Epileptic seizure detection using DWT based fuzzy approximate entropy and support vector machine

AbstractEpilepsy is a common neurological condition which affects the central nerve system that causes people to have a seizure and can be assessed by electroencephalogram (EEG). A wavelet based fuzzy approximate entropy (fApEn) method is presented for the classification of electroencephalogram (EEG) signals into healthy/interictal versus ictal EEGs. Discrete wavelet transform is used to decompose the EEG signals into
different sub-bands. The fuzzy approximate entropy of different sub-bands is employed to measure the chaotic dynamics of the EEG signals. In this work it is observed that the quantitative value of fuzzy approximate entropy drops during the ictal period which proves that the epileptic EEG signal is more ordered than the EEG signal of a normal subject. The fApEn values of different sub-bands of all the data sets are used to form feature vectors and these vectors are used as inputs to classifiers. The classification accuracies of radial basis function based support vector machine (SVMRBF) and linear basis function based support vector machine (SVML) are compared. The fApEn feature of different sub-bands (D1 – D5, A5) and classifiers is desired to correctly discriminate between three types of EEGs. It is revealed that the highest classification accuracy (100%) for normal subject data versus epileptic data is obtained by SVMRBF; however, the corresponding accuracy between normal subject data and epileptic data using SVML is obtained as 99.3% and 99.65% for the eyes open and eyes closed conditions, respectively. The similar accuracies, while comparing the interictal and ictal data, are obtained as 99.6% and 95.85% using the SVMRBF and SVML classifiers, respectively. These accuracies are not 100%; however, these are quite higher than earlier results published. The results are discussed quite in detail towards the last section of the present paper < final year projects >