Ea4301770773

Krunal J Pimple et al Int. Journal of Engineering Research and Applications www.ijera.com
ISSN : 2248-9622, Vol. 4, Issue 3( Version 1), March 2014, pp.770-773
www.ijera.com 770 | P a g e
Bilateral Symmetry Information for Brain Tumor Detection
Krunal J Pimple, Asst. Prof.Prateek Nahar, Dr.Sanjay Thakur, Rajesh Patil
Abstract
Image segmentation is used to separate an image into several “meaningful” parts. It is an old research topic,
which started around 1970, but there is still no robust solution toward it. There are two main reasons; the first is
that the content variety of images is too large, and the second one is that there is no benchmark standard to judge
the performance. Various subjects that are paired usually are not identically the same, asymmetry is perfectly
normal but sometimes asymmetry can benoticeable too much.
Structural and functional asymmetry in the human brain and nervous system is reviewed in a historical
perspective. Brain asymmetry is one of such examples, which is a difference in size or shape, or both.
Asymmetry analysis of brain has great importance because it is not only indicator for brain cancer but also
predict future potential risk for the same. In our work, we have concentrated to segment the anatomical regions
of brain, isolate the two halves of brain and to investigate each half for the presence of asymmetry of anatomical
regions in MRI.
Keywords: MRI, Asymmetry Relation, Region of Interest
I. Introduction
Segmentation subdivides an image into its
regions of components or objects and an important
tool in medical image processing. As an initial step
segmentation can be used for visualization and
compression. Through identifying all pixels (for two
dimensional image) or voxels (for three dimensional
image) belonging to an object, segmentation of that
particular object is achieved. In medical imaging,
segmentation is vital for feature extraction, image
measurements and image display [3]. Segmentation
of the brain structure from MRI has received
paramount importance as MRI distinguishes itself
from other modalities and MRI can be applied in the
volumetric analysis of brain tissues such as multiple
sclerosis, schizophrenia, epilepsy, Parkinson’s
disease, Alzheimer’s disease, cerebral atrophy, etc.
Other important aspect of the segmentation
method is the color space from which color features
are computed (for instance RGB space with
Euclidean color distance). Each segmentation
technique is usually based on some mathematical
model (theory) and/or algorithmic approach (for
instance fuzzy clustering, Markov random field,
recursive procedure, bottom-up algorithm etc.). Most
of segmentation techniques assume something about
the scene which is seen in the image (for instance
objects are polyhedral made of dielectric
materials).This is an additional knowledge attribute
of the given segmentation method [4].
Image segmentation is a process of pixel
classification. An image is segmented into subsets by
assigning individual pixels to classes. It is an
important step towards pattern detection and
recognition. Segmentation is one of the first steps in
image analysis. It refers to the process of partitioning
a digital image into multiple regions (sets of pixels).
Each of the pixels in a region is similar with respect
to some characteristic or computed property, such as
color, intensity, or texture. The level of segmentation
is decided by the particular characteristics of the
problem being considered. Image segmentation could
be further used for object matching between two
images. An object of interest is specified in the first
image by using the segmentation result of that image;
then the specified object is matched in the second
image by using the segmentation result of that image
[5].
II. Literature Survey
As the first step in image analysis and
pattern recognition, image segmentation is always a
crucial component of most image analysis and pattern
recognition systems, and determines the quality of
the final result of analysis. So image segmentation
has been intensively and extensively studied in the
past years. And a wide variety of methods and
algorithms are available to deal with the problem of
segmentation of images. According to existing
automatic image segmentation techniques can be
classified into four categories, namely, (1) Clustering
Methods, (2) Thresholding Methods, (3) Edge-
Detection Methods, and (4) Region-Based Methods
[6].
1. Clustering Methods
Clustering is a process whereby a data set
(pixels) is replaced by cluster; pixels may belong
together because of the same color, texture etc. There
are two natural algorithms for clustering: divisive
RESEARCH ARTICLE OPEN ACCESS

clustering and agglomerative clustering. The
difficulty in using either of the methods directly is
that there are lots of pixels in an image. Also, the
methods are not explicit about the objective function
that is being optimized. An alternative approach is to
write down an objective function and then build an
algorithm. The K-means algorithm is an iterative
technique that is used to partition an image into K
clusters, where each pixel in the image is assigned to
the cluster that minimizes the variance between the
pixel and the cluster center and is based on pixel
color, intensity, texture, and location, or a weighted
combination of these factors. This algorithm is
guaranteed to converge, but it may not return the
optimal solution. The quality of the solution depends
on the initial set of clusters and the value of K [7].
2. Thresholding Methods
Thresholding is the operation of converting
a multilevel image into a binary image i.e., it assigns
the value of 0 (background) or 1 (objects or
foreground) to each pixel of an image based on a
comparison with some threshold value T (intensity or
color value). When T is constant, the approach is
called global thresholding; otherwise, it is called
local thresholding. Global thresholding methods can
fail when the background illumination is uneven.
Multiple thresholds are used to compensate for
uneven illumination. Threshold selection is typically
done interactively.
3. Edge Detection Methods
Edge detection methods locate the pixels in
the image that correspond to the edges of the objects
seen in the image. The result is a binary image with
the detected edge pixels. Common algorithms used
are Sobel, Prewitt, Robert, Canny and Laplacian
operators. These algorithms are suitable for images
that are simple and noise free; and will often produce
missing edges, or extra edges on complex and noisy
images.
4. Region-Based Methods
The goal of region-based segmentation is to
use image characteristics to map individual pixels in
an input image to sets of pixels called regions that
might correspond to an object or a meaningful part of
one. The various techniques are: Local techniques,
Global techniques and Splitting and merging
techniques. The effectiveness of region growing
algorithms depends on the application area and the
input image. If the image is sufficiently simple,
simple local techniques can be effective. However,
on difficult scenes, even the most sophisticated
techniques may not produce a satisfactory
segmentation.
Edge-based techniques are based on the assumption
that pixel values change rapidly at the edge between
two regions Operators such as Sobel or Roberts
operators can be used to detect the edges. And some
post procedures such as edge tracking, gap filling can
be used to generate closed curves. Region-based
techniques are based on the assumption that adjacent
pixels in the same region should be consistent in
some properties. Namely, they may have similar
characteristic such as grey value, color value or
texture. The deformable models are based on curves
or surfaces defined within an image that moves due
to the influence of certain forces [8]. And the global
optimization approaches use a global criterion when
segmenting the image.
III. Problem Statement
Brain tumors are a heterogeneous group of
central nervous system neoplasms that arise within or
adjacent to the brain. Moreover, the location of the
tumor within the brain has a profound effect on the
patient's symptoms, surgical therapeutic options, and
the likelihood of obtaining a definitive diagnosis. The
location of the tumor in the brain also markedly alters
the risk of neurological toxicities that alter the
patient's quality of life.
At present, brain tumors are detected by
imaging only after the onset of neurological
symptoms. No early detection strategies are in use,
even in individuals known to be at risk for specific
types of brain tumors by virtue of their genetic
makeup. Current histopathological classification
systems, which are based on the tumor's presumed
cell of origin, have been in place for nearly a century
and were updated by the World Health Organization
in 1999. Although satisfactory in many respects, they
do not allow accurate prediction of tumorbehaviour
in the individual patient, nor do they guide
therapeutic decision-making as precisely as patients
and physicians would hope and need. Current
imaging techniques provide meticulous anatomical
delineation and are the principal tools for establishing
that neurological symptoms are the consequence of a
brain tumor. There are many techniques for brain
tumor detection. I have used edge detection
technique for brain tumor detection.
IV. The Proposed Mechanism
4.1. Our algorithm
Our algorithm composes of two steps. The
first is to define the bilateral symmetrical axis. The
second is to detect the region of brain tumor.
4.1.1 Symmetry axis defining
The first step of our algorithm is mainly
based on symmetry character of brain MRI.

If without tumor in the brain or the size of tumor is
very small, the symmetry axis can be defined with a
straight line x = k,(y >= 0) , which separates the
image into two bilateral symmetry parts, show as
Figure.2.
Y x=k
0 x
Figure.1. The bilateral symmetry axis is defined with
a straight line.
This kind of symmetry is not very strictly.
And, compared with normal brain MRI, the
symmetry characteristic is distorted for the existing
of brain tumor, such as the circumstance shown in
Figure.3.a.
Y=f(x) y
y
x
Figure:-3(a) The bilateral symmetry axis is defined
with a curve line.
The symmetry axis can’t be defined with a straight
line in the brain MRI with tumors, so a curve line is
more convenient to describe it.
For more convenient to describing symmetry axis, a
curve line ( y= f (x), x > 0, y > 0 ) is defined, which is
shown in Figure.3.b. From the edge map, the edge
point set Pecan be obtained. And then, we calculate
the edge-centroid Giof every line according to
equation (1).
k
Gi=1/k ∑ Pi, j, P i,j, Є Pe...................................(1)
j=1
where, Giis the abscissa of centroid in the i th line, k
is the edge point number in the i th line, whose
abscissas are Pi,1…Pi,k . So, based on the edge-
centroids, we can use the least square method to get
the symmetry curve line y approximatively.
V. Methodology Used
There are many techniques for brain tumor
detection. We have used edge detection technique
for brain tumor detection. Edge-based method is by
far the most common method of detecting boundaries
and discontinuities in an image. The parts on which
immediate changes in grey tones occur in the images
are called edges. Edge detection techniques transform
images to edge images benefiting from the changes
of grey tones in the images.
VI. Performance Evaluation
 If cutting of brain image gives symmetry by axis
then there will not be chances of tumor this is
detected by first algorithm otherwise there will
be chances of tumor.
 As in others there are various steps are required
to just identify whether there is tumor or not but
in this it shows exact region where tumor is
occurred.
 The color image is changes into gray scale image
and then by reiterative processing the tumor is
getting identified.
 Our purpose is to detect the tumor of brain
automatically.
 Compared with other automatic segmentation
methods, more effective the system model was
constructed and less time was consumed.
Table 6.1: Number of detected edges
Patient
ID
Grade
Number of Detected Edges
Robert Prewitt Canny
121 High 5259 4382 1997
122 High 5120 4323 1836
123 High 6807 5757 2302
124 Low 1491 649 317
125 Low 2509 1080 433
Table 6.2: Areas of tumor
Patient ID Lesion Volume
of tumor
areas
(Pixels)
% of
Dama
ge
areas
121 Left Frontal
Parietal
4315 17.26
122 Left High
Parietal
1068 4.27
123 Left Temporal
Lobe
435 1.74

124 Left Frontal
Parietal
1776 7.10
125 Left Thalamus 1060 4.24
VII. Conclusion
At first, it checks the image can be divided
into symmetric axis or not. If it is divided into
Symmetric part then no tumor in brain and it can be
divided in curve shape then chances of tumor in
human brain. However, if there is a macroscopic
tumor, the symmetry characteristic will be weakened.
Canny edge detecton technique use for the edge
detection because it find true edges in the images and
also work on the real images that reason result is
better than other techniques.
References
[1] Yu-Ning Liu Chung-Han Huang, Wei-Lun
Chao, “DIP: Final project report Image
segmentation based on the normalized cut
framework”, Volume-2, pp124-189.
[2] Chin-Ya Huang, Mon-Ju Wu , “Image
Segmentation”, ECE 533 Final Project, Fall
2006 University of Wisconsin- Madison, pp
1-3.
[3] B. Peng, L. Zheng and J. Yang, “Iterated
Graph Cuts for Image Segmentation”, Asian
Conference on Computer Vision
(ACCV’09), Xi'an, China, September 2009,
pp.23-27.
[4] Sharma, G., Yrzel, M.J., Trussel, H.J. “
Color imaging for multimedia” ,
Proceedings of the IEEE, 86(6), 1998, pp-
1088-1108.
[5] Wenbing Tao, Hai, Yimin Zhang Jin, “Color
Image Segmentation Based on Mean Shift
and Normalized Cuts”, IEEE Trans. On
System Man and Cybernetics-Part B,
Volume 37, No. 5, Oct 2007, pp-1382-1389.
[6] F. Z. Kettaf, D. Bi, and J. P. Asselin de
Beauville , "A comparison study of image
segmentation by clustering techniques",
IEEE Trans, Vol. 2,1996, pp. 1280-1283.
[7] Jai- Nan Wang, Jun Kong, Ying- Hualu, “ A
Region- Based SRG Algorithm For Color
Image Segmentation”,Proceedings of the
Sixth International Conference on Machine
Learning and Cybernetics, Hong Kong, 19-
22 August 2007 ,pp. 1-2.
[8] Kass, M., Witkin, A., Terzopoulos, D.,
“Snakes: Active contour models”,
International Journal Computer Vision
,Volume 1 (4), 1988, pp.321–331.
[9] Samir K. Bandyopadhyay,” Finding
Bilateral Symmetry of the Human Brain
from MRI”,Journal of Global Research in
Computer Science, Volume 2, No. 2,
February 2011,pp. 33-34.
[10] Sabina Breitenmoser,”Evalutation and
implementation of neural brain activity
detection methods for fMRI”, Linkoping
University, Sweden,21 February
2005,pp.13-18.

Ea4301770773

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to Ea4301770773 (20)

Recently uploaded (20)

Ea4301770773