Detection of epilepsy based on discrete wavelet transform and Teager-Kaiser energy operator

This paper presents a novel technique for detection of electroencephalogram (EEG) signals based on discrete wavelet transform (DWT) and Teager-Kaiser energy operator (TKEO). In this study, the EEG signals representing healthy and epileptic seizure activity, taken from an existing database are at first decomposed into different frequency sub bands using DWT. Following this, TKEO is applied on each frequency sub bands and suitable statistical features corresponding to each sub band, in particular mean and standard deviation of TKEO are extracted for effective discrimination of healthy and seizure EEG signals. Finally, the selected features sets are used as inputs to a support vector machines (SVM) classifier for classification of different types of EEG signals. It is been observed that the mean classification accuracy of 99.56% is obtained in discriminating between healthy and seizure EEG signals using polynomial kernel function of SVM classifier, which proves the efficiency of the proposed computer aided diagnostic system (CADS) for detection of epilepsy.