Epilepsy is a chronic neurological disorder which is
identified by successive unexpected seizures. Electroencephalogram (EEG) is the electrical signal of brain which contains valuable information about its normal or epileptic activity. In this work EEG and its frequency sub-bands have been analyzed to detect epileptic seizures. A discrete wavelet transform (DWT) has been applied to decompose the EEG into its sub- bands.Statistical features Energy, Covariance Inter- quartile range (IQR) and Median Absolute Deviation (MAD) are calculated for each sub-band. The extracted features are applied tofuzzy system for classifications Accuracies of 2 types of member ship functions are compared. and got classication accuracy of 97%.
Epilepsy is a neurological disorder with prevalence of about 1-2% of the worlds population It is characterized by sudden recurrent and transient disturbances of perception or behavior resulting from excessive synchronization of cortical neuronal networks; it is a neurological condition in which an individual experiences chronic abnormal bursts of electrical discharges in the brain. The hallmark of epilepsy is recurrent seizures termed "epileptic seizures". Epileptic seizures are divided by their clinical manifestation into partial or focal, generalized, unilateral and unclassified seizures Focal epileptic seizures involve only part of cerebral hemisphere and produce symptoms in corresponding parts of the body or in some related mental functions. Generalized epileptic seizures involve the entire brain and produce bilateral motor symptoms usually with loss of consciousness. Both types of epileptic seizures can occur at all ages. Generalized epileptic seizures can be subdivided into absence (petit mal) and tonic-clonic (grand mal) seizures . Monitoring brain activity through the electroencephalogram (EEG) has become an important tool in the diagnosis of epilepsy. The EEG recordings of patients suffering from epilepsy show two categories of abnormal activity inter- ictal, abnormal signals recorded between epileptic seizures; and ictal, the activity recorded during an epileptic seizure (Fig. 1). The EEG signature of an inter-ictal activity is occasional transient waveforms, as isolated spikes, spike trains, sharp waves or spike-wave complexes. EEG signature of an epileptic seizure (ictal period) is composed of a continuous discharge of polymorphic waveforms of variable amplitude and frequency, spike and sharp wave complexes, rhythmic hyper synchrony, or electro cerebral inactivity observed over a duration longer than the average duration of these abnormalities during inter-ictal periods
Figure 1. EEG signal with Epileptic Seizures
There are number of researches present in literature and Still going on regarding automated detection of seizures. The work proposed by Gotman [1] was the first widely applicable method for seizure detection. Later on modified in 1990 [2]. This method detected 70% 80% of seizures. S.R. Mousavi et al. [3] have introduced a new method based on autoregressive model, with classification claimed at the range of 91% -96%. Khan & Gotman [4] proposed a wavelet based method to capture rhythmic nature of seizure discharges. The sensitivity of detection was about 90%.The work proposed by Omar Farooq [5] wavelet based technique is adopted to extract features.
Classication of epileptiform activity in the human electroencephalogram-based Wavelet Transforms and Fuzzy logic
Shaik.Jakeer Husain ,Associate professor Departmet of Electronics and CommunicationEngineering VIGNAN INSTITUTE OF TECHNOLOGY AND SCIENCE Hyderabad E-mail: jk.shaik@gmaill.com
Dr.K.S.Rao Director CVSR College of Engineering Hyderabad
In this work the EEG signal has been decomposed into five sub-bands by discrete wavelet transform (DWT)and Mandeni fuzzy system model for the classification.A fuzzy system was implemented to classify the EEG signal to one of the categories: epileptic or normal. The aim of this work was to develop a simple algorithm for the detection of epileptic seizure, which could also be applied to real-time.
2. MATERIALS AND METHODS
2.1. Description of the EEG Database
EEG data considered for this work is obtained from University of Bonn EEG database which is available in public domain [7]. containsthree differentcases:1) healthy,2)epilepticsubjects duringseizure- freeinterval(interictal),3)epileptic subjectsduring seizureinterval (ictal) [7].Eachcasehasfivedatasets named:O,Z,F,N,andS. SetsOandZareobtainedfrom healthysubjectsunder condition ofeyesopenand closed;respectivelybyexternalsurface electrodes.SetsFandNare attainedfrom interictal subjects. SetF taken from epileptogenic zone of the brain showsfocalinterictalactivity;setNobtainedfrom hip- pocampalformationofthe oppositehemisphereofthe brain indicates non-focal interictal activity, and set S is gotfromanictal subject. Eachsetcontains100singlechannelEEGsegments of23.6secduration.Samplingfrequencyis173.61Hz, soeach segmentcontainsN= 4096samples[7].All these EEGsegmentsarerecorded withthesame128-channel amplifierthatconverts by 12A/Dconvertorwith bitrate of 12,andthen weresampled on 173.61 Hz [7].
2.2. Wavelet Decomposition Wavelet transform is a spectral estimation technique in which any general function can be expressed as an infinite series of wavelets. The basic idea underlying wavelet analysis consists of expressing a signal as a linear combination of a particular set of functions (wavelet transform, WT), obtained by shifting and dilating one single function called a mother wavelet. The decomposition of the signal leads to a set of coefficients called wavelet coefficients. Therefore the signal can be reconstructed as a linear combination of the wavelet functions weighted by the wavelet coefficients. In order to obtain an exact reconstruction of the signal, adequate number of coefficients must be computed. The key feature of wavelets is the time-frequency localization.
In this work fifth-order Daubechies (DB5) DWT has been applied to the band-limited EEG (0-60 Hz). After the first level of decomposition, the band-limited EEG has been decomposed into its high resolution frequency band, D1 (30-60 Hz), and low resolution frequency band, A1 (0-30 Hz), which should be decomposed in next level. In the second level of decomposition, A1 has been de- composed into its high, D2 (15-30), and low, A2 (0-15 Hz) resolution bands. This process has been repeated four times. After full decomposition five sub-bands have been attained: high frequency sub-bands (details) of levels 1 to 4 (D1 (30-60 Hz), D2 (15-30 Hz), D3 (8-15 Hz), D4 (4-8 Hz)) as well as the low frequency sub-band (approximate) of the last level (A4 (0-4 Hz)).Figure 2 illustrates this multi- level decomposition process schematically. These five frequency sub-bands are almost corresponding to five physiological EEG bands, delta (0-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-30), and gamma (30-60 Hz).
Figure 2 Schematic of multi-level decomposition.
2.3 Feature Extraction The extracted wavelet coefficients provide a compact representation that shows the energy distribution of the EEG signal in time and frequency.The following statistical features were used to represent the time frequency distribution of the EEG signals:
I. Energy of the wavelet coefficients in each sub- band E(l)= 2 1 n i i x (1) Where, xirepresents thevalueofsignal, nis thetotalnumberofsamplesandlrepresentsthedecomposedlevel . II. Coefficient of variation of the wavelet coefficients in each sub-band EEG(0-60Hz) ( LPF A1(0-30Hz)
HPF D1(30-60Hz)
LPF A2(0-15Hz)
HPF D2(15-30Hz)
LPF A3(0-8Hz)
HPF D3(8-15Hz)
LPF A4(0-4Hz)
HPF D4(4-8Hz)
2 2 ov C (2)
III. Interquartile range of the wavelet coefficients in each sub-band TheIQRisdefinedby IQR=Q3Q1 (3) Where,Q1 andQ3arethefirst andthirdquartile respectively.
IV. Median absolute deviation of the wavelet coefficients in each sub-band Themedianabsolutedeviationisthemean of the absolutedeviationsofasetofdataaboutthedata'smean.For asamplesizeNandthemean distributionx,themedian absolutedeviationisdefinedby(MAD) 1 1 | | N i i x x N .(4)
2.4Fuzzy Techniques
Fuzzy systems are being used successfully in an increasing number of application areas. They are linguistic rules to describe a system. The rule based systems are more suitable for complex systems where it is very difficult, but not impossible to describe the system mathematically The Mamdani fuzzy system with four inputs and one output is the basic fuzzy model. The centre of gravity defuzzification method is employedThe following cascaded block diagram shows the generalized System for the classification of epilepsy.
Figure 3 Block Diagram of Epileptic Seizure Classification systems
Table-I Range of values of fuzzy variables Variable Range Low Medium High Energy (Input) 0 to 90000 0 to 2000 1500 to 5000 4500 to 90000 Covariance (Input) 0 to 6000 0 to 100 90 to 600 185 to 5895 IQR (Input) 0 to 600 0 to 25 20 to 75 60 to 300 MAD (Input) 0 to 300 0 to 15 10 to 40 30 to 300 Epilepsy Level(Output) 0 to 1 0to 0.2 0.1 to 0.5 0.4 to 1 Figure 4 to 8 shows Input/ Output membership functions Of the Fuzzy System with linguistic members such as low, medium and high, the range of values are chosen as per table-I. Total eight rules are formed Figure 9 shows the view of all rules
EEG Signal Pre- processing Feature Extraction Classif ication
3. RESULTS & DISCUSSION
The performance of the classification is evaluated in terms of specificity,Sensitivityand classification accuracy which are defined as:
Specificity =
Sensitivity=
Overall classification accuracy (CA) %=
Table-II Methods Specificity (%) Sensitivity (%) classification accuracy(%) Fuzzy logic With triangular membership function
90
94
92 Fuzzy logic With Bell membership function
97
97
97
4. CONCLUSION In this paper, we presented a method of classifying epileptic seizure using fuzzy logic system EEG signals are decomposed using discrete wavelet transforms (db5)and Sub-bands(delta, theta, alpha, beta, and gamma) coefficients are used for feature extraction(Energy,Covariance,IQR and MAD). The extracted features are applied to The Mamdani fuzzy system with four inputs and one output is the basic fuzzy model. Triangular andbell shaped membership function are used. Performance evaluation done in terms of specificity, sensitivity and classification accuracy (Table II).
The classification performance obtained by fuzzy model is very competitive. But, there are still many issues that need to be considered in future research The performance of this can be further enhanced by using Adaptive Neural Fuzzy InferenceSystems (ANFIS) and Aggregation operator based fuzzy classification.
5. References
[1] J. Gotman., Automatic recognition of epileptic seizures in the EEG, Clinical Neurophysiology, vol. 54, pp. 530540, 1982 [2] J.Gotman., Automatic seizure detection: improvements and evaluation, Clinical Neurophysiology, vol. 76, pp. 317324, 1990. [3] S.R. Mousavi, M. Nikanazar, B. V. Vahdat, Epileptic Seizure detection using AR model on EEG signals, Cairo International Biomedical Engineering Conference, (CIBEC), pp. 1 -4, 2008. [4] Y.U. Khan, J. Gotman, Electroencephalogram Wavelet based automatic seizure detection iintracerebral, Clinical Neurophysiology, vol. 114, pp. 899-908, 2003 [5] YNidal Rafiuddin1Yusuf Uzzaman Khan and Omar Farooq,Feature Extraction and Classification of EEG for Automatic Seizure Detection, IEEE International Conference on Multimedia, Signal Processing and Communication Technologies, pp. 184-187,2011.
[6] Ahmad Mirzaei, Ahmad Ayatollahi, Hamed Vavadi "Statistical analysis of epileptic activities based on histogram and wavelet-spectral entropy"J. Biomedical Science and Engineering, 2011, 4, 207- 213 [7] EEG time series (epileptic data) (2005, Nov.) [Online],http://www.meb.unibonn.de/epileptologie/ science/physik/eegdata.html
[8] R.Harikumar, Dr. (Mrs).R.SukaneshFuzzy Techniques for Classification of Epilepsy risk level in Diabetic Patients Using Cerebral Blood Flowand Aggregation Operators
International League Against Epilepsy Classification and Definition of Epilepsy Syndromes With Onset in Childhood Position Paper by The ILAE Task Force On Nosology and Definitions