of the flex-coil for the pelvic region a mounting- frame has to be attached to the overlay (Figs. 16, 17). This can be done with hook-and-loop tapes (Fig. 18). Now the patient set-up is completed and the MRI scan can be started (Fig. 19). 14 15 16 17 19 Results Two examples are shown in the following pictures. In Fig. 20 you can see a brain tumor in two corresponding slices. The left picture shows the CT-slice and the right picture shows the corresponding MRI slice obtained with a T1-weighted sequence with contrast agent. It is clear to see that tumor boundary is much more pronounced in the MRI image. Figure 21 shows the same slices with structures created by the radiotherapists. It is also helpful to create some control structures, such as brain and ventricles, to check the accuracy of the registration. Figures 22 and 23 give an example of a patient with prostate cancer. In this case the MRI images on the right 6 7 A custom-made vacuum cushion for the lower extremities is latched to the MRI-Overlay with two index bars. 12 A second vacuum cushion is positioned on the table to fix the arms and shoulders and keep the patient in a comfortable position. 13 Now the patient can be positioned. The accurate position of the patient can be adjusted with the LAP laser system. A mounting-frame for the flex coil has to attached to the MRI-Overlay. The mounting-frame from a side view. The flex coil is fixed to the mounting-frame with hook-and- loop tapes. 18 The patient is ready to start the scan. 12 13 14 15 16 17 18 19 Two flex coils (Flex4 Large) are prepared. The flex coils have to be positioned partly under the mask system, because the whole head of the patient should be covered. It is possible to adjust the head angle in an appropriate and reproducible position that is comfortable for the patient. Now the patient is immobilized using a custom-made mask made from thermoplastic material. 9 The flex coils are closed with hook-and-loop tapes. The patient is ready for the scan. 8 9 10 11 26 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 27
- 15. Clinical Radiation OncologyRadiation Oncology Clinical Contact Thomas Koch, Ph.D. Sozialstiftung Bamberg – Medizinisches Versorgungszentrum am Bruderwald Praxis für Radioonkologie und Strahlentherapie Head Medical Physics Buger Straße 80 96049 Bamberg Germany Phone: +49 951 503 12931 [email protected] are acquired using a T2-weighted TrueFISP sequence. The boundary of the prostate and the differentiation between prostate and rectum is much more easier to define in the MRI images. The control structures in this case are the femoral heads. For the head scans we normally use 3 sequences, a T1w SE with contrast agent, a T2w TSE and a FLAIR sequence. For the pelvis scans we normally use a T2w SPACE, a T2w TrueFISP and a T2w TSE sequence. The coordinate system should be the same for all sequences, that means same slices and same field-of-view. Hence one can use the same registration parameters for all sequences. References 1 Njeh C. F. Tumor delineation: the weakest link in the search for accuracy in radiotherapy. J. Med. Phys. 2008 Oct-Dec; 33(4): 136-140. 2 Dean C.J. et al. An evaluation of four CT-MRI co-registration techniques for radiotherapy treatment planning of prone rectal cancer patients. Br. J. Radiol. 2012 Jan; 85: 61-68. 3 Hanvey S. et al. The influence of MRI scan position on image registration accuracy, target delineation and calculated dose in prostatic radiotherapy. Br. J. Radiol. 2012 Dec; 85: 1256-1262. Conclusion and outlook We can now look back over a period of two years working with the CIVKO IPPS MRI-Overlay. Our experience is very promising. The modifications on the table of the MRI scanner are very easy and can be executed and finished in only a couple of minutes. The procedure is well accepted by the radiologic technologists. To date, we have scanned more than 100 radiotherapy patients, mainly with diseases in the pelvis (rectum and prostate cancer) and in the head (brain tumors and metastasis). So far we have only used MRI dataset as a secondary image dataset. The co-registration with the CT datasets is now much easier because we have nearly identical transversal slices in both image datasets. As a conclusion we can say that we are very happy with the options we have to create MRI scans in the treatment positions. It has been demonstrated that the MRI dataset is now much more helpful in the radiotherapy planning process. We should mention the need for a quality assurance program to take possible image distortions into consideration. Our next step is to install such a program, which involves the testing of suitable phantoms. A further step will be to assess whether we can use MRI datasets alone for RTP. 4 Brunt J.N.H. Computed Tomography – Magnetic Resonance Imaging Registration in Radiotherapy Treatment Planning. Clin. Oncol. 2010 Oct; 22: 688-697. 5 Beavis A.W. et al. Radiotherapy treatment planning of brain tumours using MRI alone. Br. J. Radiol. 1998 May; 71: 544-548. 6 Prabhakar R. et al. Feasibility of using MRI alone for Radiation Treatment Planning in Brain Tumors. Jpn. J. Clin. Oncol. 2007 Jul; 37(6): 405-411. 7 Lambert J. et al. MRI-guided prostate radiation therapy planning: Investigation of dosimetric accuracy of MRI-based dose planning. Radiother. Oncol. 2011 Mar 98: 330-334. 8 Raymakers B.W. et al. Integrating a 1.5 T MRI scanner with a 6 MV accelerator: proof of concept. Phys. Med. Biol. 2009 May; 54: 229-237. 9 Hu Y. et al. Initial Experience with the ViewRay System – Quality Assurance Testing of the Imaging Component. Med. Phys. 2012 Jun; 39:4013. 10 ViewRay. Available at: http://www.viewray.com 20 Two corresponding slices of a brain scan: (20A) CT slice and (20B) MRI slice obtained using a T1-weighted sequence with contrast agent. 20A 20B 21 The same slices as in figure 20, but now with delineated tumor and help structures. 21A 21B 22 Two corresponding slices in the pelvic region of a patient with a prostate cancer: (22A) CT slice and (22B) MRI slice obtained with a T2-weighted TrueFISP sequence. 22A 22B 23 The important structures rectum and prostate as defined in the MRI slice are shown. The accuracy of the registration can be tested with the coincidence of help structures – like the femoral heads in this case – in both datasets. 23A 23B 28 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 29
- 16. Technology Technology 2A 2D 2B reference TSE qTSE qTSE-G 2E 2C 2F 2 Representative slices from 2 patients acquired with the reference TSE sequence (left column), the qTSE sequence (center column) and the qTSE-G sequence (right column). Making MRI Scanning Quieter: Optimized TSE Sequences with Parallel Imaging Eric Y. Pierre1; David Grodzki2; Bjoern Heismann2; Gunhild Aandal3, 5; Vikas Gulani1, 3; Jeffrey Sunshine3; Mark Schluchter4; Kecheng Liu6; Mark A. Griswold1, 3 1 Biomedical Engineering, Case Western Reserve University, Cleveland, Ohio, USA 2 Siemens Healthcare, Erlangen, Germany 3 Radiology, Case Western Reserve University, Cleveland, Ohio, USA 4 Division of Biostatistics, Case Western Reserve Univeristy, Cleveland, Ohio, USA 5 Haraldsplass Deaconess Hospital, Bergen, Norway 6 Siemens Medical Solutions, USA Inc., Malvern, Pennsylvania, USA Introduction Turbo Spin-Echo sequences at 1.5T can generate noise at over 100dBA inside the bore [1–3]. This noise is equivalent to standing 5 meters away from a jackhammer [3], and would be even louder on higher field systems. Despite the use of ear-protective equipment, reducing the Sound Pressure Level (SPL) generated by these standard clinical sequences could noticeably improve patient comfort [4]. MRI pulse sequences mostly generate acoustic noise because of rapidly varying gradient waveforms: The resulting Lorentz forces applied on the gradient coils make the entire scanner structure vibrate [5]. To circumvent this issue, several hardware solutions have been proposed. For example, the whole gradient coil can be enclosed in a vacuum chamber [6–8], or gradient field rotation can be performed mechanically [9]. While these solutions achieve significant noise reduction for all types of sequences, they can noticeably increase manufacturing cost, and can even decrease gradient efficiency. Mechanical and acoustic balanced designs of gradient coil systems including windings performing active acoustic control have also been considered and investigated [10, 11]. of acoustic noise in Echo Planar Imaging (EPI) [14, 15]. The reduction was achieved by counterbalancing lengthened gradient waveforms with increased acquisition speed, thereby reducing acoustic noise without increasing acquisition time while maintaining inter-echo spacing, only at cost of signal-to-noise ratio (SNR). By extending such principles to other generally-used standard clinical MR sequences, this article demonstrates that with minor SNR reductions (≤ 10%), effective reduction in acoustic noise can be further achieved without noticeable degrade of diagnostic information or imaging time, as well as without sacrificing gradient efficiency. Two types of modifications in a T2-weighted Turbo Spin-Echo (TSE) sequence were investigated for acoustic noise reduction: First by solely modifying the gradient waveforms and second by additionally using GRAPPA at a reduction factor of two (R=2)*. Comparative SPL measurements at the bore were performed between standard TSE, quiet TSE (qTSE)* and quiet TSE with GRAPPA (qTSE-G)*. A statistical analysis of comparative scores from a reader’s study was conducted. Methods The gradient waveforms of the TSE sequence were optimized with an automatic gradient optimization algorithm that extends any slope duration to its maximum and reduces the number of slopes to their minimum. For instance, with minor changes in protocols, spoiling and crusher gradient lobes are replaced by long rising or descending slopes, while maintaining the crusher moment unchanged. To keep the same total acquisition time, the reduction of the gradient slew rate is constrained by the fixed inter-echo spacing. The decreased slew rate of readout gradient will slightly reduce readout sampling time (Fig. 1). In consequence, the readout bandwidth (BW) increases slightly, with a tradeoff between reduction of SPL and SNR loss. In addition, parallel acquisition could be further employed to reduce the echo-train length, i.e. number of echoes per train, by a factor of R. Keeping the acquisition time constant, the inter-echo spacing can be extended by R, allowing further stretching of the gradient moments. This effectively represents a benefit of parallel imaging acceleration in acoustic noise reduction rather than imaging time reduction. The acquisition protocols changes are as follows: The readout BW was increased by about 10%, from 107 Hz/pixel in the standard protocol to 125 Hz/pixel. The effective TR/TE were increased from 5000/93 ms to 5180/85 ms, which resulted in only a 3 second increase in acquisition time, from 1:37 min to 1:40 min. The qTSE-G parameters were identical to the qTSE protocol, but with use of GRAPPA with R=2. For both qTSE and qTSE-G protocols, and the gradient slopes were maximally stretched as illustrated in figure 1. Modifying and/or optimizing pulse sequences can also reduce acoustic noise effectively. One such solution is to time the ramping up and ramping down of the gradient waveforms so that the induced scanner vibrations cancel each other out [12]. Another approach is to use lower gradient amplitude and slew rates of the gradient waveforms [13]. By low-pass filtering the gradient, vibration frequencies for which the acoustic response of the gradient coil is high can be avoided. Elaborate redesigns of gradient waveforms coupled with parallel imaging have demonstrated further reduction 1 Comparison of conventional sequence (dashed lines) with quiet sequence (solid lines). The reduction of ADC represents a BW increase. 1 ADC RF / ADC GSlice GRead * WIP, the product is currently under development and is not for sale in the US and other countries. Its future availability cannot be ensured. 30 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 31
- 17. Technology Technology Table1: Comparison of dBA values Sequence type Standard TSE qTSE qTSE-G Background LAEQ (30 sec average) 92.5 81.3 72.7 53.0 Max Peak 102.8 95.8 92.0 77.7 Comparison of dBA values for standard TSE, qTSE, qTSE-G sequences, and measured background noise. Measurements were performed inside the bore at patient head position using a 2238 Mediator sound level meter (Brueel & Kjaer GmbH, Bremen, Germany). Table 2: Ratings by readers Sequence type All techniques compared to themselves Reader #1 0.35 ± 0.40 (0.06, 0.64) p = 0.02 Reader #2 -0.03 ± 0.11 (-0.11, 0.04) p = 0.34 Reader #3 Average 0.11 ± 0.14 (0.01, 0.21) p = 0.04 Mean and standard deviation, 95% confidence interval, and p-value of the scores given by each radiologist to the different types of image volume pairs after self-bias correction. Positive score show preference of the right volume over the left volume, on a -10 to +10 scale. In-vivo studies were performed on a 3T MAGNETOM Verio MRI scanner (Siemens Healthcare, Erlangen, Germany) with a 12-channel head coil with patients admitted for head examination. Informed consent was obtained from the volunteer before the start of the study in accordance with IRB protocol. A total of 10 different patient scannings were performed, each comparing standard TSE images with qTSE and qTSE-G images. The image resolution (192 × 256 matrix), number of slices (26), slice thickness (5 mm) and slice orientation were kept identical throughout the 3 different acquisitions. 0 ± 0 – To measure acoustic noise level LAEQ (Equivalent Continuous Sound Level in A-weighting) with 30 seconds average and peak values, a professional device, 2238 Mediator sound level meter (Brueel & Kjaer GmbH, Bremen, Germany), was used, which was placed inside the bore at patient head position. qTSE : TSE qTSE-G : TSE -0.20 ± 0.26 (-0.38, -0.02) p = 0.04 1.30 ± 1.96 (-0.10, 2.70) p = 0.07 0.73 ± 1.59 (-0.41, 1.86) p = 0.18 0.61 ± 1.17 (-0.23, 1.45) p = 0.13 The background noise is mainly generated by the cold-head pump and the ventilation among other sources. To evaluate the image quality, a total of 7 image-volume-pairs were assembled from each of the 10 patient datasets. The first 2 pairs compared qTSE with TSE volumes, alternatively with qTSE on the left and TSE on the right. Similarly, another 2 pairs compared qTSE-G with TSE volumes in both left-right orders randomly. Finally, 3 pairs were assembled with the same volume on the left and right, which consist of TSE vs. TSE, qTSE vs. TSE, and qTSE-G vs. qTSE-G volumes, respectively. All 70 volume pairs were presented in the same random order to 3 trained radiologists blinded to the acquisition technique, who were asked the following question: “On a scale from -10 to +10, how much better is the image quality of the volume on the right compared to the volume on the left, 0.20 ± 0.59 (-0.22, 0.62) p = 0.31 3.95 ± 0.86 (3.33, 4.57) p < 0.0001 3.08 ± 1.25 (2.18, 3.97) p < 0.0001 2.41 ± 0.80 (1.83, 2.98) p < 0.0001 with a positive score indicating superiority of the right volume, and 0 representing no difference in quality between left and right?”. The graphical user interface used for the reading allowed user-navigation through the paired-volume slices, and simultaneous image windowing of the 2 displayed images. To avoid possible left-right bias, the average of the qTSE vs. TSE score and the TSE vs. qTSE score multiplied by -1 was then calculated for each reader’s reading on each patient. The average of the corrected scores across readers was then computed for each patient. Corrected scores were calculated in the same way for the qTSE-G vs. TSE comparison. One-sample t-tests were used to test whether the mean average reader scores differed from zero, and 95% confidence intervals (CI) for the mean scores were also calculated. One-sample t-tests and CI were also carried out using each reader’s scores 10 8 6 4 2 0 -2 -4 -6 -8 -10 preference for standard A B C preference for quiet 3 separately. A reader’s average rating of these three self-comparisons using images from each patient were averaged, and then the three reader averages were averaged for each patient. A t-test was used to test whether the average of the reader ratings across patients differed from zero. One-sample t-tests and CI were also carried out for each reader separately. Results The respective average and peak SPL in [dBA] measurements for standard TSE, qTSE and qTSE-G protocols are listed in table 1. The achieved reduction of average SPL for qTSE and qTSE-G were 10 dBA and near 20 dBA (30 seconds average), respectively. Discussion Optimizing the gradient waveforms alone with a 10% increase in bandwidth achieves an 11 dBA SPL reduction (Table 1), with little cost to image quality (Fig. 3). These results are in accordance with [16] though here the measurements were made directly at the bore. This cost might be more noticeable with lower SNR systems, however in this configuration, no statistically significant difference in image quality was observed (Table 2), making gradient redesign a viable solution to make TSE sequences quieter. With additional use of Parallel Imaging, the modified quiet TSE sequence allows on average a 20 dBA reduction in SPL (Table 1). The modified sequence had an effect on in image quality A. self image comparison (all methods) (p < 0.05) B. qTSE vs. standard TSE Non significant difference for all 3 readers (p = 0.20, p = 0.06, p = 0.18) C. qTSE+GRAPPA vs. standard TSE (p < 0.0001) (Fig. 3): The average preference score across readers for standard TSE images over qTSE-G images was +2.41 (p<0.0001, Table 2), and the 95% confidence interval places its true value between +1.8 and +3. However it should be noted that this change in image quality is to be expected as Parallel Imaging was used. In compensation, the reduction of acoustic noise was highly effective: the SPL at the bore of the standard TSE sequence was 39.5 dBA higher than the background noise, compared to 19.7 dBA for the modified sequence. Conclusion In comparison with standard MR sequences, gradient wave modifications in TSE sequence coupled with Parallel Imaging can achieve over a factor 10 of acoustic noise reduction, yielding an improved patient comfort with nearly identical diagnostic information and imaging time. Without any hardware modifications or upgrade, both proposals described in this article, qTSE and qTSE-G, can be easily implemented on a conventional MRI system for routine clinical applications. In addition, scanning on a high field system with multiple channel coils, such as the 32-channel head coil, provides more flexibility to make MRI scanning quieter. * Work in progress: The product is still under development and not commercially available yet. Its future availability cannot be ensured. 3 (A) 95% confidence intervals for averages scores by readers for volumes compared to themselves; (B) qTSE vs. standard TSE; and (C) qTSE-G vs. standard TSE. Positive scores show preference for standard TSE in the last two cases. References 1 Shellock FG, Morisoli SM, Ziarati M. Measurement of acoustic noise during MR imaging: evaluation of six “worst-case” pulse sequences. Radiology 1994;191:91–93. 2 McJury M, Blug A, Joerger C, Condon B, Wyper D. Acoustic noise levels during magnetic resonance imaging scanning at 1.5 T. Br J Radiol 1994;67:413–415. 3 McJury M. Acoustic noise levels generated during high field MR imaging. Clin Radiol 1995;50:331–334. 4 Quirk ME, Letendre AJ, Ciottone RA, Lingley JF. Anxiety in patients undergoing MR imaging. Radiology 1989;170:463–466. 5 Hedeen R, Edelstein W. Characterization and prediction of gradient acoustic noise in MR imagers. Magn Reson Med 2005;37:7–10. 6 Katsunuma A, Takamori H, Sakakura Y, Hamamura Y, Ogo Y, Katayama R. Quiet MRI with novel acoustic noise reduction. MAGMA 2002;13:139–44. 7 Edelstein WA, Hedeen RA, Mallozzi RP, El-Hamamsy SA, Ackermann RA, Havens TJ. Making MRI quieter. Magn Reson Imaging 2002;20:155–63. 8 Edelstein WA, Kidane TK, Taracila V, Baig TN, Eagan TP, Cheng Y-CN, Brown RW, Mallick J a. Active-passive gradient shielding for MRI acoustic noise reduction. Magn Reson Med 2005;53:1013–7. 9 Cho ZH, Chung ST, Chung JY, Park SH, Kim JS, Moon CH, Hong IK. A new silent magnetic resonance imaging using a rotating DC gradient. Magn Reson Med 1998;39:317–21. 10 Mansfield P, Haywood B. Principles of active acoustic control in gradient coil design. MAGMA 2000;10:147–51. 11 Haywood B, Chapman B, Mansfield P. Model gradient coil employing active acoustic control for MRI. MAGMA 2007;20:223–31. 32 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 33
- 18. Technology Quiet T1-weighted3D Imaging of the Central Nervous System Using PETRA Masahiro Ida, M.D.1; Matthew Nielsen, M.A.2 1 Dept. of Radiology, Tokyo Metropolitan Ebara Hospital, Tokyo, Japan 2 Research & Collaboration Dept., Healthcare Sector, Siemens Japan K.K., Tokyo, Japan Introduction Nearly all MRI sequences in routine clinical use employ rapidly varying magnetic field gradients that generate considerable acoustic noise, one of the primary causes of patient discomfort and restlessness [1]. Eliminating such noise would provide additional comfort for all patients, and may provide particular advantages for patients with pediatric*, dementia and certain psychiatric diseases who tend to have difficulty relaxing or remaining still during MR examinations. Ultra-short echo time sequences such as zero-TE [2], SWIFT [3] and PETRA [4] require only limited gradient activity and allow for inaudible 3D scanning. However, due to ultra-short TEs, the image contrast is given by the steady state and is limited to the range of PD- to T1-weighting unless pre-pulses are used [5]. Similar to the MPRAGE sequence [6-8], stronger T1-weighting can be generated by applying an inversion pre-pulse before every nth repetition in the PETRA* sequence. A study has shown that this quiet inversion-prepared PETRA sequence is capable of T1-weighting com-parable to that of MPRAGE when measured in the same time and with the same spatial resolution [1]. In this article, examples of quiet inversion-prepared PETRA images are compared with conventional 3D T1-weighted images (MPRAGE or 3D-FLASH) from the same patients. All of the examples were obtained during brain examinations, and all but one of the examples employed contrast enhancement. * WIP, the product is currently under development and is not for sale in the US and other countries. Its future availability cannot be ensured. PETRA sequence principles and noise reduction In the PETRA sequence, gradients are already on and stable at a certain amplitude before the excitation pulse, as shown in figure 1. At the end of each repetition, the gradient strength on each axis is altered only slightly meaning that the required slew rate is extremely low (e.g., < 5 T/m/s with 1A 1B PETRA combines two different sequences, acquiring central k-space in a ‘point-wise’ fashion (one k-space point per repetition), and the rest of k-space with radial trajectories. PETRA stands for Pointwise Encoding Time reduction with Radial Acquisition. No hardware modifications or dedicated coils are needed [1]. (1A) Pulse sequence diagram for one repetition of the radial part of the PETRA sequence. Gradients are held constant during almost an entire repetition and altered only slightly at the end of each repetition without being ramped down. This leads to negligible deformation and vibration of the gradient coil. Thus, no acoustic noise is generated by the gradient coil. THW is the time required to switch from transmission mode to receive mode (in the range of 10 to 100 μs on clinical scanners) [1]. (1B) During THW, a spherical volume (dots) at the center of k-space is missed by the radial part of the sequence. Each k-space point inside that spherical volume is acquired separately in the Pointwise Encoding (PE) part of the sequence. The acquisition time of the PE part is approximately 3 to 5% of the total measurement time [1]. 1 TX THW RX Tx/Rx Gradients TR Technology MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 35 12 Shou X, Chen X, Derakhshan J, Eagan T, Baig T, Shvartsman S, Duerk J, Brown R. The suppression of selected acoustic frequencies in MRI. Appl Acoust 2010;71:191–200. 13 Hennel F, Girard F, Loenneker T. “Silent” MRI with soft gradient pulses. Magn Reson Med 1999;42:6–10. 14 De Zwart J, Vangelderen P, Kellman P, Duyn J. Reduction of Gradient Acoustic Noise in MRI Using SENSE-EPI. Neuroimage 2002;16:1151–1155. 15 Witzel T, Wald LL. Methods for Functional Brain Imaging. Massachusetts Institute of Technology; 2011. Contact Eric Y. Pierre, Ph.D. Department of Biomedical Engineering Case Western Reserve University 319 Wickenden Building 10900 Euclid Avenue Cleveland, OH 44106-7207 USA Phone: +1 (216)-368-4063 [email protected] 16 Pierre EY, Grodzki D, Heismann B, Liu K, Griswold MA. Reduction of Acoustic Noise to Improve Patient Comfort Through Optimized Sequence Design. In: Proceedings of the 21st Annual Meeting of ISRMRM. Vol. 42. Salt Lake City, USA; 2013. p. 256. Eric Pierre Gunhild Aandal Vikas Gulani Jeffrey Sunshine Mark Schluchter Kecheng Liu Mark Griswold Answers for life. www.siemens.com/quiet-suite Quiet Suite Imaging is to be seen, not heard. Scan and listen to Quiet-Suite.
- 19. PETRA) [1]. Theresulting deformation and vibration of the gradient coil is negligible and produces almost no audible sound. Completely unrelated to the gradients however, transmit-mode- to-receive-mode switching (and vice versa) in receive-only RF coils produces some noise [1], while PETRA is essentially inaudible when used with transmit-and-receive RF coils. The acoustic noise levels generated by PETRA and MPRAGE on a MAGNETOM Trio A Tim System (3T) were measured using a sound-pressure meter with A-weighting. PETRA afforded a reduction in acoustic noise of more than 25 dBA with both the 12-channel head matrix coil and the 32-channel head coil. Since both coils were receive-only, an even greater reduction can be expected with transmit-and-receive coils. PETRA versus routine sequence image comparisons While MR angiography (MRA) is undoubtedly the most commonly-used 3D sequence in brain MRI exams, other 3D sequences are used in certain circumstances at Tokyo Metropolitan Ebara Hospital. The most common one is contrast-enhanced (CE) 3D-FLASH which is employed for the following indications because of its high spatial resolution and short echo time: (1) To precisely diagnose head & neck tumors, pre- & post-operatively (with Quick FatSat), (2) to inspect blood pools (AVM, thrombosis, aneurysm, dissection), and (3) to detect cranial nerve inflammation. The second most common 3D scan other than MRA is CE MPRAGE which is employed to diagnose intracranial brain tumors. Among those, the most common indication is screening for intracranial metastases. MR imaging was performed on a 3 Tesla MAGNETOM Trio A Tim System. PETRA was added to routine patient exams that included either MPRAGE or 3D-FLASH. The parameters of the three sequences are shown in table 1. The center of k-space for MPRAGE (scan time: 5 min 56 sec) was acquired approximately 6 to 11 minutes after contrast media administration, while the center of k-space for PETRA was acquired approximately 3 minutes later than that of MPRAGE, 9 to 14 minutes after contrast media administration. PETRA acquires the central portion of k-space first (in a pointwise fashion as shown in figure 1) before acquiring the rest of 3D k-space with radial trajectories. A previous study showed that for enhancing intracranial lesions with a diameter of 5 mm or larger, enhancement reached a plateau in less than 10 minutes and lasted until at least 20 minutes after contrast media injection [9]. Thus, the 3 minute difference in the acquisition of k = 0 between MPRAGE and PETRA would not result in differing lesion enhancement, and any difference in lesion enhancement can be taken as primarily due to sequence characteristics. Protocols provided with the PETRA sequence were designed to parallel the contrast and spatial resolution (0.9 to 1.0 mm cubic voxels) typically available with MPRAGE. Therefore, initial work with PETRA at our hospital focused on comparisons with MPRAGE. One of the protocols placed the priority on signal-to-noise ratio (SNR) (TI 900 ms), and one placed the priority on contrast- to-noise ratio (CNR) (TI 500 ms, for higher contrast between gray matter (GM) and white matter (WM)). The scan times of both PETRA protocols were adjusted such that they were similar to that of the MPRAGE protocol used in patient exams. In a pilot study, volunteers and patients were scanned with both PETRA protocols and with Table 1: Sequence parameters PETRA Figs. 2, 3, 11, 12 PETRA Figs. 4–10 MPRAGE Figs. 2–8 3D-FLASH Figs. 9–12 Voxel size / mm (0.99 mm)3 for Figs. 2,3 (0.80 mm)3 for Figs. 11, 12 (0.99 mm)3 (0.94 mm)3 0.6 × 0.6 × 1.0 mm3 Matrix 288 × 288 288 × 288 256 × 256 346 × 384 Slices 288 288 176 144 TI / ms 500 for Figs. 2, 3 900 for Figs. 11, 12 700 900 n.a. TR / ms 2.79 2.79 5.61 11 TE / ms 0.07 0.07 2.4 6.2 FA / deg 6 6 10 20 FatSat No Yes No Yes Scan time 05:59 06:20 5:56 (3:54 for Fig. 2) 02:52 Technology 36 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 2A 2B MPRAGE. The SNR on PETRA images with TI 900 ms was visibly higher than on MPRAGE images, while PETRA images with TI 500 ms provided more GM-to-WM contrast than necessary for CE studies in the opinion of one radiologist (MI). A decision was made that some of the SNR could be ‘traded for’ tissue CNR, and an intermediate TI of 700 ms was chosen for further CE studies. Statistical comparisons of contrast enhancement and SNR between PETRA and MPRAGE were performed (that study is under review for publication in a peer-reviewed journal). PETRA was also compared with 3D-FLASH while remaining conscious of the fact that, compared to the PETRA implementation discussed in this article, 3D-FLASH was capable of higher spatial resolution. Clinical observations Comparisons of PETRA and MPRAGE Comparisons between PETRA and MPRAGE with similar spatial reso-lution and without fat suppression are shown in figures 2 and 3. GM-to-WM contrast is seen to be similar, both without (Fig. 2) and with (Fig. 3) contrast enhancement (CE). In the latter CE case, a small enhancing lesion appears to have the 2 3 Technology Contrast-enhanced screening for brain metastases was indicated for a 57-year-old male who had lung cancer. A tiny metastasis (diameter 3.5 mm) was detected in the medial portion of the left temporal lobe (arrows) on both sequences. (3A) CE MPRAGE, (3B) CE PETRA with an inversion time of 500 ms. same size and enhancement on the PETRA image as on the MPRAGE image. The remaining comparisons between PETRA and MPRAGE, figures 4 through 8, were contrast-enhanced studies of intracranial primary or metastatic tumors with the same resolution parameters as above. However, two PETRA parameters were modified: TI was changed to 700 ms, sacrificing some GM-to-WM contrast for a gain in SNR, and fat-suppression was added (to PETRA only, avoiding a change to the hospital’s routine MPRAGE protocol). 3A 3B MR imaging was indicated for a 19-month-old* female experiencing seizures with no associated fever. Neither sequence revealed any brain abnormality. (2A) MPRAGE, (2B) CE PETRA with a TI of 500 ms demonstrated excellent gray-to-white-matter contrast comparable with that of MPRAGE. * MR scanning has not been established as safe for imaging fetuses and infants less than two years of age. The responsible physician must evaluate the benefits of the MR examination compared to those of other imaging procedures. MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 37 1Repetition time of RF excitation pulses, which for MPRAGE is displayed on the MR console as ‘Echo spacing’.
- 20. Technology Technology 6A6B Contrast-enhanced screening for brain metastases was indicated for a 70-year-old female who had lung cancer. A small metastasis was detected in the supependymal zone of the pons (arrows) on both sequences. (6A) CE MPRAGE, (6B) fat-suppressed CE PETRA. 6 7A 7B Contrast-enhanced screening for brain metastases was indicated for a 69-year-old male who had lung cancer. A small metastasis was detected in the corticomedullary junction of the left parietal lobe on both sequences. (7A) CE MPRAGE, (7B) fat-suppressed CE PETRA. 7 4A 4B Contrast-enhanced images of gliobastoma in a 56-year-old female patient (not proven histologically). (4A) CE MPRAGE, (4B) fat-suppressed CE PETRA. 4 5A 5B Contrast-enhanced images of glioblastoma in a 33-year-old male patient. (5A) CE MPRAGE, (5B) fat-suppressed CE PETRA. 5 38 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 39
- 21. Technology Technology 8B Contrast-enhanced screening for brain metastases was indicated for a 63-year-old male. A ring-enhancing lesion was detected in the left temporal lobe on both sequences. (8A) CE MPRAGE, (8B) fat-suppressed CE PETRA. 8 8A Comparisons of PETRA and 3D-FLASH Comparisons between PETRA and 3D-FLASH are shown in figures 9 through 12. While the acquired spatial resolution was higher for 3D-FLASH (0.6 × 0.6 × 1.0 mm3) than for PETRA, the clinical findings were not affected in these cases. PETRA had a voxel size of 0.993 mm3 (Figs. 9, 10) or 0.803 mm3 (Figs. 11, 12). General clinical observations Susceptibility-related artifacts and flow voids were absent on PETRA images, while signal from cortical bone was observed. All three observations can be attributed to the ultra-short TE. The absence of susceptibility- related artifacts should allow PETRA to detect sinusitis or tumors within the paranasal sinsuses which tend to be highly distorted on 3D gradient-echo-based sequences such as MPRAGE and 3D-FLASH. Positive signal from bone may be useful in cases of head trauma for detecting fractures and in surgical planning and follow-up. While CT is normally used for this purpose in adults, its use is highly restricted in children. PETRA may be able to readily provide 3D bone images even for children, or to provide more frequent follow-ups after surgery in adults. The masticator space and the paranasal space at the skull base tended to appear ‘dirty’ or ‘messy’ on PETRA images, but this was not the result of an artifact or distortion. Rather, this appearance was caused by strong venous enhancement due to the absence of flow voids. Also on PETRA, the dura mater as well as the mucosa in the paranasal sinuses exhibited contrast enhancement, and the enhancement of the dura was uniform in most cases. This was likely due to blood pool enhancement in capillary arteries which are dense in those tissues, in combination with the absence of flow voids as a result the ultra-short TE. Such enhancement did not appear on MPRAGE and 3D-FLASH images. The uniform enhancement of the dura would prevent the use of PETRA for the detection of dural inflammation, dural metastasis, intracranial hypotension or other causes of local dural enhancement. Nevertheless, many other applications of 3D T1-weighted imaging exist such as those presented in the current article. Finally, PETRA demonstrated excellent fat suppression which would allow the sequence to be employed, not only for the diagnosis of intra-cranial tumors, but also for the diagnosis of extracranial, orbital and paranasal tumors including bone-marrow metastases of the calvaria and cranial base. 9 Contrast-enhanced images of dilated, abnormal medullary veins representing developmental venous anomaly (red arrows) in a 47-year-old female patient. Yellow arrows: Small blood pool enhancement in the combined cavernous malformation. Blue arrows: T1 shortening caused by methemoglobin. (9A) CE 3D-FLASH with a voxel size of 0.6 × 0.6 × 1.0 mm3. (9B) CE PETRA (TI 700 ms) with a voxel size of (0.99 mm)3. 9A 9B 10 Contrast-enhanced images of combined cavernous malformation and developmental venous anomaly in a 47-year-old female patient. (10A) CE 3D-FLASH with a voxel size of 0.6 × 0.6 × 1.0 mm3. (10B) CE PETRA (TI 700 ms) with a voxel size of (0.99 mm)3. Flow voids are absent due to the ultra-short TE causing the venous malformation to appear more prominently. Developmental venous malformation was apparent in the same patient, again appearing more prominently on PETRA due to the ultra-short TE and lack of flow voids. (10C) CE 3D-FLASH. (10D) CE PETRA. 10A 10B 10C 10D 40 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 41
- 22. Technology Technology 11A11B gradient-echo sequence-initial experience 11 Contrast-enhanced images of small vestibular schwannoma (arrows) localized in the right acoustic canal of a 67-year-old female patient. (11A) CE 3D-FLASH with a voxel size of 0.6 × 0.6 × 1.0 mm3. (11B) CE PETRA (TI 900 ms) with a voxel size of (0.80 mm)3. 12 Normal optic nerves and paranasal sinuses in a 74-year-old male patient. (12A) 3D-FLASH with a voxel size of 0.6 × 0.6 × 1.0 mm3. (12B) PETRA (TI 900 ms) with a voxel size of (0.80 mm)3. The septi of the paranasal sinuses are depicted clearly in the ethmoid sinuses due to the absence of susceptibility-induced artifacts. Normal paranasal sinuses in the same patient. (12C) 3D-FLASH. (12D) PETRA. Notice the absence of susceptibility-induced artifacts in the paranasal sinus. Conclusion The acoustic noise (A-weighted) generated by PETRA was drastically lower than that of MPRAGE, while contrast- enhancement and image quality were similar between the two sequences, and clinical findings did not differ, as shown in several examples. In comparisons of PETRA with 3D-FLASH, although the latter provided a higher spatial resolution, again clinical findings did not differ. Quieter MRI examinations will be more comfortable for all patients, and may have particular advantages for pediatric, dementia and certain psychiatric patients. References 1 Grodzki DM, Heismann B. Quiet T1-weighted head scanning using PETRA. Proc ISMRM 2013; 21:0456. 2 Weiger M, Pruessmann KP, Hennel F. MRI with zero echo time: hard versus sweep pulse excitation. Magn Reson Med 2011; 66(2):379-89. 3 Idiyatullin D, Corum C, Park JY, Garwood M. Fast and quiet MRI using a swept radiofrequency. J Magn Reson 2006; 181(2): 342-349. Contact Masahiro Ida, M.D. Chief Radiologist Dept. of Radiology Tokyo Metropolitan Ebara Hospital 4-5-10 Higashi-yukigaya, Ota-ku Tokyo 145-0065 Japan Phone: +81 3-5734-8000 [email protected] 4 Grodzki DM, Jakob PM, Heismann B. Ultrashort echo time imaging using pointwise encoding time reduction with radial acquisition (PETRA). Magn Reson Med 2012; 67(2):510-508. 5 Chamberlain R, Moeller S, Corum C, Idiyatullin C, Garwood M. Quiet T1- and T2-weighted brain imaging using SWIFT. Proc ISMRM 2011; 19:2723. 6 Mugler JP, Brookeman JR. Three-dimensional magnetization-prepared rapid gradient-echo imaging (3D MP RAGE). Magn Reson Med 1990; 15:152-157. 7 Brant-Zawadzki M, Gillan GD, Nitz WR. MP RAGE: a three-dimensional, T1-weighted, in the brain. Radiology 1992; 182:769-75. 8 Brant-Zawadzki MN, Gillan GD, Atkinson DJ, Edalatpour N, Jensen M. Three-dimensional MR imaging and display of intracranial disease: improvements with the MP-RAGE sequence and gadolinium. J Magn Reson Imaging. 1993; 3(4): 656-62. 9 Yuh WT, Tali ET, Nguyen HD, Simonson TM, Mayr NA, Fisher DJ. The effect of contrast dose, imaging time, and lesion size in the MR detection of intracerebral metastasis. AJNR 1995; 16:373-380. 12A 12B 12C 12D 42 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world Answers for life. www.siemens.com/quiet-suite Quiet Suite Imaging is to be seen, not heard. Scan and listen to Quiet-Suite.
- 23. Acute MR StrokeProtocol in Six Minutes Kambiz Nael; Rihan Khan; Kevin Johnson; Diego Martin University of Arizona, Department of Medical Imaging, Tucson, AZ, USA Background Stroke is a common and serious disorder, with an annual incidence of approximately 795,000. Based on American Heart Association statistics update in 2010, approximately 610,000 of these are first attacks, and 185,000 are recurrent attacks. On average, every 40 seconds, someone in the United States has a stroke with an estimated mortality rate of 5.5%, claiming approximately 1 of every 18 deaths in the United States [1]. cranial hemorrhage including tissue viability, site of occlusion, and collateral status. While computed tomography (CT) is the most widely available and faster imaging modality, some comprehensive stroke centers favor streamlined MR protocols over CT in the acute stroke setting due to the higher specificity and superior tissue characterization afforded by MRI. The success of CT in initial evaluation of AIS is due, in part, to fast acquisition time, widespread availability and ease of interpretation in the emergency setting. The introduction of multi-slice technology has dramatically increased the speed and simplicity of CT techniques and has set a high standard for alternative Neuroimaging plays a central role in the evaluation of patients with acute ischemic stroke (AIS). With improved technology over the last decade, imaging now provides information beyond the mere presence or absence of intraimaging techniques. A comprehensive CT stroke algorithm including parenchymal imaging (non-contrast head CT), CT angiography (CTA), and perfusion/penumbral imaging by CT perfusion can now be acquired and processed in less than 10 minutes [5, 6]. MRI has been demonstrated to be more sensitive for the detection of acute ischemia and more specific for delineation of infarction core volume when compared to CT [7, 8]. However, due to longer acquisition time and limited availability; it has been mainly used in large institutions and comprehensive stroke centers. A comprehensive MR protocol including parenchymal imaging, MRA and MR perfusion can now be obtained in the order of 20 minutes as demonstrated in several clinical trials [9–13]. If MRI is to compete with CT for evaluation of acute stroke, there is need for further improvements in acquisition speed. In this article we describe our modified acute stroke MRI protocol that can be obtained in approximately 6 minutes rivaling that of any comprehensive acute stroke CT protocol. We describe the technical aspects and review a few clinical examples based on our preliminary results. 92-year-old man with sudden onset of right-sided weakness and aphasia presented to our emergency department after receiving IV-tPA at an outside institution. The acute stroke protocol was performed after 9 hours from the onset in our institution and selective images are shown. 1A 1B 1C Serial aligned DWI, ADC, EPI-FLAIR and EPI-GRE images are shown. There is acute infarction of the left MCA distribution involving the left operculum and insula. Small focus of hemorrhagic conversion is present within the area of infarction seen on both EPI-FLAIR and EPI-GRE images. 1A Aligned DSC-Tmax, DSC-CBF and DSC-CBV images are shown. DSC maps show a heterogeneous pattern of perfusion deficit containing a small perfusion defect in the region of hemorrhage and predominant luxury perfusion along the left MCA territory seen on Tmax and CBF maps. 1B Coronal MIP from CE-MRA of the entire supra-aortic arteries and cropped volume-rendered reconstruction of the intracranial arteries show no evidence of hemodynamic significant arterial stenosis nor occlusion involving the proximal arteries. Note the high diagnostic image quality of the CE-MRA images which are obtained after administration of 8 ml of contrast. 1C 44 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world How-I-do-it MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 45 How-I-do-it
- 24. How-I-do-it How-I-do-it BothFLAIR and GRE images have been used to detect intraarterial clot with variable sensitivity and specificity [16, 17]. Introduction of fast imaging techniques such as parallel acquisition [18] and EPI [19, 20] has significantly enhanced the performance of MR imaging in terms of acquisition speed. The main advantage of EPI, as in the case of DWI imaging, is rapid acquisition time, which is made possible by rapid gradient switching which permits the acquisition of all frequency and phase encoding steps during a single pulse cycle. The addition of parallel imaging can further enhance the acquisition speed and may also serve to mitigate the geometric distortion and susceptibility artifacts commonly associated with long echo-train sequences such as EPI [21, 22]. If their potential is realized, the application of EPI and parallel imaging techniques to the FLAIR and GRE sequences can result in reduction of image acquisition time of the entire brain to less than a minute, a three-fold reduction in scan time over conventional imaging [23, 24]. 2. MR Angiogram An important aspect of the workup of patients with AIS is the imaging of both the intracranial and extracranial vasculature. Precise imaging of the vascular tree is required during the initial assessment of patients with acute stroke to accurately detect the site of arterial disease, which in turn can be crucial in determining the type of acute therapy they are given. Intravenous thrombolysis has been shown to be more effective in small distal vessels than in the large vessels [25]. Larger vessel occlusion may be more effectively treated with intra-arterial thrombolysis or clot retrieval devices while associated with fewer complications [26, 27]. In addition, MRA of the extracranial circulation (neck arteries) is essential to establish the mechanism of ischemia and to prevent subsequent episodes. Extracranial tandem stenoses with plaque involving the carotid or vertebral arteries can be the source of disease that triggers an acute stroke. Time-of-flight MRA (TOF-MRA) has been traditionally used in routine stroke protocols to evaluate the status of neck and brain arteries. Despite its promising results [28], TOF-MRA has significant disadvantages including spin saturation and phase dispersion due to slow or turbulent flow [29, 30]. This can result in overestimation of arterial stenosis and increase false positive rates, usually due to slow flow distal to a subocclusive thrombus or clot. Most importantly the acquisition time usually is long, typically lasting 5–7 minutes. The general consensus is that contrast-enhanced MR angiography ( CE-MRA) provides more accurate imaging of extracranial vessel morphology and of the degree of stenosis than TOF-MRA techniques [31–33]. However, CE-MRA has not been widely incorporated into acute stroke protocols for several reasons. First, CE-MRA has lower spatial resolution relative to TOF-MRA, since the competing requirements of coverage and acquisition speed generally force a compromise in spatial resolution for CE- MRA [34]. A second potential limitation to incorporation of CE-MRA into clinical stroke protocols is related to the requirement of an extra contrast dose, which would be in addition to the intravenous contrast bolus normally utilized for perfusion imaging. With introduction of high performance MR scanners and recent advances in fast imaging tools such as parallel acquisition (GRAPPA) [18], high matrices can now be spread out over a large field-of-view encompassing the entire head and neck, resulting in acquisitions with submillimeter voxel sizes and acquisition times on the order of 20 seconds [35, 36]. 3. MR Perfusion MR perfusion imaging has been used broadly in the identification of potentially salvageable tissue to determine the best treatment strategy in patients with acute ischemic stroke. Although the concept of perfusion-diffusion mismatch remains controversial [37, 38], it has been used with some success to identify patients who may respond favorably to revascularization therapies in several clinical trials [12, 13, 39]. Faster image acquisition combined with higher signal-to-noise ratio (SNR) resulting from the use of gadolinium contrast agents has helped dynamic susceptibility contrast (DSC) perfusion become a more robust and widely accepted technique in comparison to arterial spin labeling (ASL) to identify the presence of perfusion abnormalities in patients with AIS. A refined MR stroke protocol that can combine both CE-MRA and DSC-perfusion with improved acquisition time and diagnostic image quality as previously suggested [47, 48] may have important therapeutic and prognostic implications in the management of patients with acute stroke. Higher inherent SNR of higher magnetic fields such as 3T with improved multi-coil technology has resulted in acquisition of low dose CE-MRA of the supra-aortic arteries with contrast dose as low as 8 ml [40, 41]. A modified 2-phase contrast injection scheme [46] can be used to perform both CE-MRA and DSC perfusion imaging, without the need for additional contrast. The influence of contrast dose reduction on DSC perfusion has been evaluated by several investigators [42, 43] and contrast dose as low as 0.05 mmol/kg has been used to perform DSC perfusion with promising results [44, 45]. Advances in MR technology including hardware and software, faster gradient performance of MR scanners, improved sequence design and fast imaging tools such as EPI and parallel Technical consideration A comprehensive MR stroke protocol has three essential components: 1) Parenchymal imaging that identifies the presence and size of an irreversible infarcted core and determines the presence of hemorrhage; 2) MR angiogram to determine the presence of proximal arterial occlusion and/or intravascular thrombus that can be treated with thrombolysis or thrombectomy; 3) Pwerfusion imaging to determine the presence of hypoperfused tissue at risk for subsequent infarction if adequate perfusion is not restored. Below we describe each of these components in detail and explain how recent technical advances can be used to enhance the performance of the different aspects of acute stroke imaging. 1. Parenchymal imaging This encompasses three parts: 1) DWI (diffusion-weighted imaging) that can detect ischemic tissue within minutes of its occurrence and has emerged as the most sensitive and specific imaging technique for acute ischemia, far beyond NECT or any other type of MRI sequences [14]. 2) FLAIR that helps to age the infarction and permits the detection of subtle subarachnoid hemorrhage; 3) GRE to detect parenchymal hemorrhage with comparable accuracy for the acute intraparenchymal hemorrhage to CT [15]. 2A 2B 2C Serial aligned DWI, ADC, EPI-FLAIR and EPI-GRE images are shown. There is acute right hemispheric infarction involving both the ACA and MCA territories. The EPI-FLAIR images demonstrate corresponding hyperintensity suggestive of completed infarction. There is associated mass effect. No hemorrhage is identified on corresponding EPI-GRE images. 2B Coronal MIP from CE-MRA of the 2A Aligned DSC-Tmax, DSC-CBF and DSC-CBV images are shown. There is a matched perfusion defect with the region of infarction. entire supra-aortic arteries shows complete occlusion of the right cervical ICA shortly after the origin. There is some reconstitution of flow signal at the supracliniod ICA likely via collaterals. 2C 68-year-old man with left sided weakness and altered level of consciousness of unknown onset. 46 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 47
- 25. How-I-do-it How-I-do-it acquisitionhave promised the potential for a fast but comprehensive MR stroke protocol that can be performed in approximately 6 minutes rivaling those of CT protocols. Next we review our stroke protocol in terms of image acquisition and sequence parameters and show some of the clinical examples that were performed at our institution. How we do it At our institution, absent contraindication, MR is the default imaging modality for AIS. An MR safety questionnaire is administered, and MR compatible ECG leads are placed in the emergency department as patients are being evaluated by the neurology team. Patients are then placed onto an MR compatible table and wheeled to the MR magnet for rapid imaging. We use both 3T and 1.5T MR scanners (MAGNETOM Skyra and MAGNETOM Aera, Siemens Healthcare, Erlangen, Germany), with 3T the default scanner for acute stroke imaging when available. For signal reception, a combination of a 16-element array coil [head (n = 12), neck (n = 4)] will be used. The coil design allows for application of parallel acquisition in both the phase and slice encoding directions. Our 6-minute MR imaging protocol consist of DWI, EPI-FLAIR, EPI- GRE, Table 1: Imaging protocol CE-MRA and DSC perfusion. The clinical indications for using this acute MR stroke protocol are patients with acute (< 9 hours) presentation from the onset of symptoms, unknown onset of symptoms, NIHSS > 4, or aphasia. Table 1 shows the sequence parameters of our acquisition protocol. A modified 2-phase contrast injection scheme [46] is used to perform both CE-MRA and DSC perfusion imaging, without the need for additional contrast. To accomplish this, the total volume of 20 ml of gadolinium ( Multihance, Bracco Diagnostics Inc., Princeton, NJ, USA) that is used routinely for MR perfusion is diluted with normal saline to a total 50 ml volume. Using a timing bolus, a total of 3 ml of contrast solution (1.2 ml of gadolinium) is injected at 1.5 ml/s to determine the transit time from the arm vein to the cervical carotid arteries. Then, a total of 22 ml contrast solution (8.8 ml of gadolinium) is injected at the same flow rate as the timing injection for the CE-MRA acquisition. A centric ordering k-space is used for CE-MRA to minimize intracranial venous contamination. Subsequently, the remaining 25 ml of contrast solution (10 ml of gadolinium) is injected at 5 ml/s for the MR perfusion scan which is performed at the end. Image analysis Following data acquisition, CE-MRA image processing is performed on the scanner console with standard commercial software using a maximum intensity projection (MIP) algorithm. All of the reconstructed data, as well as the source images are available on the workstation for image analysis. Perfusion analysis will be performed off-line on a dedicated FDA-approved workstation (Olea-sphere, Olea Medical SA, France). The arterial input function is selected automatically and multiparametric perfusion maps including time-to-peak (TTP), time-to-maximum (Tmax) cerebral blood flow (CBF) and cerebral blood volume (CBV) are then calculated using a block-circulant singular value decomposition technique [49]. Our initial results using the described stroke MR protocol have been promising. We have scanned more than 600 patients with ASI since January 2013. More than 97% of our studies have been rated with diagnostic image quality. The EPI-FLAIR sequence has been used in parallel to conventional FLAIR in a subset of patients with comparable qualitative and quantitative results [24]. In a study of 52 patients with AIS, the mean ± SD of the signal intensity ratios on EPI-FLAIR and FLAIR for DWI positive lesions were 1.28 ± 0.16 and 1.25 ± 0.17 respectively with significant DWI EPI-FLAIR EPI-GRE CE-MRA DSC TR (ms) 4600 10000 (TI:2500) 1860 3.36 1450 TE 65 88 48 1.24 22 FA (degrees) – 90 90 25 90 Matrix 160 192 192 448 128 FOV 220 220 220 340 220 Slices (n × thickness) 30 × 4 30 × 4 40 × 3 120 × 0.8 30 × 4 Bandwidth (Hz/pixel) 1250 1488 964 590 1502 Parallel acquisition (GRAPPA) 3 3 – 3 3 Acquisition time 58 sec 52 sec 56 sec 20 sec 1 min and 30 sec correlation (r = 0.899, z value = 8.677, p < 0.0001). The EPI-GRE sequence has been also used in parallel to conventional GRE in a subset of patients with comparable results in terms of detection of hemorrhage (Fig. 1) and blood clot in proximal arteries. The combination of CE-MRA and DSC has been successfully tested in our institution [48] with diagnostic image quality. In a cohort of 30 patients with acute stroke, the specificity of CE-MRA for detection of arterial stenosis > 50% was 97% compared to 89% for TOF-MRA when compared to DSA as the standard of reference [48]. DSC perfusion imaging with reduced contrast dose is feasible with comparable quantitative and qualitative results to a full-dose control group [48]. Importantly, the presence of contrast in the circulating blood of the CE- MRA half-dose group does not negatively impact the image quality nor the quantitative analysis of perfusion data when compared to the control full-dose group. Conclusion Described multimodal MR protocol is feasible for evaluation of patients with acute ischemic stroke with total acquisition time of 6 minutes rivaling that of the multimodal CT protocol. References 1 Lloyd-Jones D, Adams RJ, Brown TM, et al. Heart disease and stroke statistics – 2010 update: a report from the American Heart Association. Circulation. Feb 23 2010;121(7):e46-e215. 2 Saver JL. Time is brain-quantified. Stroke. Jan 2006;37(1):263-266. 3 Michel P, Bogousslavsky J. Penumbra is brain: no excuse not to perfuse. Ann Neurol. Nov 2005;58(5):661-663. 4 Gonzalez RG. Imaging-guided acute ischemic stroke therapy: From “time is brain” to “physiology is brain”. AJNR. American journal of neuroradiology. Apr 2006;27(4):728-735. 15 Kidwell CS, Chalela JA, Saver JL, et al. Comparison of MRI and CT for detection of acute intracerebral hemorrhage. JAMA : the journal of the American Medical Associ-ation. Oct 20 2004;292(15):1823-1830. 16 Flacke S, Urbach H, Keller E, et al. Middle cerebral artery (MCA) susceptibility sign at susceptibility-based perfusion MR imaging: clinical importance and comparison with hyperdense MCA sign at CT. Radiology. May 2000;215(2):476-482. 17 Assouline E, Benziane K, Reizine D, et al. Intra-arterial thrombus visualized on T2* gradient echo imaging in acute ischemic stroke. Cerebrovasc Dis. 2005;20(1):6-11. 18 Griswold MA, Jakob PM, Heidemann RM, et al. Generalized autocalibrating partially parallel acquisitions (GRAPPA). Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine. Jun 2002;47(6):1202-1210. 19 Mansfield P. Real-time echo-planar imaging by NMR. Br Med Bull. Apr 1984;40(2): 187-190. 20 DeLaPaz RL. Echo-planar imaging. Radio-graphics. Sep 1994;14(5):1045-1058. 21 Pruessmann KP. Parallel imaging at high field strength: synergies and joint potential. Topics in magnetic resonance imaging : TMRI. Aug 2004;15(4):237-244. 22 Wiesinger F, Van de Moortele PF, Adriany G, De Zanche N, Ugurbil K, Pruessmann KP. Potential and feasibility of parallel MRI at high field. NMR in biomedicine. May 2006;19(3):368-378. 23 Kinoshita T, Okudera T, Tamura H, Ogawa T, Hatazawa J. Assessment of lacunar hemorrhage associated with hypertensive stroke by echo-planar gradient-echo T2*-weighted MRI. Stroke. Jul 2000;31(7): 1646-1650. 24 Meshksar A, Khan R, Carmody R, Nael K. The Role of Echo-planar Fluid-Attenuated Inversion Recovery (EPI-FLAIR) in Acute Stroke Setting: A Feasibility Study. Paper presented at: ASNR; May 22, 2013, 2013; San Diego, CA. 25 del Zoppo GJ, Poeck K, Pessin MS, et al. Recombinant tissue plasminogen activator in acute thrombotic and embolic stroke. Ann Neurol. Jul 1992;32(1):78-86. 26 Furlan A, Higashida R, Wechsler L, et al. Intra-arterial prourokinase for acute ischemic stroke. The PROACT II study: a randomized controlled trial. Prolyse in Acute Cerebral Thromboembolism. JAMA : the journal of the American Medical Association. Dec 1 1999;282(21): 2003-2011. 5 Zhu G, Michel P, Aghaebrahim A, et al. Computed tomography workup of patients suspected of acute ischemic stroke: perfusion computed tomography adds value compared with clinical evaluation, noncontrast computed tomography, and computed tomography angiogram in terms of predicting outcome. Stroke; a journal of cerebral circulation. Apr 2013;44(4):1049-1055. 6 Schaefer PW, Roccatagliata L, Ledezma C, et al. First-pass quantitative CT perfusion identifies thresholds for salvageable penumbra in acute stroke patients treated with intra-arterial therapy. AJNR. American journal of neuroradiology. Jan 2006;27(1):20-25. 7 Jauch EC, Saver JL, Adams HP, Jr., et al. Guidelines for the early management of patients with acute ischemic stroke: a guideline for healthcare professionals from the American Heart Association/ American Stroke Association. Stroke. Mar 2013;44(3):870-947. 8 Chalela JA, Kidwell CS, Nentwich LM, et al. Magnetic resonance imaging and computed tomography in emergency assessment of patients with suspected acute stroke: a prospective comparison. Lancet. Jan 27 2007;369(9558):293-298. 9 Kang DW, Chalela JA, Dunn W, Warach S, Investigators NI-SSC. MRI screening before standard tissue plasminogen activator therapy is feasible and safe. Stroke. Sep 2005;36(9):1939-1943. 10 Hjort N, Butcher K, Davis SM, et al. Magnetic resonance imaging criteria for thrombolysis in acute cerebral infarct. Stroke. Feb 2005;36(2):388-397. 11 Schellinger PD, Jansen O, Fiebach JB, Hacke W, Sartor K. A standardized MRI stroke protocol: comparison with CT in hyperacute intracerebral hemorrhage. Stroke; a journal of cerebral circulation. Apr 1999;30(4):765-768. 12 Albers GW, Thijs VN, Wechsler L, et al. Magnetic resonance imaging profiles predict clinical response to early reperfusion: the diffusion and perfusion imaging evaluation for understanding stroke evolution (DEFUSE) study. Ann Neurol. Nov 2006;60(5):508-517. 13 Davis SM, Donnan GA, Parsons MW, et al. Effects of alteplase beyond 3 h after stroke in the Echoplanar Imaging Thrombolytic Evaluation Trial (EPITHET): a placebo-controlled randomised trial. Lancet neurology. Apr 2008;7(4): 299-309. 14 Fiebach JB, Schellinger PD, Jansen O, et al. CT and diffusion-weighted MR imaging in randomized order: diffusion-weighted imaging results in higher accuracy and lower interrater variability in the diagnosis of hyperacute ischemic stroke. Stroke; a journal of cerebral circulation. Sep 2002;33(9):2206-2210. 48 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 49
- 26. Contact Kambiz Nael,M.D. Assistant Professor of Radiology Director of Neuroradiology MRI University of Arizona Medical Center Department of Medical Imaging, Neuroradiology Section 1501 N. Campbell, PO Box 245067 Tucson, AZ 85724-5067 USA Phone: +1 520-626-2138 Fax: +1 520-626-7093 [email protected] 37 Hacke W, Furlan AJ, Al-Rawi Y, et al. Intravenous desmoteplase in patients with acute ischaemic stroke selected by MRI perfusion-diffusion weighted imaging or perfusion CT (DIAS-2): a prospective, randomised, double-blind, placebo-controlled study. Lancet neurology. Feb 2009;8(2):141-150. 38 Kidwell CS, Jahan R, Gornbein J, et al. A trial of imaging selection and endovascular treatment for ischemic stroke. The New England journal of medicine. Mar 7 2013;368(10):914-923. 39 Lansberg MG, Straka M, Kemp S, et al. MRI profile and response to endovascular reperfusion after stroke (DEFUSE 2): a prospective cohort study. Lancet neurology. Oct 2012;11(10):860-867. 40 Tomasian A, Salamon N, Lohan DG, Jalili M, Villablanca JP, Finn JP. Supraaortic arteries: contrast material dose reduction at 3.0-T high-spatial-resolution MR angiography-feasibility study. Radiology. Dec 2008;249(3):980-990. 41 Nael K, Moriarty JM, Finn JP. Low dose CE-MRA. Eur J Radiol. Oct 2011;80(1): 2-8. 42 Heiland S, Reith W, Forsting M, Sartor K. How do concentration and dosage of the contrast agent affect the signal change in perfusion-weighted magnetic resonance imaging? A computer simulation. Magnetic resonance imaging. Jul 2001;19(6):813-820. 43 Alger JR, Schaewe TJ, Lai TC, et al. Contrast agent dose effects in cerebral dynamic susceptibility contrast magnetic resonance perfusion imaging. Journal of magnetic resonance imaging: JMRI. Jan 2009;29(1):52-64. 44 Manka C, Traber F, Gieseke J, Schild HH, Kuhl CK. Three-dimensional dynamic susceptibility-weighted perfusion MR imaging at 3.0 T: feasibility and contrast agent dose. Radiology. Mar 2005; 234(3):869-877. 45 Alger JR, Schaewe TJ, Liebeskind DS, Saver JL, Kidwell CS. On the feasibility of reduced dose Dynamic Susceptibility Contrast perfusion MRI for stroke. Paper presented at: Intl. Soc. Mag. Reson. Med; 7-13 May 2011, 2011; Montréal, Québec, Canada. 46 Habibi R, Krishnam MS, Lohan DG, et al. High-spatial-resolution lower extremity MR angiography at 3.0 T: contrast agent dose comparison study. Radiology. Aug 2008;248(2):680-692. 47 Ryu CW, Lee DH, Kim HS, et al. Acquisition of MR perfusion images and contrast-enhanced MR angiography in acute ischaemic stroke patients: which procedure should be done first? The British journal of radiology. Dec 2006; 79(948):962-967. 48 Nael K, Pirastehfar M, Villablanca JP, Salamon N. Addition of a Low-dose Contrast Enhanced MRA at 3.0T in the Assessment of Acute Stroke: A More Efficient and Accurate Stroke Protocol. Paper presented at: ASNR 50th Annual Meeting; April, 2012; New York, NY. 49 Wu O, Ostergaard L, Weisskoff RM, Benner T, Rosen BR, Sorensen AG. Tracer arrival timing-insensitive technique for estimating flow in MR perfusion-weighted imaging using singular value decomposition with a block-circulant deconvolution matrix. Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine. Jul 2003;50(1):164-174. How-I-do-it 27 Becker KJ, Brott TG. Approval of the MERCI clot retriever: a critical view. Stroke; a journal of cerebral circulation. Feb 2005; 36(2):400-403. 28 Yucel EK, Anderson CM, Edelman RR, et al. AHA scientific statement. Magnetic resonance angiography : update on applications for extracranial arteries. Circu-lation. Nov 30 1999;100(22):2284-2301. 29 Isoda H, Takehara Y, Isogai S, et al. MRA of intracranial aneurysm models: a comparison of contrast-enhanced three-dimensional MRA with time-of-flight MRA. J Comput Assist Tomogr. Mar-Apr 2000; 24(2):308-315. 30 Lin W, Tkach JA, Haacke EM, Masaryk TJ. Intracranial MR angiography: application of magnetization transfer contrast and fat saturation to short gradient-echo, velocity-compensated sequences. Radiology. Mar 1993;186(3):753-761. 31 Somford DM, Nederkoorn PJ, Rutgers DR, Kappelle LJ, Mali WP, van der Grond J. Proximal and distal hyperattenuating middle cerebral artery signs at CT: different prognostic implications. Radiology. Jun 2002;223(3):667-671. 32 Cosottini M, Pingitore A, Puglioli M, et al. Contrast-enhanced three-dimensional magnetic resonance angiography of atherosclerotic internal carotid stenosis as the noninvasive imaging modality in revascularization decision making. Stroke; a journal of cerebral circulation. Mar 2003; 34(3):660-664. 33 Huston J, 3rd, Fain SB, Wald JT, et al. Carotid artery: elliptic centric contrast-enhanced MR angiography compared with conventional angiography. Radiology. Jan 2001;218(1):138-143. 34 Fellner C, Lang W, Janka R, Wutke R, Bautz W, Fellner FA. Magnetic resonance angiography of the carotid arteries using three different techniques: accuracy compared with intraarterial x-ray angiography and endarterectomy specimens. Journal of magnetic resonance imaging : JMRI. Apr 2005;21(4):424-431. 35 Nael K, Villablanca JP, Pope WB, McNamara TO, Laub G, Finn JP. Supraaortic arteries: contrast-enhanced MR angiography at 3.0 T-highly accelerated parallel acquisition for improved spatial resolution over an extended field of view. Radiology. Feb 2007;242(2):600-609. 36 Phan T, Huston J, 3rd, Bernstein MA, Riederer SJ, Brown RD, Jr. Contrast-enhanced magnetic resonance angiography of the cervical vessels: experience with 422 patients. Stroke; a journal of cerebral circulation. Oct 2001;32(10): 2282-2286. 50 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world TOGETHER WE MOVE FORWARD IN THE FIGHT AGAINST CANCER When two leading companies join forces in the fight against cancer, it broadens the realm of what’s possible. That’s why Varian and Siemens have partnered. Siemens' advanced diagnostic imaging capabilities coupled with Varian's powerful delivery systems and treatment planning tools give even more of an edge in the pursuit of our common goal: to EnVision better cancer care. Together we offer more personalized treatment and expanded care options that aid you in making the best possible decisions for your patients—with confidence. By gathering our strengths, we have the energy and vision to better help healthcare professionals detect, diagnose and treat cancer while paving the way for the future of cancer care. © 2013 Varian, Varian Medical Systems, Trilogy, and ARIA are registered trademarks, and TrueBeam, Edge Radiosurgery and Eclipse are trademarks of Varian Medical Systems, Inc. All other trademarks are property of Siemens AG. Find out more at varian.com/envision Varian Medical Systems International AG Zug, Switzerland Tel: +41 - 41 749 88 44 Fax: +41 - 41 740 33 40 varian.com [email protected] Global Siemens Healthcare Headquarters Siemens AG Healthcare Sector Henkestrasse 127 91052 Erlangen, Germany Tel: +49 9131 84-0 siemens.com/healthcare
- 27. Clinical Neurology NeurologyClinical at the same time, but is generally ignored. A filter is applied to the phase image (High-pass Hamming Window Filter) on a 64 × 64 matrix to reduce aliasing artifacts. A new phase mask is created which, when added to the magnitude image, creates the susceptibility image. In order to obtain a better interpretation, minimum intensity projections (minIP) are used [6]. During post-processing the phase contrast image is filtered to reduce undesirable low spatial frequency components, leaving the high frequency field variations. The phase mask created can be ‘positive’ or ‘negative’. The phase mask is multiplied using the original magnitude image to produce images that maximize the negative intensity of the mineralization of the parenchyma. Minimum intensity projection (usually from 2 to 4 slices) is used to display the processed data [1]. MAGNETOM ESSENZA 1.5 Tesla MRI unit with the following settings: TR 49 ms, TE 40 ms, FA 15°, number of slices 60, slice thickness 2 mm, acquisition matrix 256 ×157. Susceptibility-weighted imaging takes advantage of the loss of signal intensity created by alterations in a homogenous magnetic field; these disturbances can be caused by several different paramagnetic or diamagnetic substances. The loss of signal intensity in the T2*-weighted sequence is a result of the difference in the precession rate of the spins [5]. The susceptibility image is obtained during the acquisition process by combining the magnitude and phase of the images. Routine MR images are magnitude images where the signal’s intensity is converted to a gray scale. Phase information is obtained Introduction Susceptibility-weighted imaging (SWI) is a sequence that utilizes a phenomenon in which the phase and change in the local magnetic field of the tissues are proportional to one another, provided the echo time is constant [1]. It uses magnitude and phase images, as well as a summation of these in a three-dimensional gradient echo sequence with flow compensation [2]. It offers very high sensitivity for visualizing calcium, non-heme iron (ferritin) and hemoglobin degradation products (deoxyhemoglobin and hemosiderin) [3, 4]. Initial experience By means of a series of cases we will illustrate the clinical usefulness of SWI with certain neurological conditions. The studies reviewed were performed in the Neurological Scanography Magnetic Resonance Imaging Service using a Siemens The method is highly sensitive for purposes of visualizing venous circulation, blood products and iron content, and is also useful for evaluating the vascularization of tumors and for identifying brain tissue that has been compromised by a stroke, vascular dementia or trauma, and can also be used in functional imaging [1, 4, 7-9] (Fig. 1). Hemorrhage Oxyhemoglobin, formed by the binding of an oxygen and an iron atom contained in the Hem group, is a diamagnetic substance. When the oxygen is released from the iron atom it forms deoxyhemoglobin, which is paramagnetic because of its unpaired electrons. Metahemoglobin is produced when deoxyhemoglobin oxidizes, making it less stable; in this state there is little susceptibility effect and thus it is more easily visualized in T1w images. Hemosiderin is the final product of the degradation of hemoglobin when it degrades within phagocytic cells, and is a highly paramagnetic [3, 4, 10] substance. Diamagnetic substances produce a weak local magnetic field, while paramagnetics generate a stronger magnetic field that leads to a signal de-phase and therefore a signal reduction in the T2*w sequence [4]. The ferritin produced by different metabolic processes also has para-magnetic characteristics and is associated with Parkinson’s disease, Huntington’s disease and Alzheimer’s disease [9-11]. Trauma In the detection of diffuse axonal damage, this approach is more sensitive than conventional imaging for detecting microhemorrhages in the deep and subcortical white matter, which can be obscured in computed tomography (CT) scans [12, 13]. It is three to six times more sensitive than gradient echo images for detecting the number, size, and location of the lesions associated with this clinical status of the patient [1, 13-16]. It is equally useful in detecting brain-stem lesions, subarachnoid and intraventricular hemorrhage, as well as other types of hemorrhagic lesions of different origins [17] (Fig. 2). Calcifications Calcium is also diamagnetic and can lead to changes in the susceptibility image [12, 18]. SWI differentiates iron from calcium based on their diamagnetic or paramagnetic characteristics in the filtered-phase image. Calcium appears brilliant in this latter image, while the hemorrhage and its derivative products have low signal intensity. This differentiation is important when dealing with neurodegenerative and metabolic diseases, trauma, and tumors [12, 18]. Vascular malformations Venous blood causes non-homogeneity in the magnetic field due to the paramagnetic effect of the deoxygenated blood due to T2* reduction, depending on the oxygen saturation, the hematocrit and the condition of the erythrocytes; thus, the deoxyhemoglobin present in venous blood allows for the visualization of the latter [4] as well as the phase difference between the vessels and surrounding structures [19]. The susceptibility image provides contrast similar to that of a functional image (BOLD blood oxygen level-dependent). SWI is more sensitive in the detection of vascular structures that are hidden to T2* and low-flow malformations that are not detected by MR angiography, such as venous development malformations, telangiectasias and cavernomas, as well as vascular abnormalities and calcifications related to Sturge-Weber Syndrome, since it is not affected by flow velocity or direction [20-24]. In dural sinus thrombosis they show venous statis and collateral flow, as well as early detection of venous hypertension before infarcts or hemorrhages occur [7, 8, 19] (Figs. 3–5). Susceptibility-Weighted Imaging. Initial Experience José Luis Ascencio L.1; Tania Isabel Ruiz Z.2 1 Escanografia Neurologica, Medellin, Colombia 2 Universidad CES, Radiology, Medellín, Antioquia, Colombia 1A Patient with bilateral frontal hemorrhagic contusion. (1A) T2w axial; no lesions observed. (1B) Axial gradient echo shows a low-signal lesion in left frontal lobe with a slight blooming effect. (1C) SWI magnitude, two bilateral frontal hemorrhagic contusions are observed. 1 1B 1C 2A 2B 2C Patient with diffuse axonal lesion. (2A) T2w axial; no lesions observed. (2B) Low-signal, puntiform lesions. (2C) SWI minIP makes the multiple microhemorrhagic lesions more apparent. 2 52 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 53
- 28. Clinical Neurology NeurologyClinical 3A 3B 3C Venous development anomaly. (3A) Axial gradient echo; anomaly not visible. (3B) Axial contrast-enhanced image shows right frontal venous development anomaly that is more evident in the susceptibility image (3C). 3 4A 4B 4 Left frontal cavernoma. (4A) Axial proton density-weighted image; (4B) minIP SWI. 5A 5B 5C Left parietal arteriovenous malformation. (5A) Axial PDw show serpinginous images with absence of flow signal. (5B) mIP SWI. (5C) MIP TOF shows the AVM and the cortical drainage vein. 5 Brain tumors This approach provides information that supplements T1 with contrast for detecting margins, internal architecture, hemorrhage and vascularization of a tumor that are not visible with conventional sequences. This aids in differentiating between a recurring 6A 6B 7A 7B 7C Metastatic melanoma. (7A) T2w axial; large mass displacing the midline, with major edema and hypointense zone due to hemorrhage in the medial portion. (7B) Magnitude image, (7C) MIP SWI shows a greater hemorrhagic component of the mass, on the contralateral side, as well as intraventricular hemorrhaging. 7 6C Hemorrhagic metathesis. (6A) T1w axial gadolinium-enhanced, (6B) T2w axial show a left parietal mass with heterogeneous enhancement, perilesional edema and mass effect on the lateral ventricles. (6C) MIP SWI shows hypervascularity and hemorrhage in the interior of the mass. 6 tumor and post-operative changes. The use of susceptibility imaging before and after the administration of gadolinium can differentiate areas of enhancement of the vessels. Because of its suppression of cerebrospinal fluid, it enhances contrast between edema and normal tissue, similarly to what is provided by FLAIR, thus facilitating the detection of space-occupying lesions [4, 7, 25] (Figs. 6–8). 8A 8B 8C Oligodendroglioma. (8A) T1w axial gadolinium shows mass with enhanced foci and a cystic component, (8B) MIP SWI right parietal hypervascular mass with increased relative flow (8C). 8 54 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 55
- 29. Clinical Neurology NeurologyClinical 9A 9B 9C CVD with hemorrhagic transformation. (9A) Axial T2w, (9B) gradient echo in patient with left parietal hemorrhagic infarct with surrounding edema and hemorrhage. (9C) SWI makes the greater hemorrhagic component more obvious. 9 10A 10B 10C Right MCA aneurism with bleeding. (10A) Axial T2w, (10B) TOF demonstrating aneurysm with bleeding, (10C) SWI aneurism with greater bleeding than that shown in the T2w sequence. 10 Children: CT versus MRI versus Susceptibility Weighted Imaging (SWI). Journal of Neurotrauma. 2011;28(6):915-27. 14 Wang M, Dai Y, Han Y, Haacke EM, Dai J, Shi D. Susceptibility weighted imaging in detecting hemorrhage in acute cervical spinal cord injury. Magnetic resonance imaging. 2011;29(3):365-73. 15 Tong KA, Ashwal S, Holshouser BA, Shutter LA, Herigault G, Haacke EM, et al. Hemorrhagic Shearing Lesions in Children and Adolescents with Posttraumatic Diffuse Axonal Injury: Improved Detection and Initial Results1. Radiology. 2003 May 1, 2003;227(2):332-9. 16 Babikian T, Freier MC, Tong KA, Nickerson JP, Wall CJ, Holshouser BA, et al. Susceptibility weighted imaging: neuropsychologic outcome and pediatric head injury. Pediatr Neurol. 2005;33(3):184-94. 17 Wu Z, Li S, Lei J, An D, Haacke EM. Evaluation of Traumatic Subarachnoid Hemorrhage Using Susceptibility-Weighted Imaging. AJNR Am J Neuroradiol. 2010 August 1, 2010;31(7):1302-10. 18 Wu Z, Mittal S, Kish K, Yu Y, Hu J, Haacke EM. Identification of calcification with MRI using susceptibility-weighted imaging: A case study. Journal of Magnetic Resonance Imaging. 2009;29(1):177-82. 19 Tsui Y-K, Tsai FY, Hasso AN, Greensite F, Nguyen BV. Susceptibility-weighted imaging for differential diagnosis of cerebral vascular pathology: A pictorial review. Journal of the neurological sciences. 2009;287(1):7-16. 20 Hu J, Yu Y, Juhasz C, Kou Z, Xuan Y, Latif Z, et al. MR susceptibility weighted imaging (SWI) complements conventional contrast enhanced T1 weighted MRI in characterizing brain abnormalities of Sturge-Weber Syndrome. Journal of Magnetic Resonance Imaging. 2008;28(2):300-7. 21 Deistung A, Dittrich E, Sedlacik J, Rauscher A, Reichenbach JR. ToF-SWI: Simultaneous time of flight and fully flow compensated susceptibility weighted imaging. Journal of Magnetic Resonance Imaging. 2009;29(6):1478-84. 22 Koopmans P, Manniesing R, Niessen W, Viergever M, Barth M. MR venography of the human brain using susceptibility weighted imaging at very high field strength. Magnetic Resonance Materials in Physics, Biology and Medicine. 2008;21(1):149-58. 23 de Champfleur NM, Langlois C, Ankenbrandt WJ, Le Bars E, Leroy MA, Duffau H, et al. Magnetic Resonance Imaging Evaluation of Cerebral Cavernous Malformations With Susceptibility-Weighted Imaging. Neurosurgery. 2011;68(3): 641-8 10.1227/NEU.0b013e31820773cf. 24 Jagadeesan BD, Delgado Almandoz JE, Moran CJ, Benzinger TLS. Accuracy of Susceptibility-Weighted Imaging for the Detection of Arteriovenous Shunting in Vascular Malformations of the Brain. Stroke. 2011 January 1, 2011;42(1): 87-92. 25 Hori M, Ishigame K, Kabasawa H, Kumagai H, Ikenaga S, Shiraga N, et al. Precontrast and postcontrast susceptibility- weighted imaging in the assessment of intracranial brain neoplasms at 1.5 T. Japanese Journal of Radiology. 2010; 28(4):299-304. 26 Cherian A, Thomas B, Kesavadas C, Baheti N, Wattamwar P. Ischemic hyperintensities on T1-weighted magnetic resonance imaging of patients with stroke: New insights from susceptibility weighted imaging2010 January 1, 2010 Contract No.: 1. 27 Mittal P, Dua S, Kalia V. Pictorial essay: Susceptibility-weighted imaging in cerebral ischemia2010. 28 Santhosh K, Kesavadas C, Thomas B, Gupta AK, Thamburaj K, Kapilamoorthy TR. Susceptibility weighted imaging: a new tool in magnetic resonance imaging of stroke. Clinical Radiology. 2009;64(1):74-83. 29 Hermier M, Nighoghossian N. Contribution of Susceptibility-Weighted Imaging to Acute Stroke Assessment. Stroke. 2004 August 1, 2004;35(8):1989-94. 30 Haacke EM, Makki M, Ge Y, Maheshwari M, Sehgal V, Hu J, et al. Characterizing iron deposition in multiple sclerosis lesions using susceptibility weighted imaging. Journal of Magnetic Resonance Imaging. 2009;29(3):537-44. 31 Niwa T, de Vries L, Benders M, Takahara T, Nikkels P, Groenendaal F. Punctate white matter lesions in infants: new insights using susceptibility-weighted imaging. Neuroradiology. 2011:1-11. 32 Vinod Desai S, Bindu PS, Ravishankar S, Jayakumar PN, Pal PK. Relaxation and susceptibility MRI characteristics in Hallervorden-Spatz syndrome. Journal of Magnetic Resonance Imaging. 2007;25(4):715-20. 33 Kirsch W, McAuley G, Holshouser B, Petersen F, Ayaz M, Vinters HV, et al. Serial susceptibility weighted MRI measures brain iron and microbleeds in dementia. Journal of Alzheimer’s disease: JAD. 2009;17(3):599-609. References 1 Haacke EM, Xu Y, Cheng Y-CN, Reichenbach JR. Susceptibility weighted imaging (SWI). Magnetic Resonance in Medicine. 2004;52(3):612-8. 2 Haacke EM, Mittal S, Wu Z, Neelavalli J, Cheng Y-CN. Susceptibility-Weighted Imaging: Technical Aspects and Clinical Applications, Part 1. AJNR Am J Neuroradiol. 2009 January 1, 2009;30(1):19-30. 3 Mittal S, Wu Z, Neelavalli J, Haacke EM. Susceptibility-Weighted Imaging: Technical Aspects and Clinical Applications, Part 2. AJNR Am J Neuroradiol. 2009 February 1, 2009;30(2):232-52. 4 Sehgal V, Delproposto Z, Haacke EM, Tong KA, Wycliffe N, Kido DK, et al. Clinical applications of neuroimaging with susceptibility-weighted imaging. Journal of Magnetic Resonance Imaging. 2005;22(4):439-50. 5 Tong KA, Ashwal S, Obenaus A, Nickerson JP, Kido D, Haacke EM. Susceptibility- Weighted MR Imaging: A Review of Clinical Applications in Children. AJNR Am J Neuroradiol. 2008 January 1, 2008;29(1):9-17. 6 Matsushita T AD, Arioka T, Inoue S, Kariya Y, Fujimoto M, Ida K, Sasai N, Kaji M, Kanazawa S, Joja I. Basic study of susceptibility- weighted imaging at 1.5T. Acta medica Okayama. [Journal Article]. 2008 Jun;62(3):159-68. 7 Thomas B, Somasundaram S, Thamburaj K, Kesavadas C, Gupta A, Bodhey N, et al. Clinical applications of susceptibility weighted MR imaging of the brain – a pictorial review. Neuroradiology. 2008;50(2):105-16. 8 Ong BC, Stuckey SL. Susceptibility weighted imaging: A pictorial review. Journal of Medical Imaging and Radiation Oncology. 2010;54(5):435-49. 9 Robinson RJ, Bhuta S. Susceptibility- Weighted Imaging of the Brain: Current Utility and Potential Applications. Journal of Neuroimaging. 2011:no-no. 10 Goos JDC, van der Flier WM, Knol DL, Pouwels PJW, Scheltens P, Barkhof F, et al. Clinical Relevance of Improved Microbleed Detection by Susceptibility-Weighted Magnetic Resonance Imaging. Stroke. 2011 May 12, 2011:STROKEAHA. 110.599837. 11 Gupta D, Saini J, Kesavadas C, Sarma P, Kishore A. Utility of susceptibility-weighted MRI in differentiating Parkinson’s disease and atypical parkinsonism. Neuroradiology. 2010;52(12):1087-94. 12 ZHU Wen-zhen QJ-p, ZHAN Chuan-jia, SHU Hong-ge, ZHANG Lin, WANG Cheng-yuan, XIA Li-ming, HU Jun-wu, FENG Ding-yi. Magnetic resonance susceptibility weighted imaging in detecting intracranial calcification and hemorrhage. Chinese Medical Journal. [Journal Article]. 2008 oct 20;121(20):2021-5. 13 Beauchamp MH, Ditchfield M, Babl FE, Kean M, Catroppa C, Yeates KO, et al. Detecting Traumatic Brain Lesions in Contact Jose Luis Ascencio L. Escanografia Neurologica Medellin Colombia [email protected] Cerebrovascular disease The susceptibility image can be used together with diffusion images to detect the hypoperfused region, the presence of hemorrhaging within the infarct (which could affect the treatment), detect acute thrombus and predict the likelihood of hemorrhagic transformation and hemorrhagic complications during and after thrombolysis treatment, as well as microbleeding due to amyloid angiopathy and lacunar infarcts in patients with hypertensive encephalopathy [19, 26-28] (Figs. 9, 10). Vascular occlusion can change the susceptibility of the tissue as a result of reduced arterial flow and an increase in the accumulation of deoxygenated blood, which increases the amount of deoxy-hemoglobin that can be detected by SWI [27, 29]. Neurodegenerative illnesses Certain disorders, such as Parkinson’s Disease, Huntington’s Disease, Alzheimer’s, multiple sclerosis and amyotrophic lateral sclerosis (Lou Gherig’s Disease) present with abnormal iron deposition, which can be detected and quantified using susceptibility imaging [11, 30-33]. SWI can show chronic demyelinating plaques with iron depositions that are hidden in conventional sequences, as the iron content makes the lesions more visible. It can also determine the iron content of the nucleii of deep gray matter that can also be observed in patients with multiple sclerosis, as well as the perivenular distribution of the demyelinating lesions [30]. 56 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 57
- 30. How-I-do-it How-I-do-it CurveFitting of the Lipid-Lactate Range in an MR Spectrum: Some Useful Tips Jamie Ho Xiu Mei; Helmut Rumpel Department of Diagnostic Radiology, Singapore General Hospital, Singapore Purpose In the course of a neurological spectroscopic application, questions about the lipid-lactate signals often occur: How can we accurately interpret an asymmetric pattern with ‘humps’ and partially inverted signals? How can we recognise some fundamental patterns in order to differentiate between, for example, an early membrane degradation and necrosis? In addition, questions always arise about the proportion of lipid and lactate in an overlapping pattern. This article offers some hints to curve fitting of the lipid-lactate range and to avoid incorrect or misleading labelling of the peaks of an MR spectrum. Background The lipids are a large and diverse group of naturally occurring molecules with various functions, from storing energy to being components of membranes. They include miscellaneous subgroups, such as triglycerides of subcutaneous fat or bone marrow and glycerophospholipids of membranes (Fig. 1). They have in common fatty acids comprising four different proton groups: olefinic protons at 5.3 ppm, allylic protons and protons adjacent to the carboxyl group at 2.0 ppm, aliphatic methylene groups (main peak) at 1.3 ppm, and the terminal methyl group at 0.9 ppm. However, their T2 values differ according to whether the fatty acids are part of triglycerides or cell membranes. Unlike glycerophospholipids, fatty acids in membranes are embedded in the interior of the membrane resulting in efficient spin-spin interactions, thus revealing short T2 Line curve fitting in the NUMARIS software Within the NUMARIS software, line curve fitting is done in the frequency domain [2]. It is the last of the post-processing steps. Provided that the phase correction is optimal, the adequate line curve fitting protocol has to be selected out of a set of three ‘customised’ protocols, namely: 1. lipid signal only, e.g. TE 30_lip 2. lactate signal only, e.g. TE 30_lac 3. lactate and lipid signals e.g. TE 30_lip_lac They are easily derived from the Siemens protocol (CSI or SVS) TE 30 using the interactive post-processing environment and adding new peak parameters and peak restrictions (see User Manual). The lipid-containing protocols fit both the lipid_1.3 and the lipid_0.9 peaks with parameters as shown in figure 3. An analogous set of protocols shall be customised for TE 135. How do we identify lipids and lactate? Lipid-only peak (Fig. 3A): Here, the assignment method is straightforward. Firstly, we look at the line width of the peaks. Their full width at half maximum is reciprocally proportional to T2. As a rule of thumb, T2 of lipids is shorter than that of lactate, and thus the lipid peak is broader. Secondly, one can also look at the symmetry of the peak (whether it has a symmetrical Gaussian / Lorentzian shape). Thirdly, lipids should appear as two peaks. Their relative intensities may vary depending on the stage of degradation (Fig. 4). Lactate-only peak (Fig. 3A): The scalar coupling constant JIS and phase dependency of the lactate signal, S ~ [cos (JIS TE)] can be used as a kind of lactate editing: a doublet signal of 7 Hz and a 180 degree phase shift at TE 135 ms. In contrast to the lipid peak, the lactate peak reveals a sharp doublet pattern, or at least it appears foreshadowed. H1 H H2 H H H H3 H4 H 1 (1A) Triglycerides, (1B) Phospholipid structures. Note that in MR terminology, peaks arising from methyl-/methylene protons of fatty acids are referred to as ‘lipid peaks’. Schematic MR spectrum of normal brain tissue. Myo-inositol, choline, and lipid signals are ‘iceberg-like’ as most signal is invisible due to short T2. In the event of hydrolysis of inositol phospholipids or high cellular membrane turnover, myo-inositol, choline, and lipids become more ‘MR visible’ due to change in T2 towards higher values. 2 values. As such, even in short TE spectra, signals from intact membranes are hardly visible, whereas the freely tumbling fatty acids in triglycerides and also the less restricted ‘fragments’ of fatty acids from membrane degradation produce strong signals due to a longer T2. In drawing a comparison with an iceberg, protons of membranes namely those of methyl-, methylene-groups, choline, and myo-inositol, are MR-invisible unless they ‘surface’ due to degradation processes (Fig. 2). For example, depending on the grade of degradation of membranes in brain tumours, the MR signals of the said membrane fragments are raised in a characteristic way [1]. Lipid-lactate peak (Fig. 3A): In the case of lipid-lactate overlapping, this approach of Lorentzian-Gaussian curve fitting will be inaccurate in differentiating between lipid and lactate proportions, because the best fit is driven by the method of least squares rather than taking 2 Schematic diagram of the peak pattern in the lipid-lactate range. (3A) Broad lipid peak due to short T2 (blue), sharp lactate doublet due to J coupling and long T2 (orange), superposition of the lipid and lactate peak (black) for short TE. Depending on slight resonance offsets and relative intensities, often a ‘hump’ is visible (arrow). (3B) Superposition of lipid and lactate peaks for long TE (135 ms). 3 pre-knowledge of different shapes into account. In fact the operator will identify asymmetric patterns (showing a hump on one side) or if they partially invert on TE 135 spectra (Fig. 3B). Representative cases are shown in figures 5 and 6. 1B 3A 3B 1A O Fatty acid Glycerol Fatty acid O O O O C O C H H H H H H H H H H H Choline Choline fatty acid fatty acid fatty acid fatty acid Choline Choline fatty acid fatty acid fatty acid fatty acid Choline Choline fatty acid fatty acid fatty acid fatty acid myo-Ins myo-Ins fatty acid fatty acid fatty acid fatty acid myo-Ins myo-Ins fatty acid fatty acid fatty acid fatty acid myo-Ins myo-Ins fatty acid fatty acid fatty acid fatty acid INS CHO CRE NAA LAC LIP MR visible @TE 135 MR invisible @TE 135 58 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 59
- 31. How-I-do-it Lipid I:19.3 Lac I: 6.4 Illustration of a representative case of lipid-lactate overlap. (5A) Metabolite maps. In (5B), the lactate doublet peak dominates the pattern which also shows an unequal doublet with higher signal and broadening of the right peak of the doublet. Therefore, the TE_30_lip_lac protocol has been used. Due to T2 relaxation, the lipid component became negligible at TE 135, and only the lactate has been labelled with the TE_135_lac protocol in (5C). In 5D–F, lipids overwhelm the pattern. Since lactate is neither detected on TE 135, nor a ‘hump’ is observable on TE 30, the protocol TE_30_lip_lac is inappropriate. C C H n Lipid_1.3 5 Line curve fitting in the syngo.via software. Lipid-lactate compositions are slightly different in (7A) and (7B). 7 Line curve fitting in the NUMARIS software. (4A) Entire spectrum, (4B) lactate component, (4C) lipid _1.3 component, (4D) lipid_0.9 component. Note that the proportions of lipid and lactate should be regarded with caution. 4 Line curve fitting in syngo.via In syngo.via the approach of curve fitting in the time domain has been chosen [3]. It is based on PRISMA [4] using the basis set of metabolic time signals of brain metabolites together with published values of chemical shifts and coupling constants. The delineation of a lactate-lipid overlap is based on free induction decays (FIDs), in which the distinct T2 values make the segmentation more accurate. Figure 7 depicts two examples: 4A 5A Spectra of a glioblastoma multiforme, (6A) from the peri-lesional area, (6B) from the solid enhancing area. Note the relative intensities of lipid at 1.3 ppm and 0.9 ppm as in an early stage of membrane degradation (6A) these signals do not reflect a proton density even at TE 30, rather they are T2-weighted, whilst in a more advanced stage (6B) T2 becomes long enough for both signals resulting in a proton density spectrum (6C). 6 6A 6B TE 30 TE 30_lip_lac TE 135 TE 135_lac TE 30 TE 30_lip TE 30 TE 30_lip_lac TE 135 TE 135_lip Lactate area Lipid area a) with a clearly visible hump at half maximum of the Lip_1.3 peak, and b) with only an adumbrated hump (arrow). It demonstrates that delineating highly non-uniform signal compositions is possible with only one protocol for all cases. The protocol can be easily linked together by selecting the appropriate lactate and lipid templates as shown in figure 8. 7A 7B 4B 4C 4D Lactate map Lipid map 5B 5C 5D 5E 5F 6C Methylene – Group @ 1.3ppm Methyl – Group @ 0.9ppm H H H H I: 29.34 Lac I: 1.19 Lipid_1.3 I: 18.46 Lac I: 4.84 How-I-do-it 60 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 61
- 32. How-I-do-it 8 Lipid-lactateprotocol in syngo.via. References 1 Li X, Vigneron DB, Cha S, Graves EE, Crawford F,Chang SM, Nelson SJ. 2005. Relationship of MR-Derived Lactate, Mobile Lipids, and Relative Blood Volume for Gliomas in Vivo. AJNR Am J Neuroradiol 26:760–769. 2 Mierisová Š, Ala-Korpela M.2001.MR spectroscopy quantitation: a review of frequency domain methods. NMR in Biomedicine;14, 247-259. 3 Vanhamme L, Sundin T, Van Hecke P, Van Huffel S. 2001. MR spectroscopy quantitation: a review of time-domain methods NMR in Biomedicine; 14,233–246. 4 http://elib.suub.uni-bremen.de/ publications/diss/html/E-Diss1066_HTML. html Contact Helmut Rumpel, Ph.D. Department of Diagnostic Radiology Singapore General Hospital [email protected] Conclusion Carefully labelled spectra in the lactate– lipid region are a prerequisite to sending them to a PACS system as otherwise they can be misleading for further examinations and treatment planning by clinicians. Curve fitting based on chemical shift assignments should be made with caution. The technologist is required to identify ‘what is what’, as erroneous lipid-lactate discrimination is inherent to frequency domain curve fitting of overlapping peaks. By following basic steps, the lactate-lipid overlap within the region from 0.9 to 1.3 ppm can be reliably delineated. However, the proportions of lipid and lactate remain ambiguous as various sets of model peaks, i.e. half-width, signal intensity and chemical shift, may lead to the same result of curve fitting. Incorporation of prior knowledge such as supportive model spectra automates the curve fitting of a lactate- lipid overlap. The protocol TE_30_lip_lac can be made standard practice. 8 62 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world Head-to-Toe Imaging Clinical Relevant clinical information at your fingertips From technology to clinical applications, you will find all the latest news on Siemens MR at www.siemens.com/magnetom-world Don’t miss the talks of international experts on all aspects of Magnetic Resonance Imaging. Go to Clinical Corner > Clinical Talks The centerpiece of the MAGNETOM World Internet platform consists of our users’ clinical results. Here you will find case reports and clinical methods. Go to Clinical Corner > Case Studies Just a mouse click away you will find application videos and useful tips allowing you to optimize your daily MR examinations. Go to Clinical Corner > Application Tips Liver Imaging Today Tobias Heye, M.D.1; Mustafa R. Bashir, M.D.² 1Department of Radiology, University Hospital Basel, Switzerland 2Department of Radiology, Duke University Medical Center, NC, USA Introduction Liver disease is a global burden with a growing incidence and prevalence. The World Health Organization recently esti-mated that there are 800,000 cirrhosis-related deaths per year world-wide [1]. Chronic liver disease has a great impact on public health care costs with thera-peutic options ranging from antiviral treatment for viral hepatitis to orthotopic liver transplant for end stage cirrhosis. A variety of pathogens, which can be toxic, viral, metabolic or autoimmune in nature, can induce fibrosis which may progress to cirrhosis if the disease is not detected and treated. An estimated 150 million people world-wide are chronically infected with hepatitis C virus, approximately 350,000 people die due to hepatitis C related liver disease [2]. Liver fibrosis may be revers-ible at an early stage, which indicates the 1A 1B 1089_Flash_52_Inhalt_CC.indd 111 02.04.13 10:45 How-I-do-it Abdominal Imaging Clinical importance of screening and detection of liver disease. Many forms of liver fibrosis and cirrhosis especially secondary to viral hepatitis increase the risk for the devel-opment of liver cancer, namely hepato-cellular carcinoma. Non-alcoholic steatohepatatis is emerg-ing as a major pathway into chronic liver disease and is closely related to other metabolic disease entities such as diabe-tes and morbid obesity. The incidence and prevalence of these diseases has risen steadily over recent years. In a clinical context, liver disease is often reflected by a combination of several contributing factors, fibrosis, hepatic steatosis and iron overload, each with different forms of manifestations. Although these diseases are considered ‘diffuse’, actual hepatic parenchymal involvement by any of these can be irregular and patchy, leaving other parenchymal areas unaffected. Clinical management of patients with diffuse liver disease requires tools to accurately detect and classify the various forms of liver disease. Even with decades of experience in imaging, liver biopsy and the histological workup of the specimens have traditionally been the reference standard in the characterization of liver disease [3]. However, biopsy is prone to sampling errors if less affected paren-chyma is sampled and may not reflect the true disease severity and distribution in a particular organ due to the variance in the heterogeneous pattern of histological changes on a macroscopic scale [4, 5]. Biopsy, associated with the risks of an invasive procedure, is employed for dis-ease detection and staging, but periodi-cally repeated biopsy is not a practical 1 Results of the Screening Dixon tech-nique which produces color coded maps to visualize the distribution of detected abnormal metabolites in two dif-ferent clinical examples. (1A) A patient with dif-fuse hepatic steatosis as indicated by the yellow hue. (1B) A patient with diffuse iron overload as marked by blue overlay to the affected liver. MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world 111 Faster Abdominal MRI Examinations by Limiting Table Movement Mustafa Rifaat Bashir1; Brian Marshall Dale2; Wilhelm Horger3; Daniel Tobias Boll1; Elmar Max Merkle1 1 Radiology, Duke University Medical Center, Durham, NC, USA 2 Siemens Healthcare, Cary, NC, USA 3 Siemens Healthcare, Erlangen, Germany 1 How to create a minimized shimming protocol. Modifications will be made to pulse sequences following the initial localizers (1A). For the second sequence in the scan protocol, ‘Shim mode’ is set to ‘Standard’, and ‘Adjust with body coil’ is selected (1B). For the third and subsequent se-quences, ‘Positioning mode’ is set to ‘FIX’ (1C). ‘Shim mode’ is set to ‘Standard’, ‘Adjust with body coil’ is selected, and ‘Adjustment Tolerance’ is set to ‘Maximum’ (1D). Finally, for the third sequence, a reference is created to the table position of the second sequence (1E). The same reference is created for all subsequent sequences (1F). 1A 1D 118 MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world 1089_Flash_52_Inhalt_CC.indd 118 02.04.13 10:45 Clinical Abdominal Imaging 6 configurable delay times 6 Example of the Abdomen Dot Engine user interface showing the guidance view that allows global planning of delay times within a dynamic contrast enhanced liver MRI protocol. ity. Additionally, multiple scan types which differ by only a few minor compo-nents (e.g., with or without MR Cholan-giopancreatography (MRCP), with or without diffusion-weighted imaging) can be combined into a single, efficient protocol with a few key decision points, reducing redundancy and allowing for simpler base protocol maintenance and modification when necessary. Summary MRI examinations face serious competi-tion compared to sonography and CT when categories such as robustness, acquisition time, patient comfort and health care costs are considered. An abundance of information may be acquired through high resolution imag-ing and dedicated quantitative MRI sequences, but images and measure-ments should be reproducible and reli-able in their diagnostic value. The redun-dancy of preparatory steps for the operator within an MRI protocol is an opportunity for more efficient and less time consuming imaging. In addition, the image acquisition process can be improved by means of faster imaging at higher resolution with the implementa-tion of new parallel imaging acceleration techniques, to reduce the risk of motion 116 MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world 1089_Flash_52_Inhalt_CC.indd 116 02.04.13 10:45 How-I-do-it Methods Automated algorithms to minimize table movement have already been incorpo-rated into MAGNETOM MRI systems under syngo MR D11 and later software versions. Under earlier software versions, a few simple steps can be performed to convert a standard MRI protocol into a minimized shimming protocol, in order to realize the time savings previously described. These changes can all be made via the Exam Explorer (Fig. 1A). Pulse Sequence #1 – localizer The first pulse sequence of an examination is a localizer, typically utilizing either a three-plane TrueFISP or HASTE technique. At the MRI console, under the ‘Sequence’ card, the Shim is set to ‘None’ (typically the default value), and precalibrated prescan data is used with no need to acquire new prescan data. No additional modification of this sequence is required. Pulse Sequence #2 – first and only table move Using the image data from the localizer sequence, the image volume for Pulse Sequence #2 is prescribed. This volume should be centered on the area of interest and rather large, covering the volume of interest for the entire examination; at our in patients with limited breath-hold capa-bilities. which self-optimize during the course of the examination or use initial pulse sequences to tailor subsequent sequence selection, can provide faster and more efficient examinations, which include quantitative data when appro-priate. improvements may equip liver MRI examinations with sufficient tools to remain unique in delivering disease spe-cific their diagnostic value. 2B 2E 2D institution, we typically use a coronal HASTE sequence for this purpose. The prescan data acquired in this step, includ-ing shim data, will be carried forward for the remainder of the examination. The following modifications are made to this pulse sequence: 1. In the ‘System’ card, under the ‘Adjust-ments’ tab, set ‘Shim mode’ to ‘Stan-dard’. Check the ‘Adjust with body coil’ box (Fig. 1B). Pulse Sequences #3 and higher – no further table movements For all subsequent pulse sequences, table movement is disallowed, and prescan adjustment data from Pulse Sequence #2 is carried forward, so that as little time as possible is spent acquiring new adjust-ment data. The following modifications are made: 1. In the ‘System’ card, under the ‘Miscel-laneous’ tab, set ‘Position mode’ to ‘FIX’ (Fig. 1C). 2. In the ‘System’ card, under the ‘Adjust-ments’ tab: set ‘Shim mode’ to ‘Stan-dard’; select ‘Adjust with body coil’; and set ‘Adjustment Tolerance’ to ‘Maxi-mum’ (Fig. 1D). 3. From the Exam Explorer, right-click on the sequence and select ‘Properties’. 120 MAGNETOM Flash · 2/2013 · www.siemens.com/magnetom-world Intelligent imaging protocols, Combining all of the described quantitative data while expanding Under the ‘Copy References’ tab, check the ‘Copy reference is active’ box, then select Pulse Sequence #2 in the left-hand window and ‘Table position’ in the right-hand window (Fig. 1E). In combination with step #2 above, this ensures that the MRI system table will not move when progressing to later pulse sequences in the examination, despite different prescriptions for the imaging volume. 4. Repeat steps 1–4 for all subsequent sequences (Figure 1F). Discussion Preparatory adjustments made by an MRI system are essential to realize excellent image quality. In particular, adequate shimming is necessary to ensure magnetic field homogeneity. Shimming is a process whereby the main magnetic field (B0) is fine-tuned to compensate for field fluctu-ations and inhomogeneities introduced by the presence of the human body within the scanner. These adjustments are applied specifically to a volume within the bore of the magnet (based on the anticipated imaging volume), attempting to optimize magnetic field homogeneity within that volume while sacrificing field homogene-ity outside of the volume. 2A 2C 1089_Flash_52_Inhalt_CC.indd 120 02.04.13 10:46 For the whole range of clinical MR information visit us at www.siemens.com/magnetom-world
- 33. Clinical Neurology NeurologyClinical T1-weighted Phase Sensitive Inversion Recovery for Imaging Multiple Sclerosis Lesions in the Cervical Spinal Cord MS lesions in the brain. The FLAIR technique is a T2-weighted sequence with a long TR and TE and is used to demonstrate the changes in T2 relaxation times in lesions when compared to normal tissue. As the name indicates, the signal of cerebro-spinal fluid (CSF) is attenuated. CSF has a long T1 relaxation times compared to the other tissues in both the brain and cervical spine. Therefore, a rather long inversion time is needed to null the signal of CSF (~ 2500 ms). Hence, the contrast in T2-weighted FLAIR images allows for easier assessment of (MS) lesions, especially when the lesions are close to CSF, as compared to normal T2-weighted images. However, while the FLAIR technique works well in the brain, it is hampered by flow and motion artifacts when used in the cervical spine. Double Inversion recovery The Double Inversion Recovery technique has been implemented in the SPACE-DIR sequence in the Siemens syngo MR D13 software. A protocol optimized for brain imaging is also provided. SPACE-DIR is a T2-weighted technique which uses two inversion pulses, combined with a fat saturation pulse, to null both the signal of CSF and normal white matter. Similar to the FLAIR technique, this sequence is used to exploit the changes in T2 relaxation times in lesions when compared to normal tissue. In the brain, SPACE-DIR improves visualization of Bart Schraa, MSc., Senior MR Application Specialist Inversion Recovery Sequences used for imaging Multiple Sclerosis Several inversion recovery techniques are used for imaging lesions in MS. Among these are Fluid Attenuated Inversion Recovery (FLAIR), Sampling Perfection with Application optimized Contrasts using different flip-angle Evolutions Double Inversion Recovery (SPACE-DIR), and T1-weighted Phase Sensitive Inversion Recovery (PSIR). Fluid Attenuated Inversion Recovery FLAIR is commonly used to assess white matter lesions and in particular Siemens LTD Canada Introduction Multiple sclerosis (MS) is an inflammatory disease in which the insulating covers of nerve cells in the brain and spinal cord are damaged. Magnetic resonance imaging (MRI) was first used to visualize multiple sclerosis (MS) in the upper cervical spine in late 1980 [1]. Spinal MS is often associated with concomitant brain lesions; however, as many as 20% of patients with spinal lesions do not have intracranial plaques [2]. This article describes the experiences with a T1-weighted phase sensitive inversion recovery sequence for the detection of MS lesions in the cervical spinal cord using the MAGNETOM Skyra with syngo MR D13A software. 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Magnitude Reconstruction 110 160 210 260 310 360 410 460 510 560 610 Normal Cord Lesion Inversion time Relative signal intensity the cortex and reveals cortical lesions as hyperintense relative to normal surrounding gray matter. It also provides a high contrast between white matter lesions and the surrounding normal white matter. Initial studies have investigated the applicability of DIR for lesion imaging in the spinal cord with positive results [3]. Nevertheless, while SPACE-DIR provides a high contrast and isotropic voxels, its rather long acquisition time (~8 min) may prove challenging within a clinical setting. 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 T1-weighted phase sensitive inversion recovery A promising potential alternative for imaging MS lesions in the cervical spinal cord [4], is the T1-weighted true or phase sensitive inversion recovery (PSIR) sequence. This technique has been used to detect MS lesions both in white and cortical gray matter in the brain [5, 6]. This sequence exploits the differences in T1 relaxation times of tissues rather than the differences in T2 relaxation times as for both FLAIR and SPACE-DIR. Contrast Since the inversion time used is chosen such that it nulls the signal of normal white matter (~350–400 ms @ 3T), normal white matter is displayed as intermediate gray. All other tissues will have either lower or higher signal intensity than normal white matter depending on their T1 relaxation time relative to normal white matter. This provides a high contrast between MS lesions and surrounding tissue. Moreover, because PSIR uses a short TE, it is less sensitive to flow artifacts. High resolution imaging can also be achieved within reasonable scan times. Based on these advantages, T1-weighted PSIR is now being explored for the detection of MS lesions in the cervical spinal cord. Reconstruction methods The T1-weighted PSIR images can be reconstructed as a magnitude or a phase sensitive (real) image (Fig. 1). Magnitude reconstruction The magnitude reconstruction does not consider the sign of the signal. Therefore, the tissue which is nulled by the inversion time will have a signal intensity of zero and all other tissues will have higher signal intensity (ranging from 0 to +4096), regardless of whether they have shorter or longer T1 relaxation time than the nulled tissue (Fig. 2). However, there is a range of inversion times where the contrast between two different Signal behavior in an inversion recovery sequence using magnitude reconstruction. 2 Parameter Card for choosing magnitude or phase sensitive (real) reconstruction method in an Inversion Recovery sequence. 1 1 2 Contrast behavior in an inversion recovery sequence using magnitude or phase sensitive (real) reconstruction. 3 4 T1-weighted PSIR images using (4A) magnitude and (4B) phase sensitive (real) reconstructions. 4A 4B Magnitude Real Inversion time Relative signal intensity 110 160 210 260 310 360 410 460 510 560 610 3 64 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 65
- 34. Clinical Neurology NeurologyClinical tissues, e.g., lesion and normal tissue, can be decreased or even disappear. This range depends on T1 relaxation times of the two tissues and range between the two inversion times that would null one or the other tissue. In the example shown in figure 3, it ranges from approximately 390 to 430 ms. An example of the magnitude image is shown in figure 4A. Phase sensitive reconstruction In the phase sensitive reconstruction, the sign of the signal is taken in account for the reconstruction of the image (Fig. 5). As opposed to the magnitude reconstruction where the signal intensity in the image ranges from 0 to +4096, for the phase sensitive reconstruction it ranges from -4096 to +4096. This results in an image where the tissue which is nulled by the inversion time will be displayed as intermediate gray and all other tissues will have a lower or higher signal intensity depending on their T1 relaxation times relative to the T1 relaxation time of the nulled tissue. Tissues with a shorter T1 relaxation time will have a higher signal (e.g. fat), whereas tissues with a longer T1 relaxation time will have lower signal (e.g. CSF). Unlike the magnitude reconstruction, the contrast between tissues remains largely preserved independent of the chosen inversion time. Since the T1 relaxation time of lesions might vary from patient to patient and even from lesion to lesion, the phase sensitive reconstruction should be used to reconstruct the images. An example of the phase sensitive reconstruction is shown in figure 4B. Clinical Cases Case 1 Patient with a MS lesion at the level of C6 (Fig. 6). The lesion is difficult to see on the T2- and PD-weighted images. However, the MS lesion can be clearly seen in the T1-weighted PSIR image. Case 2 Patient with diffuse MS lesions in the spinal cord from level C3 to C6 (Fig. 7). The lesions are hardly visible on the T2- and PD-weighted images, whereas the T1-weighted PSIR shows the lesions more clearly. Case 3 Patient with a known MS lesion at the level of C3-C4 (Figs. 8 A–C) and C7-T1 (Figs. 8 D–F). The lesion at the level of C3-C4 can hardly be seen on the T2-weighted image. Both the PD- and the T1-weighted PSIR show this lesion clearly. While the lesion at the level of C7-T1 is poorly visible on the T2- and PD-weighted images, the T1-weighted PSIR shows it very clearly. Imaging Parameters The parameters for the sequences used in the clinical cases are listed in table 1. Conclusion The T1-weighted PSIR shows great potential in revealing MS lesions in the cervical spinal cord. While using this technique it is important to use the phase sensitive reconstruction to preserve the contrast between MS lesions and normal appearing tissue. Due of the nature of the reconstruction, and because T1 values of lesions can vary from patient to patient, for reliable depiction of lesions, the phase sensitive reconstruction is recommended. This is as, unlike the magnitude reconstruction, the phase sensitive reconstruction provides a contrast between different tissues that is largely independent of the chosen inversion time. Acknowledgements We acknowledge the invaluable support of Dr. Montanera, Dr. Alcaide Leon and Mrs. Karima Murji of St. Michael’s Hospital (Toronto, Canada) for providing the clinical cases and their feedback. 5 8A 8B 8C 8D 8E 8F Table 1: Imaging parameters for the sequences used in the clinical cases. t2_tse_sag_384 pd_tse_sag_p2 t1_tir_sag_ms TR 3500.0 ms 2500.0 ms 2400.0 ms TE 106.0 ms 23 ms 9.4 ms TI 400 ms Slices 15 15 15 Slice thickness 3.0 mm 3.0 mm 3.0 mm FOV Read 220 mm 220 mm 220 mm FOV Phase 100.0% 100.0% 100.0% Magn. preparation None None Slice-sel. IR Base resolution 384 320 320 5 Signal behavior in an inversion recovery sequence using phase sensitive (real) reconstruction. 6A 6B 6C 7A 7B 7C T2- (7A), PD- (7B) and T1-weighted PSIR (7C) images of a patient with known diffuse MS lesions at the level of C3–C6 (Case 2). 7 T2- (6A), PD- (6B) and T1-weighted (6C) PSIR images of a patient with a known MS lesion at the level of C6 (Case 1). 6 T2- (8A, D), PD- (8B, E) and T1-weighted PSIR (8C, F) images of a patient with known MS lesions at the level of C3–C4 (top row) and C7-T1 (bottom row) (Case 3). 8 Relative signal intensity 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 Real Reconstruction 110 160 210 260 310 360 410 460 510 560 610 Normal Cord Lesion Inversion time 66 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 67
- 35. Clinical Neurology Whole-BodyImaging Clinical References 1 Honig LS, Sheremata WA. Magnetic resonance imaging of spinal cord lesions in multiple sclerosis. J Neurol Neurosurg Psychiatry. Apr 1989;52(4):459-466. 2 Noseworthy JH, Lucchinetti C, Rodriguez M, et al. Multiple sclerosis. N Engl J Med. Sep 28 2000;343(13):938-452. 3 Shipp D. Case Report: Cervical Spine 3D Double Inversion Recovery (DIR) in Demyelination. MAGNETOM FLASH magazine 1/2012 ISMRM Edition: 49-50 4 Poonawalla AH, Hou P, Nelson FA, Wolinsky JS, Narayana PA. Cervical Spinal Cord Lesions in Multiple Sclerosis: T1-weighted Inversion-Recovery MR Imaging with Phase-Sensitive Reconstruction. Radiology. 2008 Jan; 246(1): 258-264. 5 Hou P, Hasan KM, Sitton CW, Wolinsky JS, Narayana PA. Phase-sensitive T1 inversion recovery imaging: a time-efficient interleaved technique for improved tissue contrast in neuroimaging. AJNR Am J Neuroradiol 2005;26:1432–1438. 6 Nelson F, Poonawalla AH, Hou P, Huang F, Wolinsky JS, Narayana P. Improved visualization of intracortical lesions in multiple sclerosis by phase-sensitive inversion recovery in combination with fast double inversion recovery MR imaging. Presented at the 22nd Congress of the European Committee for the Treatment and Research in Multiple Sclerosis, Madrid, September 27–30, 2006; 639. Download Visit us at www.siemens.com/ MAGNETOM-world to download the .edx file for 3T MAGNETOM Skyra Contact Bart Schraa, MSc. Siemens Canada Ltd. NAM RC-CA H CX-CS APP 1577 North Service Road East L6H 0H6 Oakville ON Canada Phone: +1 (416) 818 6795 [email protected] Save the Date Heidelberg Summer School Musculoskeletal Cross Sectional Imaging 2014 July 25th / 26th 2014 Heidelberg, Germany MAGNETOM Flash · 1/2012 · www.siemens.com/magnetom-world 69 The Heidelberg Summer School offers advanced learning opportunities and promotes the academic exchange of knowledge, ideas, and experiences by bringing together physicians and professional staff from all over the world. Excellent speakers will cover a wide range of medical, physical, and technical topics in musculoskeletal imaging. All lectures are in English. Course director Marc-André Weber, M.D., M.Sc. Professor of Radiology, Section Head Musculoskeletal Radiology at the University Hospital Heidelberg CME Accreditation The symposium will be accredited by the ‘Landesärztekammer Baden-Württemberg’ with CME credits (category A). Also, the symposium is accredited for 1 category 3 credit point for the ESSR diploma by the European Society of Musculoskeletal Radiology. Registration Mrs. Marianne Krebs, Secretary of the Section Musculoskeletal Radiology [email protected] For further information please visit: www.heidelbergsummerschool.de Expert Talks Don’t miss the talks of experienced and renowned experts covering a broad range of MR imaging Highest quality imaging in an optimized clinical workflow Johan Dehem, M.D. VZW Jan Yperman, Ieper, Belgium MR/PET and radiology as information business Dieter Enzman, M.D. University of California Los Angeles, Los Angeles, CA, USA Visit us at www.siemens.com/magnetom-world Go to Clinical Corner > Clinical Talks
- 36. Pictorial Essay Benignand Malignant Bone Tumors: Radiological Diagnosis and Imaging Features Katharina Grünberg, M.D.; Christoph Rehnitz, M.D.; Marc-André Weber, M.D., M.Sc. Section Musculoskeletal Radiology, Diagnostic and Interventional Radiology, University Hospital Heidelberg, Germany Topics The learning objectives of this review article are to identify benign vs. malig-nant criteria in bone tumor diagnosis and also to differentiate the types of bone tumors and their characteriza-tion. Based on the Lodwick classifica-tion an overview of the three main types of bone destruction patterns visible on radiographs will be given with many examples. Typical examples of benign and malignant bone tumors will be demonstrated, the various imaging modalities will be compared, and their utility will be dis-cussed. The image gallery comprises pearls and pitfalls. Presentation of standardized magnetic resonance imaging (MRI) protocols will be given. Of course, this pictorial essay does not have the focus of comprehensively presenting all bone tumor entities. Introduction Primary bone tumors are categorized according to their tissue of origin into cartilage, osteogenic, fibrogenic, fibrohistiocytic, haematopoietic, vas-cular, lipogenic tumors and several other tumors, like Ewing sarcoma and giant cell tumor [1]. Thy are also classified as either benign, malignant or semi-malignant, as well as tumor-like lesions [2]. They are rare, but found on radiographs during an investigation of a painful skeletal region or incidentally, e.g. when performing a joint or whole-body MRI. You will need four diagnostic columns to make a diagnosis of a bone tumor. 1. Malignant vs. benign? X-rays: Aggressiveness: Analysis of growth rate (Lodwick classification), periosteal reaction? Further imaging modality CT, MRI? Make a specific diagnosis: 2. Analysis of tumor matrix: X-rays, CT: Osteolytic, osteoblastic, mixed 3. Location within the tumor-bearing bone: Epi-, meta-, diaphysis 4. Patient’s age, (affected bone) in 80% correct specific diagnosis [4] Four diagnostic columns Four diagnostic columns (Fig. 1) Tumor’s aggressiveness The radiograph is the first method to distinguish benign from malignant lesions: at first by analysing the aggres-siveness (analysis of growth rate) of a lesion according to the classification of Lodwick [3]. In radiographs there is a correlation between bone tumor‘s growth rate and dignity. If you identify an aggressive growth pattern and/or malignant periosteal reaction another imaging modality like computed tomography (CT) or magnetic reso-nance imaging (MRI) is needed. MRI is important for defining the extension of tumor before biopsy. Tumor matrix In a second step, it is essential to analyze the mineralisation of tumor matrix in radiographs or CT. The matrix may be osteolytic, osteoblastic, or mixed, i.e. osteolytic with matrix mineralisation. Lodwick classification (Fig. 2) Based on the Lodwick classification, an overview of the three main types of bone destruction patterns visible on radiographs are given with a repre-sentative example: Type 1: geographic (with a: well-defined border with sclerotic rim, b: well-defined and sharp border but without sclerotic rim, c: ill-defined and blurred border); Type 2: geographic with moth-eaten or permeated pattern (patchy lysis); Type 3: small, patchy, ill-defined areas of lytic bone destruction with moth-eaten or permeated pattern (patchy lucencies) [3, 6]. Lodwick classification IA IB IC II III Non-ossifying fibroma Aneurysmal bone cyst Giant cell tumor Ewing’s sarcoma Osteosarcoma Lodwick classification: An o 2 verview of the three main types of bone destruction patterns with representative image examples. 1 Tumor’s aggressiveness Tumor matrix Tumor location Patient’s age 1 The four diagnostic columns needed to achieve a correct, specific diagnosis in about 80% of cases [4]. Tumor location and patient age To make a specific tumor diagnosis, the location within the tumor-bear-ing bone (epi-, meta- and diaphysis) and the patient’s age are also impor-tant. With an optimized combination of the different parameters, the expert achieves a correct, specific diagnosis in about 80% of cases [4, 5]. In other words, even a dedicated musculo-skeletal radiologist fails to predict the correct histological diagnosis in one fifth of all cases. 2 Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 71 Clinical Orthopedic Imaging 70 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world Geographic, well-defined & sclerotic rim Geographic well-defined & sharp border but without sclerotic rim Geographic but blurred border Geographic & moth-eaten damage with patchy lysis Permeated lytic damage with small patchy lucencies
- 37. Criteria of malignancy(Fig. 3) Periosteal reactions are also indicators of lesion aggressiveness and can be differentiated according to a benign (thick, dense, wavy) type or an aggressive (lamellated, amorphous, sunburst) type. Figures 3A–D show an example of an 80-year-old man with a cloudy inhomogeneous tumor of the distal humerus with perpendicular periosteal reaction of a malignant sunburst type, partial cortical destruction and a big soft tissue component, best seen in MRI. All these criteria suggest a malignant process. Differential diagnoses are osteosarcoma or bone metastasis. Biopsy results in the diagnosis of metastasis of rectal cancer (adenocarcinoma). Types of bone tumors According to their type of matrix (osteolytic, osteoblastic, or osteolytic with matrix mineralization) and to their tissue of origin, bone tumors are categorized into different types: osteoid, chondroid, fibrous, lipoid/ fatty, other, cystic (solitary bone cyst, aneurysmal bone cyst), vascular (hemangioma), special cell type: Giant cell (osteoclastoma), small cell (Ewing’s sarcoma), histiocytes (eosinophilic granuloma), plasma cells (multiple myeloma), notochordal cells (chordoma) and metastases. Osteoid type Osteoid osteoma and Osteoblastoma (Fig. 4) This entity is frequent: around 13.5% of all benign bone tumors are osteoid osteomas. The patients are usually younger than 30 years and suffer „night pain relieved by aspirin“ and other platelet aggregation inhibitors. The main location is in more than 50% within diaphysis of long bones and in 10% within the vertebral column with painful scoliosis. Osteoid osteomas show in CT and X-ray a perifocal sclerotic lesion with a central lucency (nidus) that is cortically based in 80%. Medullary, subperiosteal and articular locations also occur. Calcification of the nidus is possible. The nidus is extremely vascular in contrast-enhanced MRI and it is important to identify the nidus as the tumor itself; surrounding sclerosis and bone marrow edema pattern is just reactive. It should be noted that lesions may have less or no sclerosis if the nidus is located in the marrow or in/adjacent to a joint (Fig. 5). Osteoid osteoma resembles osteomyelitis: For example if a Brodie’s abscess is in an eccentric position, e.g. cortically located, it is difficult to differ Brodie’s abscess from osteoid osteoma. The differentiation can then only be done by biopsy or radionuclide bone scan: Osteoid osteoma shows – in contrast to osteomyelitis – the ‘double density sign’ (i.e. a high intense central activity surrounded by an area of medium activity). A lesion larger than 1.5 cm is called osteoblastoma [7, 11]. Radiofrequency ablation (RFA) is a successful treatment [8, 9, 10]. Keys to diagnosis: Sclerotic lesion with a small lucency in X-ray. The nidus shows a high signal on T2-weighted MR images and has a strong contrast-enhancement. 3 80-year-old man with a cloudy inhomogeneous tumor of the left distal humerus. (3A) Radiograph with lateral projection shows the perpendicular periosteal reaction of a malignant sunburst type of the humerus with partial cortical destruction (yellow arrow). (3B) The corresponding antero-posterior radiograph shows the partial cortical destruction and a big soft tissue component (orange arrows), best demonstrated in MRI (orange arrows). (3C) Sagittal T1w, (3D) sagittal PDw with fat saturation. 3A 3B 3C 3D 4A 4B 16-year-old male patient with osteoidblastoma (OB) of the right fibula. (4A) lateral radiograph shows well the cortical swelling of the fibula in the patient with OB but without lucency. The reason for that is best seen in 4B, C (axial CT in different positions) and 4D (coronal CT), where the upper part of the nidus is completely calcified whereas the smaller lower part shows only little calcification (*). (4E) shows a coronal CT with the ablation cannula (*) in the upper calcified part of the nidus during CT-guided radiofrequency ablation and the second ablation channel thereunder (#). T2w STIR images (4F axial and 4G coronal) as well as T1w axial images (4H, I) demonstrate the vascularisation of the lower and the calcification of the upper nidus part with low T2w signal. 4 4C 4F 4D 4G 4E 4H 4I Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 73 Clinical Orthopedic Imaging 72 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 38. 5A 5B 5C 5D 5E 5F 17-year-old male patient with articular osteoid osteoma (OO) of the left knee joint. (5A–C) Axial, sagittal and coronal CT with the articular position of the OO show no sclerosis of the nidus-margin (orange circle). In CT you see well that the nidus shows some central ossifications. The axial T2w MRI with fat saturation (5D) demonstrates the joint effusion and synovitis (yellow arrow). You can also see well that because of the central nidus calcifications the OO has only isointense to less hyperintense signal in T2w (orange circle) instead of the typical strong hyperintense signal. (5E) Sagittal PDw MRI also shows the calcification. (5F) Axial contrast-enhanced T1w MRI with fat saturation demonstrates the enhancing nidus (orange circle). Osteosarcoma The patients are usually younger than 20 years. A 2nd peak exists in the 5th decade and these cases are mostly secondary in Paget‘s disease and after irradiation. Osteosarcoma has a predilection for sites of rapid bone growth, usually the metaphyseal region. Typical symptoms are pain and local swelling. This entity shows typically destructive periosteal reactions as mentioned above (Fig. 6). Their X-ray morphology is very variable: Osteosarcomas may be osteogenic (i.e. the tumor induces new bone formation), lytic or mixed, which is the common manifestation form (Fig. 7) [12]. If such a lesion is lytic, consider also teleangiectatic osteosarcoma! From origin, sclerosis grade and soft tissue component, osteosarcomas are separated into a central, parosteal (originates from the periosteum) and a periosteal variant, which is very rare (1% of osteosarcomas). In periosteal osteosarcomas the process starts either in the periosteum or adjacent soft tissue. Typical – in contrast to parosteal osteosarcoma – the periosteal osteogenic sarcoma does not have large amounts of calcification in the soft tissue (Fig. 8) [11]. Osteosarcomas may produce osteoblastic lung metastases (Fig. 9). Keys to diagnosis are to detect criteria of malignancy in X-ray and further imaging modalities: CT is the best for identifying periosteal reaction versus tumor matrix because you can already see faint mineralization in CT. In MRI the signal depends on the degree of matrix mineralization. But MRI is important for assessing the tumor extent and for staging purposes, i.e. to identify skip-lesions, to assess the soft tissue, nerve and vessel involvement, and a potential joint infiltration. 5 6A 6B 6C 21-year-old man with a central high-grade osteosarcoma in the distal left femur. Conventional osteosarcomas are the central osteosarcomas placed in the center of the metaphysis. Figure (6A) shows the antero-posterior and (6B) the lateral radiograph. In this case you can see in addition to periosteal reactions (orange arrows in 6A) the channel-shaped lucency in the radiograph correlating with the biopsy channel (yellow arrow in 6B) within a disorganization of the bone pattern and osteoid formation (orange circle in 6B). You also see the biopsy channel in the coronal T1-weighted MRI (orange arrow in 6C). Figure 6D demonstrates the heterogeneity of the tumor mass (axial T2w MRI with fat saturation). Performing MRI is important for preoperative local staging, e.g. in this case the vessel infiltration (orange arrows in 6E) is visible in the axial post-contrast T1-weighted MRI. 6 6D 6E 7A 7B 7C These images demonstrate well the difference between osteogenic versus lytic osteosarcoma. Figures 7A (antero-posterior radiograph of the lower leg) and 7B (sagittal contrast-enhanced T1w MRI with fat saturation of the lower leg) show an osteogenic parosteal osteosarcoma of the left tibia, a bone forming tumor with fluffy, amorphous, cloudlike mineralization (orange arrow in 7A) beside sunburst periosteal reaction as a criterion of malignancy (* in 7A). This tumor has a big soft tissue component (yellow arrow in 7B) with large amounts of calcification (yellow arrow in 7A). Figures 7C (antero-posterior radiograph of the pelvis) and 7D (axial contrast-enhanced T1w MRI with fat saturation) show a more lytic osteosarcoma of a 61-year-old female patient in the right os ileum with no mineralization (orange arrow). 7 7D MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 75 Clinical Orthopedic Imaging 74 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 39. 8A 8B 8D 57-year-old female patient with a periosteal osteosarcoma (G3) of the right lower leg. Figure 8A, antero-posterior radiograph of the right knee shows saucerization of the tibial metaphysic (orange circle) and also a bone prominence (yellow arrow). 8B (coronal T1w MRI of the lower right leg) and 8C (sagittal T1w MRI of the lower right leg) show the big inhomogeneous tumor with a large soft tissue component. Keep in mind that the periosteal osteogenic sarcoma does not have large amounts of calcification in soft tissue as shown in 8A–C (orange circle). Figures 8D (axial T2w MRI with fat saturation) and 8E (axial contrast-enhanced T1w MRI with fat saturation) clearly demonstrate that the tumor inexplicably will not invade the medullary space of the tibia (orange circle) and that the fibula is not involved (yellow arrow). 63-year-old female with osteoblastic lung metastases one year after resection of an osteosarcoma of the left thigh. Axial CT images (9A–C) show several osteoblastic lung metastases of a bone producing primary tumor: an osteosarcoma. Therefore the lung metastases may be also sclerotic. 8 9 Chondroid type Enchondroma Enchondroma is a benign lytic lesion typically placed in the hand and chiefly centrally located, often with endosteal scalloping. It must have calcification except in the phalanges (Fig. 10). A typical size of enchondroma is around 1–2 cm; low grade chondrosarcoma is larger than 4–5 cm. The enchondroma shows no periosteal reaction. An important differential diagnosis is the bone infarction (Fig. 12). Keys to diagnosis are: In T1-weighted MR imaging the lesion has a low signal. The T2-weighted signal depends on the degree of calcification. After contrast-enhancement the tumor shows in T1-weighting MR imaging a lobulated appearance with septa (Figs. 10 and 11). Suspicious of malignancy in chondroid tumors are pain, a size larger than 5 cm, the presence of a soft tissue mass and a growing surrounding edema on T2-weighted images. Multiple enchondromas occur on occasion, a condition called Ollier‘s disease. This is not hereditary and with no increased rate of malignant degeneration. By contrast, Maffucci‘s syndrome is a condition with multiple enchondromas associated with soft tissue hemangiomas. Maffucci‘s syndrome is likewise not hereditary, but is characterized by an increased incidence of malignant degeneration of the enchondromas [11]. 9A 9B 8C 8E 9C 10 A 10 B 29-year-old female with an enchondroma of the phalanx D1. (10A) Lateral radiograph of the right D1 shows the lytic lesion in the proximal metacarpus of D1 which is hardly to identify and without sclerotic rim, according to a Lodwick IB lesion (orange circle) and without calcifications. (10B-F) show the typical signal characteristics of an enchondroma in MRI (orange circles) and that the lesion is smaller than 2 cm: low signal in T1-weighted imaging (10B, coronal), high signal in T2-weighted imaging because of absent calcification as shown in 10A (10E, axial T2w MRI with fat saturation). Coronal contrast-enhanced T1w MRI (10C), coronal T1w MRI subtraction (10D), and axial contrast-enhanced T1w MRI with fat saturation (10F), show that the tumor has a lobulated appearance with septa. 10 10 E 10 C 10 F 10 D Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 77 Clinical Orthopedic Imaging 76 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 40. 11 A 11B 12 A 44-year-old female with an enchondroma within the humeral head. (11A) Antero-posterior radiograph of the shoulder, (11B) coronal CT of the shoulder, (11C) coronal T2w MRI with fat saturation, (11D) coronal T1w MRI, (11E) coronal contrast-enhanced T1w MRI. 11A and B show a lesion bigger than 2 cm with a sharp border (Lodwick IB) in the humerus head with the following different forms of calcification of the chondral tissue: Punctate, comma-shaped, arc like, ring like mineralization (orange circle). In T1w the tumor shows low signal (11D), in T2w with fat saturation high signal with some low signals according to the calcifications, thus containing no fat (11C) and post-contrast a homogenous contrast-enhancement with a rough lobulated pattern (11E) (yellow arrows). 11 40-year-old female with bone infarction in the right tibia and femur. (12A) Antero-posterior, (12B) lateral radiograph of the knee, (12C) coronal contrast-enhanced T1w MRI, (12D) coronal T1w subtraction MRI, (12E) axial contrast-enhanced T1w MRI with fat saturation, (12F) axial T2w MRI with fat saturation, (12G) coronal STIR MRI. An infarct usually has a well-defined, densely sclerotic, serpiginous border as well shown in 12A and B (orange circles) and in MRI in 12C-G (yellow arrow), whereas an enchondroma does not. Fat in the lesion as seen in 12C, E and G (yellow star) is a hint of bone infarction and speaks against an enchondroma. 12 E 12 12 B 12 F 12 C 12 D 12 G 11 C 11 D 11 E Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 79 Clinical Orthopedic Imaging 78 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 41. Chondroid type Osteochondroma(Fig. 13) A synonym for osteochondroma is cartilaginous exostosis. It is a common benign tumor of the extremities (10%–15% of all bone tumors) and is located in 50% of cases in the lower extremities, in 10–20% in the humerus, but rarely in the spine. For the diagnosis it is important to identify the continuation 13 B 20-year-old male patient with an osteochondroma of the proximal right humerus and a loose body within joint space. (13A) Axial radiograph of the right shoulder shows the sharp-bordered tumor of the humerus, (13B) coronal CT shows the osteochondroma of the right shoulder, (13C) sagittal T2w MRI shows another part of the osteochondroma with its cartilage cap, where the cap seems to be much larger than 8 mm, see also 13F, (13D) axial T2w MRI with fat saturation shows the measurement of the T2w hyperintense cartilage cap with a distance of 8 mm, (13E) coronal T1w MRI shows the cartilage cap that is hypointense in T1w, (13F) axial contrast-enhanced T1w MRI with fat saturation show the same part of figure 13C with a chondroid lobated pattern after contrast-enhancement, (13G) 9 MHz ultrasound shows the structure seen in 13C and G as a round non-cystic structure. This lesion is with all imaging modalities suspicious of a low grade chondrosarcoma. After surgery and histologic examination revealed it to be no more than an osteochondroma and a neighboring loose body within joint space with caplike borders and nodose lobulated chondroid tissue with kept structure of the lobules. 13 of bone marrow and trabecular bone structures into the exostosis as well as the cartilage cap. The malignant degeneration occurs mainly in tumors near the trunk. Key to diagnosis is a mushroom-like tumor. The thickness of the cartilage cap is 8 mm or more (threshold in our institution, see also explanation in the next chapter) (Fig. 13) [13]. Contrast media is not needed to determine the thickness of the cartilage cap, because it is clearly visible on T2-weighted images. Osteochondroma vs. chondrosarcoma A malignant transformation is more likely if the cartilage cap thickness is 8 mm or more, which is the threshold of our clinic. Further publicized threshold values are 1.5 cm according to Murphey et al. [14] and 2.0 cm according to Bernard et al. [15]. Proximity to trunk (location in the pelvis with highest malignant transformation rate!) and hereditary multiple exostoses (autosomal dominant inheritance) (Fig. 14) are correlated with a higher risk of malignant transformation (3–5% of tumors develop into chondrosarcomas). A further criterion is a cartilage cap growth, especially beyond age 20. Note: It is extremely difficult for either a radiologist or a pathologist to differentiate a low-grade chondrosarcoma from enchondroma (Fig. 15). 12-year-old male patient with hereditary multiple exostoses. (14A) Axial radiograph of the left shoulder, (14B) antero-posterior radiograph of the left shoulder, (14C) axial T2w MRI with fat saturation of the left humerus, (14D) sagittal T2w MRI with fat saturation of the left humerus, (14E) lateral radiograph of the left knee, (14F) antero-posterior radiograph of the left knee, (14G) coronal STIR MRI of the upper extremities. In 14A, B, E and F the continuation of bone marrow and trabecular bone structures into the exostosis are clearly depicted. The cartilage cap can be well evaluated in T2-weighted images as seen in 14C, D and also in 14G. 14 13 A 13 D 13 E 13 F < 8 mm 13 C 13 G 14 A 14 E 14 C 14 B 14 F 14 D 14 G Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 81 Clinical Orthopedic Imaging 80 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 42. Chondrosarcoma (Figs. 15–17) Patients are mostly older than 40 years and experience pain. Tumors are near the trunk and have a chondroid matrix. Chondrosarcomas are characterized by slow growth. Primary chondrosarcomas are lytic, per-meative and destructive lesions with calcification in 50%. Secondary chondrosarcomas have a cartilage cap‘s thickness larger than 8 mm as a sign of malignant transformation of an osteochondroma (see also the comments to threshold value in the last chapter) [13-15]. Keys to diagnosis are lytic, destructive lesion with flocculent, snowflake or popcorn calcification in patients older than 40 years. MRI: soft tissue mass or edema. The following criteria are in favor of a chondrosarcoma as opposed to an enchondroma: Pain, tracer uptake in bone scan, growth, cortical bone penetration. 15 B 15 C 31-year-old male patient with grade 1 chondrosarcoma of the right os ilium. (15A) The antero- posterior radiograph of the right hip joint shows a geographic well-defined lytic lesion in the right acetabulum with a sharp border but without sclerotic rim according to a Lodwick IB lesion (orange arrow), in the center there is some flocculent calcification. (15B) Coronal CT and (15C) axial CT of the right hip show the lytic lesion with sharp border, thinned cortex and central punctuate calcification without cortical destruction. (15D, E) Contrast-enhanced T1w MRI with fat saturation (coronal in 15D and axial in 15E) show a central contrast-enhancement. Figure 15F shows a coronal STIR MRI with a high signal in the border area of the tumor. 15 Orthopedic Imaging Clinical 16 B 16 C 16 D 16 E 16 F 16G 33-year-old female with grade 2 chondrosarcoma of the left olecranon. 16A shows an antero-posterior radiograph of the left olecranon with a Lodwick type IC lesion: geographic but blurred border (orange circle). Figure 16B shows the lateral radiograph of the left olecranon and reveals a cortical destruction (orange arrow). (16C) Coronal T1-weighted MR image of the olecranon clearly shows the intraosseous borders of the tumor (orange arrows). (16D) Coronal STIR MR image clearly shows the muscle edema (orange arrow). (16E) Axial T2w MRI with fat saturation shows the chondroid matrix of the tumor (orange circle). (16F) Sagittal T2w MRI shows the hypointense calcification (orange arrow) in the center of the chondroid tumor. (16G) Sagittal contrast-enhanced T1w MRI with fat saturation shows the infiltration of the surrounding soft tissue (orange arrow). 16 58-year-old male patient with grade 3 chondrosarcoma of the right humerus. (17A) The antero-posterior radiograph of the right humerus shows in addition to a patchy lysis pattern (Lodwick II) the cortex destruction (orange circle). (17B) Coronal STIR MRI clearly shows the extension of this large amorphous lesion (size of 10 cm, long yellow arrow) and the soft tissue infiltration (small yellow arrow). The tumor has a predominant chondroid matrix with low signal in T1w (17C coronal T1w MRI) and an inhomogeneous high signal in T2w (17D sagittal T2w MRI) as a further hint of a high-grade chondrosarcoma. (17D) Also shows well the cortex destruction and soft tissue infiltration (orange circle). (17E) Coronal contrast-enhanced MRI shows necrotic tumor areas within the tumor (yellow arrows). Also areas without chondroid matrix are a hint of a high-grade chondrosarcoma. 17 15 A 15 D 15 E 15 F 16 A 17 A 17 B 17 C 17 D 17 E MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 83 Clinical Orthopedic Imaging 82 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 43. Chondroblastoma (Fig. 18) Patients are usually younger than 20 years (i. e. skeletally immature patients). The lesion must be located epiphyseally and is rare in metaphysis. This entity also occurs in carpal and tarsal bones and rarely in the patella (which with regard to the 18 C 18 D 12-year-old female patient with a chondroblastoma of the right lateral tibia epiphysis. (18A) Antero-posterior radiograph of the right knee shows the epiphyseal located lytic lesion of the tibia with discreet sclerotic rim and some central located calcification (orange circle). (18B) Axial CT shows the puncture needle in that lytic lesion having a sharp border (orange arrow). (18C) Coronal and (18D) sagittal T1-weighted MR images show the bordered lesion having a discreet hypointense sclerotic rim and a central hypointense punctual calcification (orange circle). (18E) Axial PD-weighted MR image shows a chondroid component with high signal and central calcification with low signal (orange circle). (18F) Coronal STIR MR image also shows a chondroid component with high signal and central calcification with low signal (orange circle) and a bone marrow edema in the circumference (orange arrows). Therefore biopsy was performed. 18 differential diagnosis of lytic lesions behaves like an epiphysis) [11]. Usually it appears in long bones and shows in 40–60% calcification. Differential diagnoses are the sequestrum (osteomyelitis) and the eosinophilic granuloma. Key to diagnosis: Chondroblastomas are lytic epiphyseal lesions with sclerotic rim. In MRI it shows a chondroid component with high signal in T2-weighted imaging and calcification with low signal in T2-weighted imaging [4]. Fibrous type Non-ossifying fibroma – NOF (Figs. 19 and 20) Patients are usually younger than 20 years and have no pain or periosteal reaction. This lesion is located in the metaphysis of long bone in eccentric position and emanates from the cortex, so that the cortex will be replaced with benign fibrous tissue. Non-ossifying fibromas ‘heal’ with sclerosis and disappear in the following years. Lesions smaller than 3 cm in length are called fibrous cortical defect and lesions larger than 3 cm in length are called non-ossifying fibroma. Key to diagnosis: A lytic lesion with expansive growth and scalloped, well-defined sclerotic border. The MRI appearance of an NOF is somewhat variable. Although they are essentially always low signal on T1-weighted MR imaging, they can have high or low signal on T2-weighted imaging. NOF has partly homogeneous or partly non-homogeneous contrast-media enhancement. During the ‘healing period’ the non-ossifying fibroma can be hot on radionuclide bone scans indicating the osteoblastic activity. 19 14-year-old male patient with a non-ossifying fibroma of the left distal tibia. (19A) Antero-posterior radiograph shows a classic example of a non-ossifying fibroma that is slightly expansile and lytic and has a scalloped, well-defined sclerotic border (Lodwick IA, orange circle). (19B) Coronal T1-weighted MR image shows the typical low signal of the lesion (orange circle). (19C) Axial T2-weighted MR image shows in this case a low signal in T2-weighting what is variable (orange arrow). (19D) Axial contrast-enhanced T1-weighted MR image with fat saturation shows that in this case the NOF has a partly inhomogeneous contrast-media enhancement (orange arrow). 18 A 18 B 18 E 18 F 19 A 19 B 19 C 19 D Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 85 Clinical Orthopedic Imaging 84 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 44. Fibrous dysplasia (Figs.21–23) Patients have usually no pain or periosteal reaction. Fibrous dysplasia can be either monostotic (most commonly) or polyostotic (McCune-Albright syndrome) and has a predilection for the pelvis, the proximal femur, the ribs and skull. In its classic description, fibrous dysplasia has a, ‘ground-glass appearance’ or ‘smoky appearing’ in X-ray and/or CT (Fig. 21), but the ground glass appearance is not always present. Lesions may be mixed lytic and sclerotic [11] and bone may be deformed. Keys to diagnosis are: No periosteal reaction. Fibrous dysplasia shows lytic lesions, as the matrix calcifies it has a hazy, smoky and ground-glass look to the point of sclerotic lesion. The signal alterations of fibrous dysplasia in MRI follow the uniform pattern of all tumors (low signal in T1-weighted and intermediate to high signal in T2-weighted images). The fibrous tissue enhances contrast media. If the lesion is located in the tibia, consider also adamantinoma, which has malignant potential, i.e. a mixed lytic and sclerotic lesion in anterior cortex of tibia that resembles the fibrous dysplasia. 20 A 20 C 20 D 20 E Image gallery of the non-ossifying fibroma. (20A) Antero-posterior radiograph of the right knee of a 17-year-old male patient clearly shows the various appearance of a NOF. Two healing periods can be seen: Lateral, a lesion with proceeding sclerosis indicating that the lesion is in a progressed healing stadium (orange circle) and medial, a classical scalloped lesion with well- defined sclerotic border (orange arrows). (20B) Antero-posterior radiograph of the left knee of an 18-year-old male patient shows the typical appearance of a NOF: Scalloped lesion with well-defined sclerotic border. (20C) Antero-posterior radiograph of the knee of a 12-year-old male patient shows a lytic lesion with sclerotic rim (orange arrow) and below an exostosis (yellow arrow). (20D) Coronal T1-weighted MR image shows the typical low signal of a NOF (orange circle) and (20E) a coronal STIR MR image shows in this case also a low signal compatible to the diagnosis of a NOF. 20 21 A 21 E 21 F 21 G 21 H 21 B 21 C 21 D 20 B 21 44-year-old male patient with fibrous dysplasia of the left femur. (21A) Antero-posterior radiograph of the left femur shows well the ground glass appearance of a sclerotic lesion in the proximal diaphysis (orange circle). (21B) Coronal CT, (21C) axial CT and (21D) 3D figure also clearly show the ground glass appearance of that lesion (orange circle). (21E) Coronal T1-weighted MR image shows low signal of the lesion. (21F) Axial T2-weighted MR image with fat saturation shows that the lesion contains only point-shaped lipoid and calcified parts (orange arrow). (21G) Sagittal T2-weighted MR image shows in this case a homogenous low signal, (21H) axial contrast-enhanced T1-weighted MR image with fat saturation shows a relatively homogenous contrast enhancing of the lesion. An inhomogeneous contrast-enhancement occurs in lesions with bigger parts of blood, fat and calcifications leading to signal alterations. Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 87 Clinical Orthopedic Imaging 86 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 45. 22 Image galleryof fibrous dysplasia: “Fibrous dysplasia…can look like almost anything!” [11], as is clearly visible when you compare the following three cases. (22A) Antero-posterior radiograph of the pelvis of a 36-year-old male patient clearly shows that the ipsilateral proximal femur is always affected when the pelvis is involved with fibrous dysplasia (orange circles). The lesion in the pelvis is more lytic than the lesion in the femur which is more sclerotic. (22B) Antero-posterior radiograph of the right knee of a 33-year-old female patient shows a circumscripted lytic lesion of the distal femur with smoky parts (orange arrow). (22C) Lateral radiograph of the left lower leg of a 22-year-old male patient with a fibrous dysplasia of the tibia shows a lytic lesion in the tibia with cortical destruction (orange arrows). MRI and biopsy were needed to confirm the diagnosis. Figures 22D–F show the corresponding MRI images to this case: (22D) Coronal T1-weighted MR image shows the classically low signal of lesion. (22E) Sagittal T2-weighted MR image and (22F) axial T2-weighted MR image with fat saturation show that the lesion is inhomogeneous. 22 A 22 D 22 E 22 F 22 B 22 C 23 A 23 B 23 C 23 F 23 D 23 E 34-year-old male patient with a polyostotic fibrous dysplasia in pelvis and proximal femur (Albright-syndrome). (23A) Antero-posterior radiograph of the pelvis, (23B) antero-posterior radiograph of the left femur and (23C) lateral radiograph of the left femur show lots of lesions with smoky appearance in the right os ileum and the left femur (orange arrow). (23D) Coronal T1-weighted MR image of the left femur shows a low signal of the lesions (orange arrow). (23E) Coronal STIR MR image of the left femur shows an intermediate to high signal of the lesions (orange arrow) and (23F) axial contrast-enhanced T1-weighted MR image shows an inhomogeneous contrast-enhancement of the fibrous tissue (orange arrow). 23 Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 89 Clinical Orthopedic Imaging 88 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 46. Lipoid/fatty type Calcaneuslipoma (Fig. 24) A common location is the calcaneus. It is a rare entity and a so-called ‘leave-me-alone lesion’. Key to diagnosis: Fat signal in all MRI sequences. Other types Solitary bone cyst (Figs. 25, 26) Patients are usually younger than 20 years. Common location: calcaneus, proximal humerus and femur with central location of the lesion. Patients have no pain or periosteal reaction unless they suffer a fracture through this lesion. The fracture often produces fragments 24 C 49-year-old male patient with an intra-osseous lipoma of the calcaneus as typical location. (24A) Lateral radiograph of the calcaneus shows the geographic lesion with sclerotic rim, Lodwick IA (orange arrow). (24B) Axial contrast-enhanced T1w MRI with fat saturation, (24C) coronal T2w MRI, (24D) sagittal T1w MRI and (24E) coronal contrast-enhanced T1w MRI. MR images show well that the lesion contains fat, especially seen in 24B and E (orange circle). Notice the synovial cyst between calcaneus and talus in 24C as auxiliary diagnosis (orange arrow). 24 24 A 24 B 24 D 24 E that sink to the bottom of the lesion, well known as the ‘fallen fragment sign’ visible on radiographs. Key to diagnosis: Lytic centrally located lesion, well-defined with sclerotic rim (Lodwick type IA). The MRI shows non-enhancing pure fluid (in contrary to aneurysmal bone cyst). Orthopedic Imaging Clinical 25-year-old male patient with a solitary bone cyst of the calcaneus. (25A) Lateral and (25B) antero-posterior radiographs of the calcaneus show both the geographic lesion with sclerotic rim, Lodwick IA (orange arrows). Typically for the location in the anterior to the midportion of the calcaneus and on the inferior border is: only in this position the solitary bone cyst has a characteristic triangular appearance. 25 25 B If the lesion is located in the calcaneus think about the differential diagnosis of an intra-osseous lipoma. A differentiation by X-ray is then only possible if the lipoma has a central calcification. But this differentiation is not relevant, because both lesions are ‘leave-me-alone lesions’ [11]. 25 A Save the Date 3rd Heidelberg Summer School Musculoskeletal Cross Sectional Imaging 2014 July 25/26, 2014 Heidelberg, Germany Please visit: www.heidelbergsummerschool.de MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 91 Clinical Orthopedic Imaging 90 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 47. 26 Image galleryof solitary bone cyst with the ‘fallen fragment sign’. (26A) Antero-posterior radiograph of the right shoulder of a 9-year-old female patient with the ‘fallen fragment sign’ in a solitary bone cyst of the humerus (orange arrow). Figures 26B–E show the case of an 11-year-old patient with a solitary bone cyst also in the right humerus. (26B) Antero-posterior radiograph of the right shoulder demonstrates well the pathognomonic ‘fallen fragment sign’ of the cystic lesion. (26C) Shows a coronal T1w MRI with low signal of the lesion and (26E) shows an axial T2w MRI with a small fluid level between the cystic fluid and the blood after the occurred fracture (orange arrow). Usually solitary bone cysts show no fluid-fluid levels as it is typical for the aneurysmal bone cyst. (26D) Coronal contrast-enhanced T1w MRI with fat saturation shows non-enhancing pure fluid. Aneurysmal bone cyst (ABC) (Fig. 27) The patients are usually younger than 20 years. At the vertebral column, this entity often occurs at the posterior elements of the vertebral bodies. It shows an aneurysmal, expansive growth with thinned cortex or neo- cortex (ballooned cortices) called ‘blow-out’ phenomenon in CT. Key to diagnosis: The aneurysmal bone cyst is a lytic geographic lesion, eccentrically located with extensive thinning of the cortex. Sedimentation effects of blood-filled cysts with fluid-fluid levels and contrast-enhancement of the cystic wall and the septa are typical signs in MRI. If there are solid 27 B 17-year-old male patient with an aneurysmal bone cyst of the right glenoid. (27A) Lateral radiograph of the right shoulder shows a geographic lesion in the glenoid without sclerotic rim, Lodwick IB (orange arrows). (27B) Coronal CT of the right shoulder shows a lytic expansible lesion with a thinned cortex (yellow arrow). (27C) Axial T2-weighted MRI shows the cystic parts with fluid-fluid level (yellow arrow). (27D) Axial contrast-enhanced T1w MRI shows the enhancement of the septa (orange circle). 27 contrast-enhancing parts consider secondary ABC with other tumors (e.g. giant cell tumor, osteosarcoma, chondrosarcoma, chondroblastoma). 26 A 26 C 26 B 26 D 26 E 27 A 27 C 27 D Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 93 Clinical Orthopedic Imaging 92 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 48. Giant cell tumor(Fig. 28) A precondition is that the epiphysis is closed. This tumour often abuts the articular surface and most often has an eccentric localization. This is often a well defined lesion with a non-sclerotic margin (Lodwick IB). Local aggressive growth and lung metastasis in 5–10% occur. Key to diagnosis: Osteolytic eccentric, epiphyseal lesion without matrix calcification and extensive thinning of the cortex. The tumor shows low signal in T1-weighting, inhomogeneous or low signal in T2-weighting and contrast-enhancement. If the tumor contains necrosis and hemosiderin, this results in an inhomogeneous contrast-enhancement pattern. Ewing‘s sarcoma (Figs. 29, 30) The classic Ewing‘s sarcoma is a, ‘permeative lesion in the diaphysis of long bone in a child’, [11], with osteodestruction in CT and a very high signal in T2-weighted imaging indicating infiltration of bone marrow. The location of Ewing‘s sarcoma tends to follow the distribution of red marrow. In histology small round blue cells are visible. A large soft tissue mass is possible. Important differential diagnoses are osteomyelitis and eosinophilic granuloma, which have a benign periosteal reaction and sometimes a sequestrum. Keys to diagnosis are: A permeative lesion or lesion with sclerotic and 29 D 29 E 29 F 16-year-old male patient with Ewing‘s sarcoma of the proximal left forearm. (29A) Bone scan shows an overview of the involvement of the proximal radius, the ulna and parts of the distal humerus. (29B) Lateral radiograph of the forearm shows the onion-skinned, multilamellated periosteal reaction of the proximal radius (yellow arrows). (29C) Three-phase radionuclide bone scan with Tc-99m MDP shows the tracer uptake in the big tumor mass. (29D, E) Contrast-enhanced T1-weighted MRI with fat saturation axial (D) and coronal (E) and (29F) coronal STIR MRI shows the big tumor involving radius and ulna. 29 3-year-old male patient with Ewing’s sarcoma of the left distal femur. (30A) Antero-posterior radiograph of the femur, (30B) lateral radiograph of the femur, (30C) coronal contrast-enhanced T1-weighted MRI with fat saturation, (30D) coronal T1-weighted MRI, (30E) coronal STIR MRI. Figure 30A clearly shows the Codman-triangle, whereby the elevated periosteum forms an angle with the cortex, (orange circle) and 30B the onion-skinned periosteal reaction (yellow arrow). Figures 30C–E show the T1w hypo-, T2w hyperintense signal character of the tumor with contrast-enhancement, the large soft tissue component (yellow arrows) and also the Codman triangle. 30 patchy appearance and periosteal reaction which can be onion-skinned (multilamellated), sunburst or amorphous. Low signal in T1-weighted MR images, high signal in T2-weighted MR imaging with strong contrast-enhancement. More than 50% are osteolytic lesions. Edema and large soft tissue mass often occur. 29 A 28 A 28 B 29 B 29 C 30 A 30 B 30 C 30 D 30 E 28 34-year-old female patient with giant cell tumor of the distal right femur. (28A) Antero-posterior radiograph of the knee, (28B) coronal CT of the knee, (28C) coronal T1w MRI, (28D) T2w axial MRI, (28E) coronal contrast-enhanced T1w MRI. Figure 28A shows the well-defined lesion with an incomplete sclerotic rim (Lodwick IB) in the distal femur (yellow arrow). 28B shows the extensive thinning of the cortex in lateral direction with a little cortex destruction below (yellow arrow). 28C clearly shows the low signal in T1w, 28D the low signal in T2w (orange circle) and 28E an inhomogeneous contrast-enhancement pattern because of the necrosis area below (yellow arrow). 28 C 28 E 28 D Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 95 Clinical Orthopedic Imaging 94 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 49. Multiple myeloma (Fig.31) In multiple myeloma, a proliferation of monoclonal plasma cells within the bone marrow occurs. The vertebral column is mostly affected and 70% of patients are older than 60 years. Multiple lytic lesions in an adult older than 40 years almost always suggest metastases or multiple myeloma. Bone sarcomas are rare, and the most common cause of a solitary destructive lesion in an Metastases (Fig. 32) 40% of all metastases are located in the vertebral column. The most frequent primary tumors are lung, breast, prostate, renal cell, gastrointes-tinal and thyroid carcinomas. Bone marrow infiltration happens before osseous destruction. It is important to pay attention to fractures, spinal canal invasion and myelon compression. Key to diagnosis: For the diagnosis of bone metastases a low signal in T1-weighted MR images is more sen-sitive than osteolysis in CT [24]. Osteolytic metastases have a high signal in T2-weighted images, whereas osteoblastic metastases have a low to isointense signal in T2-weighted images. Take into account these factors in older patients and consider several osteoblastic and/or osteolytic lesions. adult is a metastasis. Low-dose CT is important for proving osteolytic lesions and MRI [22] for proving bone marrow affection: Decrease of T1-weighted signal in bone marrow infiltration compared to the disks, and a signal increase in the STIR images compared to muscle tissue. Whole-body MRI is suitable for demonstration of the tumor burden. It is important to think of patient’s age when interpreting T1-weighted MR imaging, because young patients still have a cell-rich red bone marrow and therefore also a low T1 signal. We differentiate three patterns of bone marrow infiltration: diffuse, multi-focal, or ‘salt-and-pepper’ pattern. Salt-and-pepper pattern indicates a low grade disease stadium. A single lesion is called plasmacytoma [22, 23]. 65-year-old female patient with multiple myeloma and pain of the backside. (31A) Antero-posterior radiograph shows osteopenic bone pattern with several punched out lytic lesions. (31B) Coronal T1-weighted MR image of the pelvis and (31C) coronal STIR MR image of the pelvis show a solitary lesion with low signal in T1w and high signal in STIR suitable to a focal lesion of the left os sacrum (yellow arrow). 31 32 55-year-old man with osteoblastic metastases of vertebral column and pelvis in prostate cancer. (32A) Lateral radiograph of the vertebral column, (32B) antero-posterior radiograph of the lower vertebral column and os sacrum show several osteoblastic metastases of the vertebral bodies and the os sacrum. (32C) Scintigraphic bone scan shows more skeletal metastases. 31 A 31 B 32 C 32 A 32 B 31 C Head R lat Head L lat Thorax R V L Thorax L D R Pelvis R V L Pelvis L D R Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 97 Clinical Orthopedic Imaging 96 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 50. Summary Role ofX-ray In addition to patient history and clinical findings, a radiograph in two orthogonal planes is still of great importance for determining the Lodwick classification and the tumor matrix, whereas the bone matrix is only poorly visualized in X-ray: You cannot differentiate between lesions containing fluid and solid lesions without mineralized matrix. In general, conventional X-ray radiography is the starting point and CT and MR images should only be interpreted with concurrent radiographic correlation. Role of CT CT is superior to MRI for the assessment of mineralized structures especially cortical integrity, matrix mineralization, and periosteal reactions [21]. Small lucency of the cortex, localized involvement of the soft tissues, and thin peripheral periosteal reaction can Contact Katharina Grünberg, M.D. Section Musculoskeletal Radiology Diagnostic and Interventional Radiology University Hospital Heidelberg Schlierbacher Landstraße 200a 69118 Heidelberg Germany [email protected] References 1 WHO classification of bone tumours 2006. 2 Vanel D, Ruggieri P, Ferrari S, Picci P, Gambarotti M, Staals E, Alberghini M. The incidental skeletal lesion: ignore or explore? Cancer Imaging. 2009 Oct 2; 9 Spec No A: S38-43. 3 Lodwick GS, Wilson AJ, Farrell C, Virtama P, Dittrich F. Determining growth rates of focal lesions of bone from radiographs. Radiology 1980; 134: 577-583. 4 Erlemann R. Basic diagnostics of bone tumors. Radiologe 2009; 49: 257–267. 5 Oudenhoven LF, Dhondt E, Kahn S, Nieborg A, Kroon HM, Hogendoorn PC, Gielen JL, Bloem JL, De Schepper A. Accuracy of radiography in grading and tissue-specific diagnosis-a study of 200 consecutive bone tumors of the hand. Skeletal Radiol. 2006; 35: 78-87. 6 Miller T. Bone tumors and tumorlike conditions: analysis with conventional radiography. Radiology 2008; 246: 662-674. 7 Lucas DR, Unni KK, McLeod RA, O’Connor MI, Sim FH. Osteoblastoma: clinicopathologic study of 306 cases. Hum Pathol. 1994 Feb;25(2):117-34. 8 Rehnitz C, Sprengel SD, Lehner B, Ludwig K, Omlor G, Merle C, Kauczor HU, Ewerbeck V, Weber MA. CT-guided radiofrequency ablation of osteoid osteoma and osteoblastoma: clinical success and long-term follow up in 77 patients. Eur J Radiol 2012 Nov;81(11):3426-34. doi: 10.1016/j. ejrad.2012.04.037. Epub 2012 Jul 6. 9 Rehnitz C, Sprengel SD, Lehner B, Ludwig K, Omlor G, Merle C, Kauczor HU, Ewerbeck V, Weber MA. CT-guided radiofrequency ablation of osteoid osteoma: correlation of clinical outcome and imaging features. Diagn Interv Radiol. 2013 Jul-Aug;19(4):330-9. doi: 10.5152/ dir.2013.096. be best seen with CT [16]. CT is the examination of choice in the diagnosis of the nidus of osteoid osteoma in dense bone [17]. CT is valuable in the diagnosis of tumors of the axial skeleton such as spinal metastasis as well as in systemic staging. Role of MRI [18-21] Without any radiation MRI can be helpful while evaluating lesions that represent a differential diagnosis dilemma between benign and malignant lesions before a biopsy. For example in aneurysmal bone cysts MRI can display fluid levels in blood filled cavities better than CT. Another example, MRI before biopsy for staging all suspected sarcomas of bone could help identifying extraosseous sarcoma better. MRI plays an important role in planning limb salvage surgeries because of its superior role for soft tissue evaluation including the presence or absence of neurovascular invasion [21]. MRI helps by identifying skip lesions and helps measure the thickness of cartilage cap. The cap is thin in benign lesions and thicker in chondrosarcomas [14, 15]. This aids evaluation of the entire compartment of long bones in acceptable time. (Important here is a large field-of- view of the MR sequence, see Fig. 33.) This in turn helps to improve the quality of life by reducing morbidity without affecting survival. MRI is most useful in evaluation of spine metastasis differentiating osteoporotic and metastatic compression fractures. In Multiple Myeloma cases whole-body MRI scans are suitable for demonstration of the tumor burden. Though not yet in clinical routine, newer techniques such as diffusion- weighted imaging and DCE-MRI may support assessment of tumor response. More studies are being conducted. 10 Omlor GW, Lehner B, Wiedenhöfer B, Deininger C, Weber MA, Rehnitz C. [Radiofrequency ablation in spinal osteoid osteoma. Options and limits]. [Article in German]. Orthopade. 2012 Aug;41(8):618-22. doi: 10.1007/ s00132-012-1907-x. 11 Clyde A. Helms, Fundamentals of Skeletal Radiology (2005),3. Edition, Elsevier inc. 12 Murphey MD, Robbin MR, McRae GA, Flemming DJ, Temple HT, Kransdorf MJ. The many faces of osteosarcoma. Radiographics 1997; 17: 1205-1231. 13 Kloth JK, Wolf M, Rehnitz C, Lehner B, Wiedenhöfer B, Weber MA. [Radiological diagnostics of spinal tumors. Part 1: general tumor diagnostics and special diagnostics of extradural tumors]. Orthopade. 2012 Aug;41(8):595-607. doi: 10.1007/s00132-012-1978-8. [Article in German]. 14 Murphey MD, Choi JJ, Kransdorf MJ, Flemming DJ, Gannon FH. Imaging of osteochondroma: variants and complications with radiologic-pathologic correlation. Radiographics. 2000 Sep-Oct;20(5):1407-34. 15 Bernard SA, Murphey MD, Flemming DJ, Kransdorf MJ. Improved differentiation of benign osteochondromas from secondary condrosarcomas with standardized measurement of cartilage cap at CT and MR imaging. Radiology 2010 Jun;255(3):857-65. 16 Brown KT, Kattapuram SV, Rosenthal DI. Computed tomography analysis of bone tumors: patterns of cortical destruction and soft tissue extension. Skeletal Radiol 1986; 15: 448-451). 17 Glass RB, Poznanski AK, Fisher MR, Shkolnik A, Dias L. MR imaging of osteoid osteoma. J Comput Assist Tomogr 1986; 10: 1065-1067. Proposed tumor MRI protocol Sequences unenhanced coronal STIR with a large field-of-view coronal T1-weighted TSE (turbo spin echo) axial T2-weighted TSE with fat saturation sagittal T2-weighted TSE Contrast-agent (0.1 mmol/kg body weight) axial contrast-enhanced T1-weighted TSE with fat saturation coronal contrast-enhanced T1-weigthed TSE + subtraction (contrast-enhanced minus native T1-weighted MRI scan) The contrast-enhanced sequences are important in biopsy planning for identifying necrotic and viable tumor tissue. The biopsy should be targeted to the viable tumor area. Optional MR-angiography Dynamic T1-weighted contrast-enhanced MRI (DCE-MRI) Dynamic sequences are important for biopsy planning to identify vital tumor tissue [19-21], to which the biopsy should be guided. 13-year-old female patient with Ewing’s sarcoma. (33A, B) Coronal STIR MR images show the tumor in the left femur diaphysis (orange circle) with a large soft tissue mass surrounded by a soft tissue edema (yellow arrows) and a skip lesion in the femoral neck (orange arrow). 33 18 Anderson SE, Steinbach LS, Schlicht S, Powell G, Davies M, Choong P. Magnetic resonance imaging of bone tumors and joints. Top Magn Reson Imaging. 2007 Dec; 18(6):457-65. 19 Fayad LM, Jacobs MA, Wang X, Carrini JA, Bluemke DA. Musculoskeletal tumours: how to use anatomic, functional, and metabolic MR techniques. Radiology 2012; 265: 340-356. 20 Alyas F, James SL, Davies AM, Saifuddin A. The role of MR imaging in the diagnostic characterisation of appendicular bone tumours and tumour-like conditions. Eur Radiol 2007; 17: 2675-2686. 21 Roberts CC, Liu PT, Wenger DE. Musculoskeletal tumor imaging, biopsy, and therapies: self-assessment module. AJR Am J Roentgenol. 2009; 193(6 Suppl): S74-78. 22 Fechtner K, Hillengass J, Delorme S, Heiss C, Neben K, Goldschmidt H, Kauczor HU, Weber MA. Staging monoclonal plasma cell disease: comparison of the Durie- Salmon and the Durie-Salmon PLUS staging systems. Radiology. 2010 Oct; 257(1):195-204. doi: 10.1148/ radiol.10091809. 23 Bäuerle T, Hillengass J, Fechtner K, Zechmann CM, Grenacher L, Moehler TM, Christiane H, Wagner-Gund B, Neben K, Kauczor HU, Goldschmidt H, Delorme S. Multiple myeloma and monoclonal gammopathy of undetermined significance: importance of whole-body versus spinal MR imaging. Radiology. 2009 Aug; 252(2):477-85. doi: 10.1148/radiol.2522081756. 24 Bohndorf K, et al. Radiologische Diagnostik der Knochen und Gelenke. (2006) 2nd edition, Thieme. 33 A 33 B Orthopedic Imaging Clinical MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 99 Clinical Orthopedic Imaging 98 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 51. Combined 18F-FDG PETand MRI Evaluation of a case of Hypertrophic Cardiomyopathy Using Simultaneous MR-PET Ihn-ho Cho, M.D.; Eun-jung Kong, M.D. Department of Nuclear Medicine, Yeungnam University Hospital, Daegu, South Korea Introduction Hypertrophic cardiomyopathy (HCM) is a common condition causing left ventricular outflow obstruction, as well as cardiac arrhythmias. Cardiac MRI is a key modality for evaluation of HCM. Apart from estimating left ventricular (LV) wall thickness, LV function and aortic flow, MRI is capable of estimating the late gadolinium enhancement in affected myocardium, which has been shown to have a direct correlation with incidence and Patient history A 25-year-old man presented to the cardiology department with incidental ECG abnormality after fractures to his left 2nd and 4th fingers. Although he had not consulted a doctor, he had been suffering from mild dyspnea with chest discomfort at rest and exacerbation at exercise since May 2012. Echocardiography revealed non-obstructive hypertrophic cardiomyopathy (Maron III) with trivial MR. The patient was referred for a simultaneous MR-PET study for 18F-FDG PET and cardiac MRI with Gadolinium (Gd) contrast for evaluation of the morphological and metabolic status of the hypertrophic myocardium. The patient was injected with 10 mCi 18F- FDG following glucose loading. Simultaneous MR-PET study performed on a Biograph mMR was started one hour following tracer injection. Following standard Dixon sequence acquisition for attenuation correction, the comprehensive cardiac MRI sequences were acquired including MR perfusion after Gd contrast infusion, as well as post contrast late Gd enhancement studies. Static 18F-FDG PET was acquired simultaneously during the MRI acquisition. Cardiovascular Imaging Clinical 2A 2 Discussion The late Gd enhancement within the hypertrophic septum along with the non-uniform glucose metabolism demonstrated by the patchy 18F-FDG uptake within the hypertrophic septum exactly corresponding to the area of Gd enhancement reflect myocardial fibrosis within the asymmetric septal hypertrophy. Myocardial fibrosis and the presence of late Gd enhancement on MRI has been shown to be associated with increased risk of cardiac arrhythmia [1] as evident from the symptoms of this patient. severity of arrhythmias in HCM [1]. In patients with HCM, late gadolinium enhancement (LGE) on CE-MRI is presumed to represent intramyocardial fibrosis. PET myocardial perfusion studies have shown slight impairment of myocardial blood flow with pharmacological stress in hypertrophic myocardium in HCM, presumably related to microvascular disease [2]. 18F-FDG PET has been sporadically studied in HCM, mostly for evaluation of the metabolic status of the hypertrophic myocardial segment, especially after interventions such as transcoronary ablation of septal hypertrophy (TASH) [3] or to demonstrate partial myocardial fibrosis [4]. This clinical example illustrates the value of integrated simultaneous 18F-FDG PET and MRI acquisition performed on the Biograph mMR system. 1A 1B Short-axis views of end diastole and end systole at 3 different sections in the left ventricle obtained from gated TrueFISP cine MRI acquisitions performed on Biograph mMR. Note the thick hypertrophic septum (white arrow), which demonstrates the degree of asymmetric septal hypertrophy. 1 1D 1F 1C 1E 1G 1 1 End Diastole End Systole 2 3 2 3 Simultaneous MR-PET acquisition provides combined acquisition of both modalities, thereby ensuring accurate fusion between morphological and functional images due to simultaneous PET acquisition for every MR sequence. The exact coregistration of the patchy 18F-FDG uptake in the area of Gd enhancement within the hypertrophic upper septum reflects the advantage of simultaneous acquisition. End diastolic and end systolic views of 2-chamber and 4-chamber views obtained from gated cine TrueFISP acquisitions showing thickness of the asymmetric septal hypertrophy (white arrow). 2C 2B 2D End Diastole 4-chamber view 2-chamber view End Systole 3 3 Static 18F-FDG PET images in short-axis, horizontal long-axis and vertical long-axis views demonstrating normal uptake in the LV myocardium except the non-uniform uptake pattern in the hypertrophied septum (white arrows). LV cavity size appears normal. MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 101 Clinical Cardiovascular Imaging 100 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 52. Clinical Cardiovascular ImagingHow-I-do-it 4 5 Contact Ihn-ho Cho, M.D. Department of Nuclear Medicine Yeungnam University College of Medicine Daegu Hyunchungro 170 South Korea [email protected] 4D T2 HASTE T2 STIR T1 FAT SAT 4B 4E References 1 Rubinstein et al. Characteristics and Clinical Significance of Late Gadolinium Enhancement by Contrast-Enhanced Magnetic Resonance Imaging in Patients With Hypertrophic Cardiomyopathy. Circ Heart Fail. 2010;3:51-58. 2 Bravo et al. PET/CT Assessment of Symptomatic Individuals with Obstructive and Nonobstructive Hypertrophic Cardiomyopathy. J Nucl Med 2012; 53:407–414. Transverse, short-axis and vertical long-axis MR and fused MR-PET images show hypertrophied septum (white arrows) and normal thickness of rest of left ventricular myocardium with corresponding normal 18F-FDG uptake. The T2-weighted STIR (fat suppression) image shows slight hyperintensity in the middle of the hypertrophied septum which shows corresponding non-uniformity in 18F-FDG uptake. Post-contrast MR short-axis images demonstrate late Gd enhancement within the hypertrophied septum (white arrow), which shows corresponding non-uniform patchy uptake of 18F-FDG. 4A 5A 5B 4C 4F 4 Funabashi N et al. Partial myocardial fibrosis in hypertrophic cardiomyopathy demonstrated by 18F-fluoro-deoxyglucose positron emission tomography and multislice computed tomography. Int J Cardiol. 2006 Feb 15;107(2):284-6. 3 Kuhn et al. Changes in the left ventricular outflow tract after transcoronary ablation of septal hypertrophy (TASH) for hypertrophic obstructive cardiomyopathy as assessed by transoesophageal echocardiography and by measuring myocardial glucose utilization and. perfusion. European Heart Journal (1999) 20, 1808–1817. 102 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world What’s your favorite Dot Feature? Dot (Day optimizing throughput) is the most comprehensive MRI workflow solution, and it helps take the complexity out of MRI. Dot has now established itself in the field and our customers have told us what they like best about Dot: “Within our environment, we just could not provide a cardiac MRI service without the Cardiac Dot Engine.” Dr. Russell Bull, MRCP, FRCR Consultant Radiologist Royal Bournemouth Hospital, Bournemouth, UK “The Dot Decisions functionality in Abdomen Dot has enabled us to schematize and simplify these protocols. With Dot, we can now ensure our examinations are far more reproducible and of excellent quality.” Arnaud Lambert Technologist Imagerie Médicale Saint Marie, Osny, France “Cardiac Dot (Engine) allows us to obtain automatic positioning of the main slices necessary to evaluate cardiac function with a high degree of reproducibility.” Professor Philippe Cluzel, MD, PhD Service de Radiologie Polyvalente Diagnostique et Interventionnelle Hôpital Pitié-Salpêtrière, Paris, France “AutoAlign is helpful especially for colleagues who rarely perform knee examinations because the slices are positioned automatically, which saves a lot of time. Furthermore, our knee examinations have become reproducible.” Linda Willeke Technologist St. Franziskus Hospital, Münster, Germany Experience a Dot workflow yourself and hear from more users at www.siemens.com/Dot Visit our site optimized for tablets and smartphones MAGNETOM Flash · 3/2011 · www.siemens.com/magnetom-world 103
- 53. Clinical Cardiology CardiologyClinical New Generation Cardiac Parametric Mapping: the Clinical Role of T1 and T2 Mapping T1 mapping Initial T1 measurement methods were multi-breath-hold. These were time consuming and clunky, but were able to measure well diffuse myocardial fibrosis, a fundamental myocardial property with high potential clinical significance [1]. Healthy volunteers and those with disease had different extents of diffuse fibrosis [2], and these were shown to be clinically significant in a number of diseases. T1 mapping methods based on the MOLLI* approach with modifications for shorter breath-holds, better heart rate independence and better image registration for cleaner maps, however, transformed the field – albeit still with a variety of potential sequences in use [3-5]. There are two key ways of using T1 mapping: Without (or Viviana Maestrini; Amna Abdel-Gadir; Anna S. Herrey; James C. Moon The Heart Hospital Imaging Centre, University College London Hospitals, London, UK designed to optimize contrast between ‘normal’ and abnormal – a dichotomy of health and disease. As a result, global myocardial pathologies such as diffuse infiltration (fibrosis, amyloid, iron, fat, pan-inflammation) are missed. Recently, rapid technical innovations have generated new ‘mapping’ techniques. Rather than being ‘weighted’, these create a pixel map where each pixel value is the T1 or T2 (or T2*), displayed in color. These new sequences are single breath-hold, increasingly robust and now widely available. With T1 mapping, clever contrast agent use also permits the measurement of the extracellular volume (ECV), quantifying the interstitium (odema, fibrosis or amyloid), also as a map. Early results with these methodologies are exciting – potentially representing a new era of CMR. Introduction Cardiovascular magnetic resonance (CMR) is an essential tool in cardiology and excellent for cardiac function and perfusion. However, a key, unique advantage is its ability to directly scrutinize the fundamental material properties of myocardium – ‘myocardial tissue characterization’. Between 2001 and 2011, the key methods for tissue characterization have been sequences ‘weighted’ to a magnetic property – T1-weighted imaging for scar (LGE) and T2-weighted for edema (area at risk, myocarditis). These, particularly LGE imaging, have changed our understanding and clinical practice in cardiology. However, there are limitations to these approaches: Both are difficult to quantify – the LGE technique in particular is very robust in infarction, but harder to quantify in non-ischemic cardiomyopathy. A more fundamental difference is that sequences are before) contrast – Native T1 mapping; and with contrast, typically by subtracting the pre and post maps with hematocrit correction to generate the ECV [6]. Native T1 Native T1 mapping (pre-contrast T1) can demonstrate intrinsic myocardial contrast (Fig. 1). T1, measured in milliseconds, is higher where the extracellular compartment is increased. Fibrosis (focal, as in infarction, or diffuse) [7-8], odema [9-10] and amyloid [11], are examples. T1 is lower in lipid (Anderson Fabry disease, AFD) [12], and iron [13] accumulation. These changes are large in some rare disease. Global myocardial changes are robustly detectable without contrast, even in early disease. In iron, AFD and amyloid, changes appear before any other abnormality – there may be no left ventricular hypertrophy, a normal electrocardiogram, and normal conventional CMR, for example – genuinely new information. In established disease, low T1 values in AFD appear to absolutely distinguish it from other causes of left ventricular hypertrophy [12] whilst in established amyloid T1 elevation tracks known markers of cardiac severity [11]. A note of caution, however. Native T1, although stable between healthy volunteers to 1 part in 30, is dependent on platform (magnet manufacturer, sequence and sequence variant, field strength) [14]. Normal reference ranges for your setup are needed. Lowest ECV Tertile Middle ECV Tertile Highest ECV Tertile p < 0.001 fortrend p < 0.015 for Middle Tertile compared to others 2 ECV in non scar areas (LGE excluded) is associated with all-cause mortality [21]. The signal acquired is also a composite signal – generated by both interstitium and myocytes. The use of an extracellular contrast agent adds another dimension to T1 mapping and the ability to characterize the extracellular compartment specifically. Extracellular volume (ECV) Initially, post-contrast T1 was measured, but this is confounded by renal clearance, gadolinium dose, body composition, acquisition time post bolus, and hematocrit. Better is measuring the ECV. The ratio of change of T1 between blood and myocardium after contrast, at sufficient equilibrium (e.g. after 15 minutes post-bolus – no infusion generally needed) [15, 16], represents the contrast agent partition coefficient [17], and if corrected for the hematocrit, the myocardial extracellular space – ECV [1]. The ECV is specific for extracellular expansion, and well validated. Clinically this occurs in fibrosis, amyloid and odema. To distinguish, the degree of ECV change and the clinical context is important. A multiparametric approach (e. g. T2 mapping or T2-weighted imaging in addition) may therefore be useful. Amyloid can have far higher ECVs than any other disease [18] whereas ageing has small changes – near the detection limits, but of high potential clinical importance [19, 20]. For low ECV expansion diseases, biases from blood pool partial volume errors need to be meticulously addressed. Nevertheless, even modest ECV changes appear prognostic. In 793 consecutive patients (all-comers but excluding amyloid and HCM, measuring outside LGE areas) followed over 1 year, global ECV predicted short term-mortality (Fig. 2) * The product is currently under development; is not for sale in the U.S. and other countries, and its future availability cannot be ensured. 1B 1C 3A 3B 3C 1 Native T1 maps of (1A) healthy volunteer (author VM): the myocardium appears homogenously green and the blood is red; (1B) cardiac amyloid: the myocardium has a higher T1 (red); (1C) Anderson Fabry disease: the myocardium has a lower T1 (blue) from lipid – except the inferolateral wall where there is red from fibrosis; (1D) myocarditis, the myocardium has a higher T1 (red) from edema, which is regional; (1E) iron overload: the myocardium has a lower T1 (blue) from iron. 1A 2 A patient with myocarditis. On the left side a native T1 map showing the higher T1 value in the inferolateral wall (1115 ms); in the centre, a post-contrast T1 map showing the shortened T1 value after contrast administration (594 ms); on the right side the derived ECV map showing higher value of ECV (58%) compared to remote myocardium. 3 1D 1E 100% 0% Proportion Surviving Years of Follow-up 1.0 0.9 0.8 0.7 0.6 0.5 0 0.5 1.0 1.5 2.0 104 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 105
- 54. Clinical Cardiology CardiologyClinical Progress is rapid; challenges remain. Delivery across sites and standardization is now beginning with new draft guidelines for T1 mapping in preparation. Watch this space. References 1 Flett AS, Hayward MP, Ashworth MT, Hansen MS, Taylor AM, Elliott PM, McGregor C, Moon JC. Equilibrium Contrast Cardiovascular Magnetic Resonance for the measurement of diffuse myocardial fibrosis: preliminary validation in humans. Circulation 2010;122:138-144. 2 Sado DM, Flett AS, Banypersad SM, White SK, Maestrini V, Quarta G, Lachmann RH, Murphy E, Mehta A, Hughes DA, McKenna WJ, Taylor AM, Hausenloy DJ, Hawkins PN, Elliott PM, Moon JC. Cardiovascular magnetic resonance measurement of myocardial extracellular volume in health and disease. Heart 2012;98:1436-1441. 3 Piechnik SK, Ferreira VM, Dall’Armellina E, Cochlin LE, Greiser A, Neubauer S, Robson MD. Shortened Modified Look- Locker Inversion recovery (ShMOLLI) for clinical myocardial T1-mapping at 1.5 and 3 T within a 9 heartbeat breathhold. J Cardiovasc Magn Reson 2010;12:69. 4 Messroghli DR, Greiser A, Fröhlich M, Dietz R, Schulz-Menger J. Optimization and validation of a fully-integrated pulse sequence for modified look-locker inversion-recovery (MOLLI) T1 mapping of the heart. J Magn Reson Imaging 2007;26:1081–1086. 5 Fontana M, White SK, Banypersad SM, Sado DM, Maestrini V, Flett AS, Piechnik SK, Neubauer S, Roberts N, Moon JC. Comparison of T1 mapping techniques for ECV quantification. Histological validation and reproducibility of ShMOLLI versus multibreath-hold T1 quantification equilibrium contrast CMR. J Cardiovasc Magn Reson 201;14:88. 6 Kellman P, Wilson JR, Xue H, Ugander M, Arai AE. Extracellular volume fraction mapping in the myocardium, part 1: evaluation of an automated method. J Cardiovasc Magn Reson 2012;14:63. 7 Dass S, Suttie JJ, Piechnik SK, Ferreira VM, Holloway CJ, Banerjee R, Mahmod M, Cochlin L, Karamitsos TD, Robson MD, Watkins H, Neubauer S. Myocardial tissue characterization using magnetic resonance non contrast T1 mapping in hypertrophic and dilated cardiomyopathy. Circ Cardiovasc Imaging. 2012; 6:726-33. 8 Puntmann VO, Voigt T, Chen Z, Mayr M, Karim R, Rhode K, Pastor A, Carr-White G, Razavi R, Schaeffter T, Nagel E. Native T1 mapping in differentiation of normal myocardium from diffuse disease in hypertrophic and dilated cardiomyopathy. J Am Coll Cardiovasc Imgaging 2013;6:475–84. Contact Dr. James C. Moon The Heart Hospital Imaging Centre University College London Hospitals 16–18 Westmoreland Street London W1G 8PH UK Phone: +44 (20) 34563081 Fax: +44 (20) 34563086 [email protected] 4A 4B 4C 4D (4A) T2 mapping in a normal volunteer (author VM). (4B) High T2 value in patient with myocarditis – here epicardial edema. (4C) Edema in acute myocardial infarction – here patchy due to microvascular obstruction – see LGE, (4D). 4 [21]. The same group also found (n ~1000) higher ECVs in diabetics. Those on renin-angiotensin-aldosterone system blockade had lower ECVs. ECV also predicted mortality and/or incident hospitalization for heart failure in diabetics [22]. The use and capability of ECV quantification is growing. T1 mapping is getting better and inline ECV maps are now possible where each pixel carries directly the ECV value (Fig. 3) – a more biologically relevant figure than T1 [6]. T2 mapping T2-weighted CMR identifies myocardial odema both in inflammatory pathologies and acute ischemia, delineating the area at risk. However, these imaging techniques (e. g. STIR) are fragile in the heart and can be challenging, both to acquire and to interpret. Preliminary advances were made with T2-weighted SSFP sequences, which reduce false negatives and positives [23, 24]. T2 mapping seems a further increment [25] (Fig. 4). As with T1 mapping, global diseases such as pan-myocarditis may now be identified by T2 mapping, and preliminary results are showing this in several rheumatologic diseases (lupus, systemic capillary leak syndrome) and transplant rejection, detecting early rejection missed by other modalities [26, 27]. Conclusion Mapping – T1, T2, ECV mapping of myocardium is an emerging topic with the potential to be a powerful tool in the identification and quantification of diffuse myocardial processes without biopsy. Early evidence suggests that this technique detects early stage disease missed by other imaging methods and has potential as a prognosticator, as a surrogate endpoint in trials, and to monitor therapy. 9 Ferreira VM, Piechnik SK, Dall’Armellina E, Karamitsos TD, Francis JM, Choudhury RP, Friedrich MG, Robson MD, Neubauer S. Non-contrast T1-mapping detects acute myocardial edema with high diagnostic accuracy: a comparison to T2-weighted cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2012; 14:42. 10 Dall’Armellina E, Piechnik SK, Ferreira VM, Si Ql, Robson MD, Francis JM, Cuculi F, Kharbanda RK, Banning AP, Choudhury RP, Karamitsos TD, Neubauer S. Cardiovascular magnetic resonance by non contrast T1-mapping allows assessment of severity of injury in acute myocardial infarction. J Cardiovasc Magn Reson 2012;14:15. 11 Karamitsos TD, Piechnik SK, Banypersad SM, Fontana M, MD, Ntusi NB, Ferreira VM, Whelan CJ, Myerson SG, Robson MD, Hawkins PN, Neubauer S, Moon JC. Non-contrast T1 Mapping for the Diagnosis of Cardiac Amyloidosis. J Am Coll Cardiol Img 2013;6:488–97. 12 Sado DM, White SK, Piechnik SK, Banypersad SM, Treibel T, Captur G, Fontana M, Maestrini V, Flett AS, Robson MD, Lachmann RH, Murphy E, Mehta A, Hughes D, Neubauer S, Elliott PM, Moon JC. Identification and assessment of Anderson-Fabry Disease by Cardiovascular Magnetic Resonance Non-contrast myocardial T1 Mapping clinical perspective. Circ Cardiovasc Imaging 2013;6:392-398. 13 Pedersen SF, Thrys SA, Robich MP, Paaske WP, Ringgaard S, Bøtker HE, Hansen ESS, Kim WY. Assessment of intramyocardial hemorrhage by T1-weighted cardiovascular magnetic resonance in reperfused acute myocardial infarction. J Cardiovasc Magn Reson 2012; 14:59. 14 Raman FS, Kawel-Boehm N, Gai N, Freed M, Han J, Liu CY, Lima JAC, Bluemke DA, Liu S. Modified look-locker inversion recovery T1 mapping indices: assessment of accuracy and reproducibility between magnetic resonance scanners. J Cardiovasc Magn Reson 2013; 15:64. 15 White SK, Sado DM, Fontana M, Banypersad SM, Maestrini V, Flett AS, Piechnik SK, Robson MD, Hausenloy DJ, Sheikh AM, Hawkins PN, Moon JC. T1 Mapping for Myocardial Extracellular Volume measurement by CMR: Bolus Only Versus Primed Infusion Technique, 2013 Apr 5 [Epub ahead of print]. 16 Schelbert EB, Testa SM, Meier CG, Ceyrolles WJ, Levenson JE, Blair AJ, Kellman P, Jones BL, Ludwig DR, Schwartzman D, Shroff SG, Wong TC. Myocardial extravascular extracellular volume fraction measurement by gadolinium cardiovascular magnetic resonance in humans: slow infusion versus bolus. J Cardiovasc Magn Reson 2011, Mar 4;13-16. 17 Flacke SJ, Fischer SE, Lorenz CH. Measurement of the gadopentetate dimeglumine partition coefficient in human myocardium in vivo: normal distribution and elevation in acute and chronic infarction. Radiology 2001;218:703-10. 18 Banypersad SM, Sado DM, Flett AS, Gibbs SDG, Pinney JH, Maestrini V, Cox AT, Fontana M, Whelan CJ, Wechalekar AD, Hawkins PN, Moon JC. Quantification of myocardial extracellular volume fraction in systemic AL amyloidosis: An Equilibrium Contrast Cardio-vascular Magnetic Resonance Study. Circ Cardiovasc Imaging 2013;6:34-39. 19 Ugander M, Oki AJ, Hsu LY, Kellman P, Greiser A, Aletras AH, Sibley CT, Chen MY, Bandettini WP, Arai AE. Extracellular volume imaging by magnetic resonance imaging provides insights into overt and sub-clinical myocardial pathology. Eur Heart J 2012; 33: 1268–1278. 20 Liu CY, Chang Liu Y, Wu C, Armstrong A, Volpe GJ, van der Geest RJ, Liu Y, Hundley WG, Gomes AS, Liu S, Nacif M, Bluemke DA, Lima JAC. Evaluation of age related interstitial myocardial fibrosis with Cardiac Magnetic Resonance Contrast-Enhanced T1 Mapping in the Multi-ethnic Study of Atherosclerosis (MESA). J Am Coll Cardiol 2013 Jul 3 [Epub ahead of print]. 21 Wong TC, Piehler K, Meier CG, Testa SM, Klock AM, Aneizi AA, Shakesprere J, Kellman P, Shroff SG, Schwartzman DS, Mulukutla SR, Simon MA, Schelbert EB. Association between extracellular matrix expansion quantified by cardiovascular magnetic resonance and short-term mortality. Circulation 2012 Sep 4;126(10):1206-16. 22 Wong TC, Piehler KM, Kang IA, Kadakkal A, Kellman P, Schwartzman DS, Mulukutla SR, Simon MA, Shroff SG, Kuller LH, Schelbert EB. Myocardial extracellular volume fraction quantified by cardiovascular magnetic resonance is increased in diabetes and associated with mortality and incident heart failure admission. Eur Heart J 2013 Jun 11 [Epub ahead of print]. 23 Giri S, Chung YC, Merchant A, Mihai G, Rajagopalan S, Raman SV, Simonetti OP. T2 quantification for improved detection of myocardial edema. J Cardiovasc Magn Reson 2009; 11:56. 24 Verhaert D, Thavendiranathan P, Giri S, Mihai G, Rajagopalan S, Simonetti OP, Raman SV. Direct T2 Quantification of Myocardial Edema in Acute Ischemic Injury. J Am Coll Cardiol Img 2011;4: 269-78. 25 Ugander M, Bagi PS, Oki AB, Chen B, Hsu LY, Aletras AH, Shah S, Greiser A, Kellman P, Arai AE. Myocardial oedema as detected by Pre-contrast T1 and T2 CMR delineates area at risk associated with acute myocardial infarction. J Am Coll Cardiol Img 2012;5:596–603. 26 ThavendiranathanP, Walls M, Giri S, Verhaert D, Rajagopalan S, Moore S, Simonetti OP, Raman SV. Improved detection of myocardial involvement in acute inflammatory cardiomyopathies using T2 Mapping. Circ Cardiovasc Imaging 2012;5:102-110. 27 Usman AA, Taimen K, Wasielewski M, McDonald J, Shah S, Shivraman G, Cotts W, McGee E, Gordon R, Collins JD, Markl M, Carr JC. Cardiac Magnetic Resonance T2 Mapping in the monitoring and follow-up of acute cardiac transplant rejection: A Pilot Study. Circ Cardiovasc Imaging. 2012; 6:782-90. 120 ms 0 ms 106 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 107
- 55. Clinical Cardiovascular ImagingCardiovascular Imaging Clinical Preliminary Experiences with Compressed Sensing Multi-Slice Cine Acquisitions for the Assessment of Left Ventricular Function: CV_sparse WIP G. Vincenti, M.D.1; D. Piccini2,4; P. Monney, M.D.1; J. Chaptinel3; T. Rutz, M.D.1; S. Coppo3; M. O. Zenge, Ph.D.4; M. Schmidt4; M. S. Nadar5; Q. Wang5; P. Chevre1, 6; M.; Stuber, Ph.D.3; J. Schwitter, M.D.1 1 Division of Cardiology and Cardiac MR Center, University Hospital of Lausanne (CHUV), Lausanne, Switzerland 2 Advanced Clinical Imaging Technology, Siemens Healthcare IM BM PI, Lausanne, Switzerland 3 Department of Radiology, University Hospital (CHUV) and University of Lausanne (UNIL) / Center for Biomedical Imaging (CIBM), Lausanne, Switzerland 4 MR Applications and Workflow Development, Healthcare Sector, Siemens AG, Erlangen, Germany 5 Siemens Corportate Technology, Princeton, USA 6 Department of Radiology, University Hospital Lausanne, Switzerland ize pathological myocardial tissue was the basis to assign a class 1 indication for patients with known or suspected heart failure to undergo CMR in the new Heart Failure Guidelines of the European Society of Cardiology [3]. decision making [3] e.g. to start [4] or stop [5] specific drug treatments or to implant devices [6]. CMR is generally accepted as the gold standard method to yield most accurate measures of LV ejection fraction and LV volumes. This capability and the additional value of CMR to characterIntroduction Left ventricular (LV) ejection fraction is one of the most important measures in cardiology and part of every cardiac imaging evaluation as it is recognized as one of the strongest predictors of outcome [1]. It allows to assess the effect of established or novel treatments [2], and it is crucial for The evaluation of LV volumes and LV ejection fraction are based on well-defined protocols [7] and it involves the acquisition of a stack of LV short axis cine images from which volumes are calculated by applying Simpson’s rule. These stacks are typically acquired in multiple breath-holds. Quality criteria [8] for these functional images are available and are implemented e.g. for the quality assessment within the European CMR registry which currently holds approximately 33,000 patients and connects 59 centers [9]. Recently, compressed sensing (CS) techniques emerged as a means to considerably accelerate data acquisition without compromising significantly image quality. CS has three requirements: 1) transform sparsity, 2) incoherence of undersampling artifacts, and 3) nonlinear reconstruction (for details, see below). Based on these prerequisites, a CS approach for the acquisition of cardiac cine images was developed and tested*. In particular, the potential to acquire several slices covering the heart in different orientations within a single breath-hold would allow to apply model-based analysis tools which theoretically could improve the motion assessment at the base of the heart, where considerable through-plane motion on short-axis slices can introduce substantial errors in LV volume and LV ejection fraction calculations. Conversely, with a multi-breath- hold approach, there are typically small differences in breath-hold positions which can introduce errors in volume and function calculations. The pulse sequence tested here allows for the acquisition of 7 cine slices within 14 heartbeats with an excellent temporal and spatial resolution. Such a pulse sequence would also offer the advantage to obtain functional information in at least a single plane in patients unable to hold their breath for several heartbeats or in patients with frequent extrasystoles or atrial fibrillation. However, it should be mentioned that accurate quantitative measures of LV volumes and function cannot be obtained in highly arrhythmic hearts or in atrial fibrillation, as under such conditions volumes and ejection fraction change from beat to beat due to variable filling conditions. Nevertheless, rough estimates of LV volumes and function would still be desirable in arrhythmic patients. In a group of healthy volunteers and patients with different LV pathologies, the novel single-breath-hold CS cine approach was compared with the standard multi-breath-hold cine technique with respect to measure LV volumes and LV ejection fraction. The CV_sparse work-in-progress (WIP) The CV_sparse WIP package implements sparse, incoherent sampling and iterative reconstruction for cardiac applications. This method in principle allows for high acceleration factors which enable triggered 2D real-time cine CMR while preserving high spatial and/or temporal resolution of conventional cine acquisitions. Compressed sensing methods exploit the potential of image compression during the acquisition of raw input data. Three components [10] are crucial for the concept of compressed sensing to work I. Sparsity: In order to guarantee compressibility of the input data, sparsity must be present in a specific transform domain. Sparsity can be computed e.g. by calculating differences between neighboring pixels or by calculating finite differences in angiograms which then detect primarily vessel contours which typically 1 * Work in progress: The product is still under development and not commercially available yet. Its future availability cannot be ensured. 1 Display of the represent a few percent of the planning of the 7 slices (4 short axis and 3 long axis slices) acquired within a single breath-hold with the three localizers. 1 2A 2B Displays of the data analysis tools for the conventional short axis stack of cine images covering the entire LV (2A) and the 4D analysis tool (2B), which is model-based and takes long axis shortening of the LV, i.e. mitral annulus motion into account. Note that with both analysis tools, LV trabeculations are included into the LV volume, particularly in the end-diastolic images (corresponding images on the left of top row in 2A and 2B). 2 108 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 109
- 56. Clinical Cardiovascular ImagingCardiovascular Imaging Clinical entire image data only. Furthermore, sparsity is not limited to the spatial domain: the acquisition of cine images of the heart can be highly sparsified in the temporal dimension. II. Incoherent sampling: The aliasing artifacts due to k-space undersampling must be incoherent, i.e. noise-like, in that transform domain. Here, it is to mention that fully random k-space sampling is suboptimal as k-space trajectories should be smooth for hardware and physiological considerations. Therefore, incoherent sampling schemes must be designed to avoid these concerns while fulfilling the condition of random, i.e. incoherent sampling. III. Reconstruction: A non-linear iterative optimization corrects for sub-sampling artifacts during the process of image reconstruction yielding to a best solution with a sparse representation in a specific transform domain and which is consistent with the input data. Such compressed sensing techniques can also be combined with parallel imaging techniques [11]. WIP CV_sparse Sequence The current CV_sparse sequence [12] realizes incoherent sampling by initially distributing the readouts pseudo-randomly on the Cartesian grid in k-space. In addition, for cine-CMR imaging, a pseudo-random offset is applied from frame-to-frame which results in an incoherent temporal jitter. Finally, a variable sampling density in k-space stabilizes the iterative reconstruction. To avoid eddy current effects for balanced steady-state free precession (bSSFP) acquisitions, pairing [13] can also be applied. Thus, the tested CV_sparse sequence is characterized by sparse, incoherent sampling in space and time, non-linear iterative reconstruction integrating SENSE, and L1 wavelet regularization in the phase encoding direction and/or the temporal dimension. With regard to reconstruction, the ICE program runs a non-linear iterative reconstruction with k-t regularization in space and time specifically modified for compressed sensing. The algorithm derives from a parallel imaging type reconstruction which takes coil sensitivity maps into account, thus supporting predominantly high acceleration factors. For cine CMR, no additional reference scans are needed because – similar to TPAT – the coil sensitivity maps are calculated from the temporal average of the input data in a central region of k-space consisting of not more than 48 reference lines. The extensive calculations for image reconstruction typically running 80 iterations are performed online on all CPUs on the MARS computer in parallel, in order to reduce reconstruction times. Volunteer and Patient studies In order to obtain insight into the image quality of single-breath-hold multi-slice cine CMR images acquired with the compressed sensing (CS) approach, we studied a group of healthy volunteers and a patient group with different pathologies of the left ventricle. In addition to the evaluation of image quality, the robustness and the precision of the CS approach for LV volumes and LV ejection fraction was also assessed in comparison with a standard high-resolution cine CMR approach. All CMR examinations were performed on a 1.5T MAGNETOM Aera (Siemens Healthcare, Erlangen, Germany). The imaging protocol consisted of a set of cardiac localizers followed by the acquisition of a stack of conventional short-axis SSFP cine images covering the entire LV with a spatial and temporal resolution of 1.2 x 1.6 mm2, and approximately 40 ms, respectively (slice thickness: 8 mm; gap between slices: 2 mm). LV 2-chamber, 3-chamber, and 4-chamber long-axis acquisitions were obtained for image quality assessment but were not used for LV volume quantifications. As a next step, to test the new CS-based technique, slice orientations were planned to cover the LV with 4 short-axis slices distributed evenly over the LV long axis complemented by 3 long-axis slices (i.e. a 2-chamber, 3-chamber, and 4-chamber slice) (Fig. 1). These 7 slices were then acquired in a single breath-hold maneuver lasting 14 heart beats (i.e. 2 heart beats per slice) resulting in an acceleration factor of 11.0 with a temporal and spatial resolution of 30 ms and 1.5 x 1.5 mm2, respectively (slice thickness: 6 mm). As the reconstruction algorithm is susceptible to aliasing in the phase-encoding direction, the 7 slices were first acquired with a non-cine acquisition to check for correct phase-encoding directions and, if needed, to adjust the field-of-view to avoid fold-over artifacts. After confirmation of correct imaging parameters, the 7-slice single-breath- hold cine CS-acquisition was performed. In order to obtain a reference for the LV volume measurement, a phase-contrast flow measurement in the ascending aorta was performed to be compared with the LV stroke volumes calculated from the standard and CS cine data. The conventional stack of cine SSFP images was analyzed by the Argus software (Siemens Argus 4D Ventricular Function, Fig. 2A). The CS cine data were analyzed by the 4D-Argus software (Siemens Argus, Fig. 2B). Such software is based on an LV model and, with relatively few operator interactions, the contours for the LV endocardium and epicardium are generated by the analysis tool. Of note, this 4D analysis tool automatically tracks the 3-dimensional motion of the mitral annulus throughout the cardiac cycle which allows for an accurate volume calculation particularly at the base of the heart. Results and discussion Image quality – robustness of the technique Overall, a very good image quality of the single-breath-hold multi-slice CS acquisitions was obtained in the 12 volunteers and 14 patient studies. All CS data sets were of adequate quality to undergo 4D analysis. Small structures such as trabeculations were visualized in the CS data sets as shown in Figures 3 and 4. However, very small structures, detectable by the conventional cine acquisitions, were less well discernible by the CS images. Therefore, it should be mentioned here, that this accelerated single- breath-hold CS approach would be adequate for functional measurements, i.e. LV ejection fraction assessment (see also results below), whereas assessment of small structures as present in many cardiomyopathies is more reliable when performed on conventional cine images. Temporal resolution of the new technique appears adequate to even detect visually the dyssynchroneous contraction pattern in left bundle branch block. Also, the image contrast between the LV myocardium and the blood pool was high on the CS images allowing for an easy assessment of the LV motion pattern. As a result, the single-breath-hold cine approach permits to reconstruct the LV in 3D space with high temporal resolution as illustrated in Figure 5. Since these data allow to correctly include the 3D motion of the base of the heart during the cardiac cycle, the LV stroke volume appears to be measurable by the CS approach with higher accuracy than with the conventional multi-breath-hold approach (see results below). With an accurate measurement of the LV stroke volume, the quantification of a mitral insufficiency should theoretically benefit (when calculating mitral regurgitant volume as ‘LV stroke volume minus aortic forward-flow volume’). As a current limitation of the CS approach, its susceptibility for fold-over artifacts should be mentioned (Figs. 6A). Therefore, the field-of-view must cover the entire anatomy and thus, some penalty in spatial resStandard cine 9 heartbeats CV_SPARSE 3 heartbeats CV_SPARSE 2 heartbeats CV_SPARSE 1 heartbeat Examples of visualization of small trabecular structures in the LV (in the rectangle) with the standard cine SSFP sequence (image on the left) and the accelerated compressed sensing sequences (images on the right). Despite increasing acceleration most information on small intraluminal structures remains visible. 3 RCA Example demonstrating the performance of the compressed sensing technique visualizing small structures such as the right coronary artery (RCA) with high temporal and spatial resolution acquired within 2 heartbeats. Short-axis view of the base of the heart (1 out of 17 frames). 4 3 4 110 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 111
- 57. Clinical Cardiovascular ImagingCardiovascular Imaging Clinical olution may occur in relation to the patient’s anatomy. In addition, the sparsity in the temporal domain may be limited in anatomical regions of very high flow, and therefore, in some acquisitions, flow-related artifacts occurred in the phase-encoding direction during systole (Figs. 6B, C). Also, in its current version, the sequence is prospective, thus it does not cover the very last phases of the cardiac cycle and the reconstruction times for the CS images lasted several minutes precluding an immediate assessment of the image data quality or using this image information to plan next steps of a CMR examination. Performance of the single-breath- hold CS approach in comparison with the stan-dard multi-breath-hold cine approach From a quantitative point-of-view, the accurate and reliable measurement of LV volumes and function is crucial as many therapeutic decisions directly depend on these measures [3–6]. In this current relatively small study group, LV end-diastolic and end-systolic volumes measured by the single-breath-hold CS approach were comparable with those calculated from the standard multi-breath-hold cine SSFP approach. LVEDV and LVESV differed by 10 ml ± 17 ml and 2 ml ± 12 ml, respectively. Most importantly, LV ejection fraction differed by only 1.3 ± 4.7% (50.6% vs 49.3% for multi-breath-hold and single-breath-hold, respectively, p = 0.17; regression: r = 0.96, p < 0.0001; y = 0.96x + 0.8 ml). Thus, it can be concluded that the single-breath-hold CS approach could potentially replace the multi-breath- hold standard technique for the assessment of LV volumes and systolic function. What about the accuracy of the novel single-breath-hold CS technique? To assess the accuracy of the LV volume measurements, LV stroke volume was compared with the LV output measured in the ascending aorta with phase-contrast MR. As the flow measurements were performed distally to the coronary arteries, flow in the coronaries was estimated as the LV mass multiplied by 0.8 ml/min/g. An excellent agreement was found with a mean of 86.8 ml/beat for the aortic flow measurement and 91.9 ml/beat for the LV measurements derived from the single-breath-hold CS data (r = 0.93, p < 0.0001). By Bland-Altman analysis, the stroke volume approach overestimated by 5.2 ml/beat versus the reference flow measurement. For the conventional stroke volume measurements, this difference was 15.6 ml/beat (linear regression analysis vs aortic flow: r = 0.69, p < 0.01). More importantly, the CS LV stroke data were not only more precise with a smaller mean difference, the variability of the CS data vs the reference flow data was less with a standard deviation as low as 6.8 ml/beat vs 12.9 ml/beat for the standard multi-breath-hold approach (Fig. 7). Several explanations may apply for the higher accuracy of the single-breath- hold multi-slice CS approach in comparison to the conventional multi-breath- hold approach: 1) With the single-breath-hold approach, all acquired slices are correctly co-registered, i.e. they are correctly aligned in space, a prerequisite for the 4D-analysis tool to work properly. 2) This 4D-analysis tool allows for an accurate tracking of the mitral valve plane motion during the cardiac cycle as shown in Figure 5, which is important as the cross-sectional area of the heart at its base is large and thus, inaccurate slice positioning at the base of Display of the 3D reconstruction derived from the 7 slices acquired within a single breath-hold. Note the long-axis shortening of the LV during systole allowing for accurate LV volume measurements (5A, 5B, yellow plane). Any orientation of the 3D is available for inspection of function (5A–D). 5 6A A typical fold-over artifact along the phase-encoding direction in a short axis slice, oriented superior-inferior for demonstrative purpose. 6B No flow-related artifacts are visible on the end-diastolic phases, while small artifacts in phase-encoding direction (Artif, arrows) occur in mid-systole projecting over the mitral valve (6C). 5A 5B 5C 5D 6A 6B 6C Artif Artif 112 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 113
- 58. Clinical Cardiovascular ImagingCardiovascular Imaging Clinical 50 60 70 120 130 130 120 110 100 90 80 70 the heart with conventional short-axis slices typically translate in relatively large errors. Nevertheless, we observed a systematic overestimation of the stroke volume by the CS approach of 5.2 ml/beat in comparison to the flow measurements. In normal hearts with tricuspid aortic valves, an underestimation of aortic flow by the phase-contrast technique is very unlikely [14]. Thus, overestimation of stroke volume by the volume approach is to consider. In the volume contours, the papillary muscles are excluded as illustrated in Figure 8. As these papillary muscles are excluded in both the diastolic and systolic contours, this aspect should not affect net LV stroke volume. However, as shown in Figure 8, smaller trabeculations of the LV wall are included into the LV blood pool contour in the diastolic phase, while these trabeculations, CS technique Standard technique when compacted in the end- systolic phase, are excluded from the blood pool resulting in a small overestimation of the end-diastolic volume, and thus, LV stroke volume. This explanation is likely as Van Rossum et al. demonstrated a slight underestimation of the LV mass when calculated on end-diastolic phases versus end-systolic phases, as trabeculations in end-diastole are typically excluded from the LV walls [15]. In summary, this novel very fast acquisition strategy based on a CS technique allows to cover the entire LV with high temporal and spatial resolution within a single breath-hold. The image quality based on these preliminary results appears adequate to yield highly accurate measures of LV volumes, LV stroke volume, LV mass, and LV ejection fraction. 7 Testing of this very fast multi-slice cine approach for the atria and the right ventricle is currently ongoing. Finally, these preliminary data show that compressed sensing MR acquisitions in the heart are feasible in humans and compressed sensing might be implemented for other important cardiac sequences such as fibrosis/viability imaging, i.e. late gadolinium enhancement, coronary MR angiography, or MR first-pass perfusion. The Cardiac MR Center of the University Hospital Lausanne The Cardiac Magnetic Resonance Center (CRMC) of the University Hospital of Lausanne (Centre Hospitalier Universitaire Vaudois; CHUV) was established in 2009. The CMR center is dedicated to high-quality clinical work-up of cardiac patients, to deliver state-of-the-art training in CMR to cardiologists and radiologists, and to pursue research. In the CMR center education is provided for two specialties while focusing on one organ system. Traditionally, radiologists have focussed on using one technique for different organs, while cardiologists have concentrated on one organ and perhaps one technique. Now in the CMR center the focus is put on a combination of specialists with different background on one organ. Research at the CMR center is devoted to four major areas: the study of 1.) cardiac function and tissue characterization, specifically to better understand diastolic dysfunction, 2.) the development of MR-compatible cardiac devices such as pacemakers and ICDs; 3.) the utilization of hyperpolarized 13C-carbon contrast media to investigate metabolism in the heart, and An excellent correlation is obtained for the LV stroke volume calculated from the compressed sensing data with the flow volume in the aorta measured by phase-contast technique. Variability of the conventional LV stroke volume data appears higher than for the compressed sensing data. LV stroke volume: comparison vs aortic forward flow LV short-axis slice: CV_SPARSE 4.) the development of 19F-fluorine-based CMR techniques to detect inflammation and to label and track cells non-invasively. For the latter two topics, the CMR center established tight collaborations with the Center for Biomedical Imaging (CIBM), a network around Lake Geneva that includes the Ecole Polytechnique Fédérale de Lausanne (EPFL), and the universities and university hospitals of Lausanne and Geneva. In particular, strong collaborative links are in place with the CVMR team of Prof. Matthias Stuber, a part of the CIBM and located at the University Hospital Lausanne and with Prof. A. Comment, with whom we perform the studies on real-time metabolism based on the 13C-carbon hyperpolarization (DNP) technique. In addition, collaborative studies are ongoing with the Heart Failure and Cardiac Transplantation Unit led by Prof. R. Hullin (detection of graft rejection by tissue characterization) and the Oncology Department led by Prof. Coukos (T cell tracking by19F- MRI in collaboration with Prof. Stuber, R. van Heeswijk, CIBM, and Prof. O. Michielin, Oncology). This structure allows for a direct interdisciplinary interaction between physicians, engineers, and basic scientists on a daily basis with the aim to enable innovative research and fast translation of these techniques from bench to bedside. The CMRC is also the center of competence for the quality assessment of the European CMR registry which holds currently approximately 33,000 patient studies acquired in 59 centers across Europe. The members of the CRMC team are: Prof. J. Schwitter (director of the center), PD Dr. X. Jeanrenaud, Dr. D. Locca, MER, Dr. P. Monney, Dr. T. Rutz, Dr. C. Sierro, and Dr. S. Koestner (cardiologists, staff members), Overestimation of end-diastolic LV volumes by volumetric measurements. In comparison to ejected blood from the LV as measured with phase-contrast techniques, the volumetric measurements of LV stroke volume overestimated by approximately 5 ml, most likely by overestimation of LV end-diastolic volume. Small trabculations (yellow contours in 8A) are included into the LV blood volume (red contour in 8A) in diastole, while these trabeculations (yellow contours in 8B) are typically included in the end-systolic phase (red contours in 8B). For the same reasons, LV mass (= green contour minus red contour) is often slightly underestimated in diastole vs systole. 8 7 8A 8B End-diastolic frame End-systolic frame ml/beat (aortic forward flow by PC) ml/beat (LV stroke volume) 80 90 100 110 60 114 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 115
- 59. References 1 CurtisJP, Sokol SI, Wang Y, Rathore SS, Ko DT, Jadbabaie F, Portnay EL, Marshalko SJ, Radford MJ, Krumholz HM. The association of left ventricular ejection fraction, mortality, and cause of death in stable outpatients with heart failure. Journal of the American College of Cardiology. 2003;42(4):736-42. 2 Sürder D, Manka R, Lo Cicero V, Moccetti T, Rufibach K, Soncin S, Turchetto L, Radrizzani M, Astori G, Schwitter J, Erne P, Jamshidi P, Auf Der Maur C, Zuber M, Windecker S, Moschovitis A, Wahl A, Bühler I, Wyss C, Landmesser U, Lüscher T, Corti R. Intracoronary injection of bone marrow derived mononuclear cells, early or late after acute myocardial infarction: Effects on global LV-function: 4 months results of the SWISS-AMI trial. Circulation. 2013;127:1968-79. 11 Liang D, Liu B, Wang J, Ying L. Accelerating SENSE using compressed sensing. Magnetic Resonance in Medicine. 2009;62(6): 1574-84. 12 Liu J. Dynamic cardiac MRI reconstruction with weighted redundant Haar wavelets. Magn Reson Med. 2012;Proc. ISMRM 2012,abstract. 13 Bieri O, Markl M, Scheffler K. Analysis and compensation of eddy currents in balanced SSFP. Magn Reson Med. 2005;54:129-37. 14 Muzzarelli S, Monney P, O’Brien K, Faletra F, Moccetti T, Vogt P, Schwitter J. Quantification of aortic valve regurgitation by phase-contrast magnetic resonance in patients with bicuspid aortic valve: where to measure the flow? . Eur Heart J - CV Imaging. 2013:in press. 15 Papavassiliu T, Kühl HP, Schröder M, Süselbeck T, Bondarenko O, Böhm CK, Beek A, Hofman MMB, van Rossum AC. Effect of Endocardial Trabeculae on Left Ventricular Measurements and Measurement Reproducibility at Cardiovascular MR Imaging1. Radiology. 2005 July 1, 2005;236(1):57-64. Contact Professor Juerg Schwitter Médecin Chef Cardiologie Directeur du Centre de la RM Cardiaque du CHUV Centre Hospitalier Universitaire Vaudois – CHUV Rue du Bugnon 46 1011 Lausanne Suisse Phone: +41 21 314 0012 [email protected] www.cardiologie.chuv.ch Dr. G. Vincenti (cardiologist) and Dr. N. Barras (cardiologist in training, rotation), PD. Dr. S. Muzzarellli (affiliated cardiologist), Prof. C. Beigelman and Dr. X. Boulanger (radiologists, staff members), Dr. G.L. Fetz (radiologist in training, rotation), C. Gonzales, PhD (19F-fluorine project leader), H. Yoshihara, PhD (13C-carbon project leader), V. Klinke (medical student, doctoral thesis), C. Bongard (medical student, master thesis), P. Chevre (chief CMR technician), and F. Recordon and N. Lauriers (research nurses). Acknowledgements The authors would like to thank all the members of the team of MR technologists at the CHUV for their highly valuable participation, helpfulness and support during the daily clinical CMR examinations and with the research protocols. Finally, a very important acknowledgment goes to Dr. Michael Zenge, Ms. Michaela Schmidt, and the whole Siemens MR Cardio team of Edgar Müller in Erlangen. 3 McMurray JJV, Adamopoulos S, Anker SD, Auricchio A, Böhm M, Dickstein K, Falk V, Filippatos G, Fonseca C, Sanchez MAG, Jaarsma T, Køber L, Lip GYH, Maggioni AP, Parkhomenko A, Pieske BM, Popescu BA, Rønnevik PK, Rutten FH, Schwitter J, Seferovic P, Stepinska J, Trindade PT, Voors AA, Zannad F, Zeiher A. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2012. European Heart Journal. 2012 May 19, 2012(33):1787–847. 4 Zannad F, McMurray JJV, Krum H, van Veldhuisen DJ, Swedberg K, Shi H, Vincent J, Pocock SJ, Pitt B. Eplerenone in Patients with Systolic Heart Failure and Mild Symptoms. New England Journal of Medicine. 2011;364(1):11-21. 5 Gharib MI, Burnett AK. Chemotherapy-induced cardiotoxicity: current practice and prospects of prophylaxis. European Journal of Heart Failure. 2002 June 1, 2002;4(3):235-42. 6 Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R, Domanski M, Troutman C, Anderson J, Johnson G, McNulty SE, Clapp-Channing N, Davidson- Ray LD, Fraulo ES, Fishbein DP, Luceri RM, Ip JH. Amiodarone or an Implantable Cardioverter–Defibrillator for Congestive Heart Failure. New England Journal of Medicine. 2005;352(3):225-37. 7 Schwitter J. CMR-Update. 2. Edition ed. Lausanne, Switzerland. www.herz-mri.ch. 8 Klinke V, Muzzarelli S, Lauriers N, Locca D, Vincenti G, Monney P, Lu C, Nothnagel D, Pilz G, Lombardi M, van Rossum A, Wagner A, Bruder O, Mahrholdt H, Schwitter J. Quality assessment of cardiovascular magnetic resonance in the setting of the European CMR registry: description and validation of standardized criteria. Journal of Cardiovascular Magnetic Resonance. 2013;15(1):55. 9 Bruder O, Wagner A, Lombardi M, Schwitter J, van Rossum A, Pilz G, Nothnagel D, Steen H, Petersen S, Nagel E, Prasad S, Schumm J, Greulich S, Cagnolo A, Monney P, Deluigi C, Dill T, Frank H, Sabin G, Schneider S, Mahrholdt H. European Cardiovascular Magnetic Resonance (EuroCMR) registry-multi national results from 57 centers in 15 countries. J Cardiovasc Magn Reson. 2013;15:1-9. 10 Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magnetic Resonance in Medicine. 2007;58(6):1182-95. Accelerated Segmented Cine TrueFISP of the Heart on a 1.5T MAGNETOM Aera Using k-t-sparse SENSE Maria Carr1; Bruce Spottiswoode2; Bradley Allen1; Michaela Schmidt2; Mariappan Nadar4; Qiu Wang4; Jeremy Collins1; James Carr1; Michael Zenge2 1 Northwestern University, Feinberg School of Medicine, Chicago, IL, USA 2 Siemens Healthcare 3 Siemens Corporate Technology, Princeton, United States Introduction Cine MRI of the heart is widely regarded as the gold standard for assessment of left ventricular volume and myocardial mass and is increasingly utilized for assessment of cardiac anatomy and pathology as part of clinical routine. Conventional cine imaging approaches typically require 1 slice per breath-hold, resulting in lengthy protocols for complete cardiac coverage. Parallel imaging allows some shortening of the acquisition time, such that 2–3 slices can be acquired in a single breath-hold. In cardiac cine imaging artifacts become more prevalent with increasing acceleration factor. This will negatively impact the diagnostic utility of the images and may reduce accuracy of quantitative measurements. However, regularized iterative reconstruction techniques can be used to considerably improve the images obtained from highly undersampled data. In this work, L1-regularized iterative SENSE as proposed in [1] was applied to reconstruct under-sampled k-space data. This technique* takes advantage of the de-noising characteristics of Wavelet regularization and promises to very effectively suppress sub-sampling artifacts. This may allow for high acceleration factors to be used, while diagnostic image quality is preserved. The purpose of this study was to compare segmented cine TrueFISP images from a group of volunteers and patients using three acceleration and reconstruction approaches: iPAT factor 2 with conventional reconstruction; T-PAT factor 4 with convenTable tional reconstruction; and T-PAT factor 4 with iterative k-t-sparse SENSE reconstruction. Technique Cardiac MRI seems to be particularly well suited to benefit from a group of novel image reconstruction methods known as compressed sensing [2] which promise to significantly speed up data acquisition. Compressed sensing methods were introduced to MR imaging [3, 4] just a few years ago and have since been successfully combined with parallel imaging [5, 6]. Such methods try to utilize the * Work in progress: The product is still under development and not commercially available yet. Its future availability cannot be ensured. 1: MRI conventional and iterative imaging parameters Parameters Conventional iPAT 2 Conventional T-PAT 4 Iterative T-PAT 4 Iterative recon No No Yes Parallel imaging iPAT2 (GRAPPA) TPAT4 TPAT4 TR/TE (ms) 3.2 / 1.6 3.2 / 1.6 3.2 / 1.6 Flip angle (degrees) 70 70 70 Pixel size (mm2) 1.9 × 1.9 1.9 × 1.9 1.9 × 1.9 Slice thickness (mm) 8 8 8 Temp. res. (msec) 38 38 38 Acq. time (sec) 7 3.2 3.2 Clinical Cardiovascular Imaging 116 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world Cardiovascular Imaging Technology MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 117
- 60. As outlined byLiu et al. in [1], the image reconstruction can be formulated as an unconstrained optimization problem. In the current implementation, this optimization is solved using a Nesterov-type algorithm [7]. The L1-regularization with a redundant Haar transform is efficiently solved using a Dykstra-type algorithm [8]. This allowed a smooth integration into the current MAGNETOM platform and, therefore, facilitates a broad clinical evaluation. Materials and methods Nine healthy human volunteers (57.4 male/56.7 female) and 20 patients (54.4 male/40.0 female) with suspected cardiac disease were scanned on a 1.5T MAGNETOM Aera system under an approved institutional review board protocol. All nine volunteers and 16 patients were imaged using segmented cine TrueFISP sequences with conventional GRAPPA factor 2 acceleration (conventional iPAT 2) T-PAT factor 4 acceleration (conventional T-PAT 4), and T-PAT factor 4 acceleration with iterative k-t-sparse SENSE reconstruction (iterative T-PAT 4). The remaining 4 patients were scanned using only conventional iPAT 2 and iterative T-PAT 4 techniques. Note that the iterative technique is fully integrated into the standard reconstruction environment. The imaging parameters for each imaging sequence are provided in Table 1. All three sequences were run in 3 chamber and 4 chamber views, as well as a stack of short axis slices. Quantitative analysis was performed on all volunteer data sets at a syngo MultiModality Workplace (Leonardo) using Argus post-processing software (Siemens Healthcare, Erlangen, Germany) by an experienced cardiovascular MRI technician. Ejection fraction, end-diastolic volume, end-systolic volume, stroke volume, cardiac output, and myocardial mass were calculated. In all volunteers and patients, 5 4 3 2 1 blinded qualitative scoring was performed by a radiologist using a 5 point Likert scale to assess overall image quality (1 – non diagnostic; 2 – poor; 3 – fair; 4 – good; 5 – excellent). Images were also scored for artifact and noise (1 – severe; 2 – moderate; 3 – mild; 4 – trace; 5 – none). All continuous variables were compared between groups using an unpaired t-test, while ordinal qualitative variables were compared using a Wilcoxon signed-rank test. Results All images were acquired successfully and image quality was of diagnostic quality in all cases. The average scan time per slice for conventional iPAT 2, conventional T-PAT 4 and iterative T-PAT 4 were for patients 7.7 ± 1.5 sec, 5.6 ± 1.5 sec and 2.9 ± 1.5 sec and for the volunteers 9.8 ± 1.5 sec, 3.2 ± 1.5 sec and 3.0 ± 1.5 sec, respectively. The results in scan time are illustrated in Figure 1. In both patients and volunteers, conventional iPAT 2 were significantly longer than both conventional T-PAT 4 and iterative T-PAT 4 techniques (p < 0.001 for each group). The results for ejection fraction (EF) for all three imaging techniques are provided in Figure 2. The average EF for conventional T-PAT 4 was slightly lower than that measured for conventional iPAT 2 and iterative T-PAT, but the group size is relatively small (9 subjects) and this difference was not significant (p = 0.34 and p = 0.22 respectively).There was no statistically significant difference in ejection fraction between the conventional iPAT 2 and the iterative T-PAT 4 sequences (p = 0.48). The results for image quality, noise and artifact are provided in Figure 3. The iterative T-PAT 4 images had comparable image quality, noise and artifact scores compared to the conventional iPAT 2 images. The conventional T-PAT 4 images had lower image quality, more artifacts and higher noise compared to the other techniques. Figures 4 and 5 show an example of 4-chamber and mid-short axis images from all three techniques in a patient with basal septal hypertrophy. In both series’, the conventional iPAT 2 and iterative T-PAT 4 images are comparable in quality, while the conventional T-PAT 4 image is visibly noisier. Cardiovascular Imaging Technology Discussion This study compares a novel accelerated segmented cine TrueFISP technique to conventional iPAT 2 cine TrueFISP and T-PAT 4 cine TrueFISP in a cohort of normal subjects and patients. The iterative reconstruction technique provided comparable measurements of ejection fraction to the clinical gold standard (conventional iPAT 2). The accelerated segmented cine TrueFISP with T-PAT 4, which was used as comparison technique, produced slightly lower EF values compared to the other techniques, although this was not found to be statistically significant. The iterative reconstruction produced comparable image quality, noise and artifact scores to the conventional reconstruction using iPAT 2. The conventional T-PAT 4 technique had lower image quality and higher noise scores compared to the other two techniques. The iterative T-PAT 4 segmented cine technique allows for greater than 50% reduction in acquisition time for comparable image quality and spatial resolution as the clinically used iPAT 2 cine TrueFISP technique. This iterative technique could be extended to permit complete heart coverage in a single breath-hold thus greatly simplifying and shortening routine clinical cardiac MRI protocols, which has been one of the biggest obstacles to wide acceptance of cardiac MRI. With a shorter cine acquisition, additional advanced imaging techniques, such as perfusion and flow, can be more readily added to patient scans within a reasonable protocol length. Technology Cardiovascular Imaging 12 10 8 6 4 Single slice scan time in patients and volunteers. There was a statistically significant reduction in scan time compared to the standard iPAT2 for both TPAT4 acceleration and iterative reconstruction TPAT4 acceleration. 1 Qualitative scores in patients and volunteers. Image quality was highest and noise and artifact were lowest with iterative T-PAT 4 and conventional iPAT 2 compared to conventional T-PAT 4. 3 full potential of image compression during the acquisition of raw input data. In the case of highly subsampled input data, a non-linear iterative optimization avoids sub-sampling artifacts during the process of image reconstruction. The resulting images represent the best solution consistent with the input data, which have a sparse representation in a specific transform domain. In the most favorable case, residual artifacts are not visibly perceptible or are diagnostically irrelevant. 0 Conventional iPAT 2 Scan Time (sec) Conventional T-PAT 4 Iterative T-PAT 4 2 Standard iPAT 2 T-PAT 4 Acceleration Iterative Reconstruction T-PAT 4 Accel. 0 Quality Noise Artifact 1 3 65,00 60,00 55,00 50,00 45,00 40,00 Ejection fraction in volunteers. Quantitatively measured ejection fractions were comparable across all three techniques. 2 Conventional iPAT 2 Ejection Fraction (%) Conventional T-PAT 4 Iterative T-PAT 4 2 118 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 119
- 61. Technology Cardiovascular ImagingCardiovascular Imaging Technology There are currently some limitations to the technique. Firstly, the use of SENSE implies that aliasing artifacts can occur if the field-of-view is smaller than the subject, which is sometimes difficult to avoid in the short axis orientation. But a solution to this is promised to be part of a future release of the current proto-type. Secondly, the image reconstruc-tion times of the current implemen-tation seems to be prohibitive for routine clinical use. However, we anticipate future algorithmic 6A 6B 6 Real-time cine TrueFISP T-PAT 6 images reconstructed using (6A) conventional, and (6B) iterative techniques. Contact Maria Carr, RT (CT)(MR) CV Research Technologist Department of Radiology Northwestern University Feinberg School of Medicine 737 N. Michigan Ave. Suite 1600 Chicago, IL 60611 USA Phone: +1 312-926-5292 [email protected] References 1 Liu J, Rapin J, Chang TC, Lefebvre A, Zenge M, Mueller E, Nadar MS. Dynamic cardiac MRI reconstruction with weighted redundant Haar wavelets. In Proceedings of the 20th Annual Meeting of ISMRM, Melbourne, Australia, 2002. p 4249. 2 Candes EJ, Wakin MB. An Introduction to compressive sampling. IEEE Signal Processing Magazine 2008. 25(2):21-30. doi: 10.1109/MSP.2007.914731. improvements with increased compu-tational power to reduce the recon-struction time to clinically acceptable values. Of course, iterative reconstruction techniques are not just limited to cine imaging of the heart. Future work may see this technique applied to time intense techniques such as 4D flow phase contrast MRI and 3D coronary MR angiography, making them more clinically applicable. Furthermore, higher acceleration rates might be achieved by using an incoherent sampling pattern [9]. With sufficiently high acceleration, the technique can also be used effectively for real time cine cardiac imaging in patients with breath-holding difficul-ties or arrhythmia. Figure 6 shows that real-time acquisition with T-PAT 6 and k-t iterative reconstruction still results in excellent image quality. In conclusion, cine TrueFISP of the heart with inline k-t-sparse iterative reconstruction is a promising tech-nique for obtaining high quality cine images at a fraction of the scan time compared to conventional techniques. Acknowledgement The authors would like to thank Judy Wood, Manger of the MRI Department at Northwestern Memorial Hospital, for her continued support and collabo-ration with our ongoing research through the years. Secondly, we would like to thank the magnificent Cardio-vascular Technologist’s Cheryl Jarvis, Tinu John, Paul Magarity, Scott Luster for their patience and dedication to research. Finally, the Resource Coordi-nators that help us make this possible Irene Lekkas, Melissa Niemczura and Paulino San Pedro. 3 Block KT, Uecker M, Frahm J. Unders-ampled Radial MRI with Multiple Coils. Iterative Image Reconstruction Using a Total Variation Constraint. Magn Reson Med 2007. 57(6):1086-98. 4 Lustig M, Donoho D, Pauly JM. Sparse MRI: The application of compressed sensing for rapid MR imaging. Magn Reson Med 2007. 58(6):1182-95. 5 Liang D, Liu B, Wang J, Ying L. Acceler-ating SENSE using compressed sensing. Magn Reson Med 2009. 62(6):154-84. doi: 10.1002/mrm.22161. 6 Lustig M, Pauly, JM. SPIRiT: Iterative self- consistent parallel imaging reconstruction from arbitrary k-space. Magn Reson Med 2010. 64(2):457-71. doi: 10.1002/mrm.22428. 7 Beck A, Teboulle M. A fast iterative shrinkage-thresholding algorithm for linear inverse problems. SIAM J Imaging Sciences 2009. 2(1): 183-202. 8 Dykstra RL. An algorithm for restricted least squares regression. J Amer Stat Assoc 1983 78(384):837-842. 9 Schmidt M, Ekinci O, Liu J, Lefebvre A, Nadar MS, Mueller E, Zenge MO. Novel highly accelerated real-time CINE-MRI featuring compressed sensing with k-t regularization in comparison to TSENSE segmented and real-time Cine imaging. J Cardiovasc Magn Reson 2013. 15(Suppl 1):P36. 4A 4B 4C Four chamber cine TrueFISP from a normal volunteer. (4A) Conventional iPAT 2, acquisition time 8 s. (4B) Conventional T-PAT 4, acquisition time 3 seconds. (4C) Iterative T-PAT 4, acquisition time 3 seconds. 4 5A 5B 5C End-systolic short axis cine TrueFISP images from a patient with a history of myocardial infarction. A metal artifact from a previous sternotomy is noted in the sternum. There is wall thinning in the inferolateral wall with akinesia on cine views, consistent with an old infarct in the circumflex territory. (5A) Conventional iPAT 2, (5B) conventional T-PAT 4, (5C) iterative T-PAT 4. 5 120 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 121
- 62. Subscriptions Siemens HealthcarePublications Our publications offer the latest information and background for every healthcare field. From the hospital director to the radiological assistant – here, you can quickly find information relevant to your needs. Heartbeat Everything from the world of sustainable cardiovascular care. MAGNETOM Flash Everything from the world of magnetic resonance imaging. SOMATOM Sessions Everything from the world of computed tomography. Medical Solutions Online The online version includes additional video features and greater depth to the articles in the printed healthcare leadership magazine. Read online at: www.siemens.com/medical-solutions AXIOM Innovations Everything from the world of interventional radiology, cardiology, and surgery. Imaging Life Everything from the world of molecular imaging innovations. For current and past issues and to order the magazines, please visit www.siemens.com/healthcare-magazine cannot Siemens sales subject to herein may specifications be present options information. tolerances. Europe/Africa/Middle East: Siemens AG, Healthcare Sector Henkestr. 127 91052 Erlangen, Germany Phone: +49 9131 84-0 Latin America: Siemens S.A., Medical Solutions Avenida de Pte. Julio A. Roca No 516, Piso C1067 ABN Buenos Aires, Argentina Phone: +54 11 4340-8400 USA: Siemens Medical Solutions USA, Inc. 51 Valley Stream Parkway Malvern, PA 19355-1406, USA Phone: +1 888 826-9702 09.13, Siemens AG MAGNETOM Flash ASTRO Edition | Issue Number 4/2013 54 MAGNETOM Flash The Magazine of MRI Issue Number 4/2013 | ASTRO Edition 54 Functional, Volumetric, Treatment Response Assess-ment Using MR OncoTreat Page 24 Optimizing MRI for Radiation Oncology Page 45 How-I-do-it Tissue 4D on syngo.via Page 18 Not for distribution in the US 11.09.13 09:35 122 MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world
- 63. Imprint The entireeditorial staff at Duke University Medical Center and at Simens Healthcare extends their appreciation to all the radiologists, technologists, physicists, experts and scholars who donate their time and energy – without payment – in order to share their expertise with the readers of MAGNETOM Flash. connection with such use. The Operating Instructions must always be strictly followed when operating the MR system. The sources for the technical data are the corresponding data sheets. Results may vary. Partial reproduction in printed form of individual contributions is permitted, provided the customary bibliographical data such as author’s name and title of the contribution as well as year, issue number and pages of MAGNETOM Flash are named, but the editors request that two copies be sent to them. The written consent of the authors and publisher is required for the complete reprinting of an article. We welcome your questions and comments about the editorial content of MAGNETOM Flash. Please contact us at [email protected]. Manuscripts as well as suggestions, proposals and information are always welcome; they are carefully examined and submitted to the editorial board for attention. MAGNETOM Flash is not responsible for loss, damage, or any other MAGNETOM injury to Flash unsolicited is also manuscripts or available other materials. on the internet: We reserve the right to www.edit siemens.for clarity, com/accuracy, magnetom-and world space. Include your name, address, and phone number and send to the editors, address above. MAGNETOM Flash – Imprint © 2013 by Siemens AG, Berlin and Munich, All Rights Reserved Publisher: Siemens AG Medical Solutions Business Unit Magnetic Resonance, Karl-Schall-Straße 6, D-91052 Erlangen, Germany Guest Editor: Mustafa R. Bashir, M.D. Director of MRI Assistant Professor of Radiology Division of Abdominal Imaging Duke University Medical Center Durham, NC, USA Associate Editor: Antje Hellwich ([email protected]) Editorial Board: Wellesley Were; Ralph Strecker; Sven Zühlsdorff, Ph.D.; Gary R. McNeal, MS (BME); Peter Kreisler, Ph.D. Production: Norbert Moser, Siemens AG, Medical Solutions Layout: independent Medien-Design Widenmayerstrasse 16, D-80538 Munich, Germany Printer: G. Peschke Druckerei GmbH, Schatzbogen 35, D-81829 Munich, Germany Note in accordance with § 33 Para.1 of the German Federal Data Protection Law: Despatch is made using an address file which is maintained with the aid of an automated data processing system. MAGNETOM Flash is sent free of charge to Siemens MR customers, qualified physicians, technologists, physicists and radiology departments throughout the world. It includes reports in the English language on magnetic resonance: diagnostic and therapeutic methods and their application as well as results and experience gained with corresponding systems and solutions. It introduces from case to case new principles and proce-dures and discusses their clinical poten-tial. The statements and views of the authors in the individual contributions do not necessarily reflect the opinion of the publisher. The information presented in these articles and case reports is for illustration only and is not intended to be relied upon by the reader for instruction as to the practice of medicine. Any health care practitioner reading this information is reminded that they must use their own learning, training and expertise in dealing with their individual patients. This material does not substitute for that duty and is not intended by Siemens Medical Solutions to be used for any purpose in that regard. The drugs and doses mentioned herein are consistent with the approval labeling for uses and/or indications of the drug. The treating physician bears the sole responsibility for the diagnosis and treatment of patients, including drugs and doses prescribed in MAGNETOM Flash is also available on the internet: www.siemens.com/magnetom-world Not for distribution in the US. MAGNETOM Flash | 5/2013 | www.siemens.com/magnetom-world 123
- 64. Please include mein your mailing list for the following Siemens Healthcare customer magazine(s): Medical Solutions MAGNETOM Flash SOMATOM Sessions AXIOM Innovations Stay up to date with the latest information Register for: Yes, I consent to the above information being used for future contact regarding product updates and other important news from Siemens. Please print clearly! Subscription unsubscribe from info service Please enter your business address Institution Department Function Title Name Street Postal Code City State Country MR system used E-mail
- 65. ➔Visit www.siemens.com/ magnetom-world for case reports, clinical methods, application tips, talks and much more clinical information. SUBSCRIBE NOW! – and get your free copy of future MAGNETOM Flash! Interesting information from the world of magnetic resonance – gratis to your desk. Send us this postcard, or subscribe online at www.siemens.com/MAGNETOM-World MAGNETOM Flash Siemens AG Healthcare Sector Magnetic Resonance Antje Hellwich – Marketing P.O. Box 32 60 D-91050 Erlangen Germany
- 66. On account ofcertain regional limitations of sales rights and service availability, we cannot guarantee that all products included in this brochure are available through the Siemens sales organization worldwide. Availability and packaging may vary by country and is subject to change without prior notice. Some/All of the features and products described herein may not be available in the United States. The information in this document contains general technical descriptions of specifications and options as well as standard and optional features which do not always have to be present in individual cases. Siemens reserves the right to modify the design, packaging, specifications, and options described herein without prior notice. Please contact your local Siemens sales representative for the most current information. Note: Any technical data contained in this document may vary within defined tolerances. Original images always lose a certain amount of detail when reproduced. Local Contact Information Asia/Pacific: Siemens Medical Solutions Asia Pacific Headquarters The Siemens Center 60 MacPherson Road Singapore 348615 Phone: +65 6490 6000 Canada: Siemens Canada Limited Healthcare Sector 1550 Appleby Lane Burlington, ON L7L 6X7, Canada Phone +1 905 315-6868 Europe/Africa/Middle East: Siemens AG, Healthcare Sector Henkestr. 127 91052 Erlangen, Germany Phone: +49 9131 84-0 Latin America: Siemens S.A., Medical Solutions Avenida de Pte. Julio A. Roca No 516, Piso C1067 ABN Buenos Aires, Argentina Phone: +54 11 4340-8400 USA: Siemens Medical Solutions USA, Inc. 51 Valley Stream Parkway Malvern, PA 19355-1406, USA Phone: +1 888 826-9702 Global Siemens Healthcare Headquarters Siemens AG Healthcare Sector Henkestrasse 127 91052 Erlangen Germany Phone: +49 9131 84-0 www.siemens.com/healthcare Not for distribution in the US Global Business Unit Siemens AG Medical Solutions Magnetic Resonance Henkestrasse 127 DE-91052 Erlangen Germany Phone: +49 9131 84-0 www.siemens.com/healthcare Global Siemens Headquarters Siemens AG Wittelsbacherplatz 2 80333 Munich Germany Order No. A91MR-1000-102C-7600 | Printed in Germany | CC MR 1661 111320. | © 11.13, Siemens AG www.siemens.com/magnetom-world