Epileptic Seizure Detection using An EEG Sensor

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 03 | Mar -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 1668
EPILEPTIC SEIZURE DETECTION USING AN EEG SENSOR
Prof. Sandhya Shinde1, Shradha Chavan2
1Assistant Professor, E&TC, Dr. D.Y Patil Institute of Engg, Management and Research, Pune, Maharashtra, India.
2Student, E&TC, Dr. D.Y Patil Institute of Engg, Management and Research, Pune, Maharashtra, India.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Epilepsy is an inveterate neurological disorder which is
causedbyunusualnervecellactivityinthe brainandischaracterizedby
consecutiveunexpectedseizures.Preciselyindentifyingandenumeratingthe
seizures in patients with epilepsy is essential for diagnosis, selecting the
treatmentandestimatingtheeffectsofthetherapy.Epilepsydetectiondone
justbyphysicallyanatomizingaperson’sbodyisaveryarduousjob. The
braininvolvespeculiarnervecellactivitywhichcanbeanalyzedwiththehelp
ofanEEGsensortodetectepilepticseizureinasubjectasitmeasuresthe
voltagefluctuationsresultingfromioniccurrentwithintheneuronsofthe
brain.Thispaperpresentsanapproachtoanalyzethebrainsignalusingan
EEG sensor and perform various signal processing techniques on it in
MATLAB, detect its high frequency components by wavelet analysis and
compareitwiththedatabasesignaltodetectandclassifytheseizurewiththe
helpofSVM.
Key Words: Epilepsy, EEG (Electroencephalogram),
seizure, SVM classifier, DWT.
1. INTRODUCTION
The brain is one of the crucial organs of the human
body. Epilepsy is a non-contagious, chronic neurological
disorder of the cerebral nervous system (brain) that affects
people of all ages with over 50 million patients worldwide.
Globally, each year an estimated of about 2.4 million people
are diagnosed with epilepsy [1]. The chronic peculiar bursts
of electrical discharge in the brain causes a severe disorder
of the cerebral nervous system resulting in recurrent,
unprovoked epileptic seizures. During an epileptic seizure,
the patient may be struck by numerous symptoms such as
loss of consciousness, involuntary movements,
uncontrollable twitching. Some epileptic seizures can be
milder than the others, but even minor seizures can be
perilous if occurred during activities like swimming or
driving. The seizures are classified based on the part of the
brain affected. Focal (partial) seizures and generalized
seizures are the two main types of seizures, in a partial
seizure the epileptic activity affects only one part of the
brain whereas in a generalized seizure the entire brain is
affected during the epileptic activity [2]. The classificationof
the epileptic seizure is as shown in Fig -1.
Epilepticseizurescanbe efficientlycontrolled bythe
use of appropriate medications but about 30% of the
patients do not have seizure control at the proper time.
Fig -1: Classification of Epileptic Seizure.
Also the exact type of seizure should be recognized for apt
diagnosis. Prediction of epileptic seizures at an initial stage
increases the effect of medication and more patients can be
treated accordingly thus improving the quality of life of the
patients. For the proper treatment, doctors need to know if
and when the seizure occurs as many medical decisions
depend on detailed information about the seizure type and
its origin in the brain. EEG monitoring isthegoldenstandard
for the diagnosis of epilepsy, it aids in the appropriate
classification and detection of the seizure. An EEG
investigation provides the aforementioned information
about the continuous unusual nerve cell activity in the brain
and a detailed seizure characterization in order to resolve
therapeutic options, particularlyintheabsenceofa response
to medication. During an EEG investigation, the physicians
place electrodes on the scalp which sense and record the
impulsive electrical motion taking place in your brain which
is then examined to find unusual activity, which may signal
epilepsy. The Fig-2 shows a typical EEG recording setup
wherein the electrodes are placed on the scalp.
Fig -2: EEG Electrodes placed on scalp.

1.1 Electroencephalogram:
The brain is the most multifaceted of all the other
organs of the human body, producing electricalsignalstorun
the entire body directly or indirectly. The electricalactivityis
generated in the brain by millions of neurons which are
recorded by an Electroencephalogram. EEG is a good, non-
invasive diagnostic tool that can be digitally anatomized to
detect the disorder. The EEG sensor encompassed in theEEG
headset is used asa Brain-Computer Interface (BCI) toolthat
reports the wearer’s mental state, along with raw waves and
information about the brainwave frequency bands. Two
electrodes are placed on the scalp to form an EEG channel
which measures the potential difference between the
electrodes and then records the summed potential of
neurons. A patient suffering from a seizure has a distinctly
different EEG as compared to that of the normal ones. Thus
by processing the raw waves acquired from the EEG headset
with the help of MATLAB, we can classify and detect the
epileptic seizure. The first five seconds of EEG data segments
from three different groups of data sets (normal EEG data –
healthy subject, interictal EEG data – epileptic subject during
seizure free interval, ictal EEG data – epileptic subject during
a seizure) are as shown in Fig-3.
Fig -3: Segments of EEG data: (a) Normal, (b) Interictal, (c)
Ictal.
1.2 Objective of the project:
The main objective of our project is to build a
lucrative and consumer affable system that will assist
doctors, patients and parents in supervising the epileptic
seizures. This system will detect the epilepsy even when the
patient is not having a seizure at that instance but have had
it in the past.
2. Proposed work:
The proposed plan of action of this project is to use
an EEG headset comprised of EEG sensors to detect the raw
data from the brain waves and process these raw waves
using signal processing techniques, extract features and
classify them into seizure and seizure-free sets using SVM
classifier in MATLAB. The block diagram consisting of the
hardware and software part is as shown in the Fig-4. The
hardware part consists of the EEG headset which will detect
the high frequency components of the brain waves in the
form of raw data and further send those signals for
processing; the signal processing comes under the software
part. The signal processing techniques involves the basic
steps of filtering these raw waves in order to remove noise
and artifacts associated with the irrelevant physiological
activity [3]. These techniques involve the use of low pass
filter and a notch filter. These processed signals are further
sent for feature extraction to the DWT.
Fig -4: Block Diagram.
2.1 Discrete Wavelet Transform (DWT):
There is a considerable trade-off between the time
and frequency resolutions in the traditional methods of
Fourier analysis, thus leading to show more accurateresults
with increased sensitivity and specificity in the wavelet
decomposition (wavelet transform). The wavelet
decomposition allows the best possible selection of
decomposition levels with least amount of entropy values,
high power spectral density, pertaining to epileptic
waveform [3].TheDWTdecomposesthesignal into mutually
orthogonal set of wavelets meaning that the wavelets are
sampled discretely. The EEG signal is decomposed into its
sub-bands by discrete wavelet transform to extract features
from the theta, alpha and beta bands i.e. it is used to split the
signal into five frequency bands. Variousfeaturessuchasthe
Energy, Zero Crossing Rate (ZCR), Variance and Fractal
dimensions can be extracted from the sub-bands.Thevalues

of these features are fed to a Support Vector Machine to
separate the classes into epileptic and non-epileptic seizures
after performing statistical analysis on it. Statistical analysis
and SVM classifier are used for the computation of datasets
that are too vast to be stored in the memory. The statistical
toolbox aids us to identify variables and features by
providing feature choice, stepwise reversion, principal
component analysis (PCA), formalization and additional
dimensionality reduction methods.
2.2 Support Vector Machine (SVM):
The Support Vector Machine is used for the
classification of the EEG signals, it must be able to
differentiate between the electrical activity of a healthy
subject and an epileptic one. In SVM, we plot each data item
as a point in n-dimensional space with the value of each
feature being the value of a specific coordinate, and then the
classification is done. The SVM primarily consists of
constructing an optimum hyperplane that maximizes the
margin of separation amid two different classes. To do so a
kernel is used to transform the input data to a higher
dimensional space followed by an optimization step for the
construction of an optimum hyperplane [4]. Since we are
using raw data from the brain waves we will have non-
linearly separable data, for which we will create non-linear
decision boundaries using kernel. A kernel is basically a
function which converts non-separable problem to a
separable problem by simply transforming low dimensional
input space to a higher dimensional input space. The main
motive behind the kernel is to map the data into different
feature space in order to construct linear or non-linear
classifiers in the original space. It does some extremely
complex data transformations and then finds outtheprocess
to segregate the data based on the labels or outputs we have
defined. The classifier will thus be able to classify the seizure
precisely and at the same time be robust with respect to EEG
signal variations across assorted mental states and subjects.
The SVM classifier used for the epileptic seizure detection
should be trained, cross validated and tested with the
extracted features using DWT of the EEG signals obtained
from healthy (non-epileptic) and epileptic subjects
comprised in our database.
Fig -5: Classification of EEG from Nonlinear into a Linear.
3. Flowchart:
The work flow of the project is as shown in Fig -6.
Fig -6: Flow Chart
4. Advantages:
 It is a robust device for tracking brain changes
during different phases of life, including evaluating
adolescent brain maturation and indicating
noteworthy facets of the timing of brain
development.
 EEG is tolerant to a relative extent of subject
movement and it also includes artefacts for
eliminating movement in EEG data.
 It prevents inadequate availability of technologists
to provide abrupt care in high traffic hospitals.
 It is a non-invasive technique with high temporal
resolution.
 The hardware costs are significantly lower of this
system than compared with most of the other
techniques available.

5. Disadvantages:
 Relatively many subjects are needed to extract
constructive information from EEG and
sophisticated data analysis is required since the
signal-to-noise ratio is poor.
 EEG requires intense interpretation just to
postulate which parts are activated by a specific
response.
6. Applications:
 To distinguish between the epileptic seizures and
other sorts of spells such as psychogenic non-
epileptic seizures, syncope (fainting), sub-cortical
movement disorders and migraine variants.
 It serves as an adjunct test of brain death.
 To prognosticate, in certain cases, in patients with
coma.
 To decide whether to wean anti-epileptic
medications.
7. CONCLUSIONS
This paper presents a system which extracts large
number of features from the EEG data after wavelet
transformation and performs statistical analysis before
sending it to the SVM classifier to make an objectivedecision
about the EEG data processed. This system will be able to
efficiently differentiate the epileptic features from the
normal EEG signals with high accuracy and the least false
rate will influence its ability to correctly predict the onset of
the seizure.
REFERENCES
[1] World Health Organization, “Epilepsy.”
(http://www.who.int/mediacentre/factsheets/fs999/en/).
[2] Kimberly Holland, “Epilepsy by the Numbers: Facts,
Statistics and You.”
(http://www.healthline.com/health/epilepsy/facts-
statistics-infographic).
[3] Kavya Devarajan, E. Jyostna, K. Jayasri, “EEG-Based
Epilepsy DetectionandPrediction”,IACSIT International
Journal of Engineering and Technology, Vol. 6, No. 3,
June 2014.
[4] N. Mammone, F. La Foresta, and F.C Morabito,
“Automatic artifact rejection from multichannel scalp
EEG by wavelet ICA,” IEEE Sensors J., vol. 12, no. 3, pp.
533-542, Mar. 2012.
[5] A. Kumar and M. H. Kolekar,“Machinelearningapproach
for epileptic seizure detection using wavelet analysis of
EEG signals,” Medical Imaging, m-Health and Emerging
Communications Systems (MedCom), 2014
International Conference on, Greater Noida, 2014, pp.
412-416. doi: 10.1109/MedCom.2014.7006043.
[6] S. Sanei and J.A. Chambers, “EEG Signal Processing,”
Centre of Digital Signal Processing, Cardiff University,
UK, John Wiley & Sons, Ltd., 2007.
[7] A. S. Zandi, G. a Dumont, M. Javidan, R. Tafreshi, B. a
MacLeod, C. R. Ries, and E. Puil, “A novel wavelet-based
index to detect epileptic seizures using scalp EEG
signals”, Conf. Proc.EngineeringinMedicineandBiology
Society, vol. 2008, no. 2, (2008), pp. 919–922.
[8] A. Shoeb and J. Guttag, “Application of MachineLearning
To Epileptic Seizure Detection”, Proceedings of the27th
International Conference on Machine Learning, (2010)
pp. 975–982.

Epileptic Seizure Detection using An EEG Sensor

More Related Content

What's hot (20)

Similar to Epileptic Seizure Detection using An EEG Sensor (20)

More from IRJET Journal (20)

Recently uploaded (20)

Epileptic Seizure Detection using An EEG Sensor