Computer aided technique for Epilepsy classification using cross wavelet transform and RBF-kernel based support vector machine

Abstract

Fast and automatic identification and analysis of different bio-medical signals is of growing importance nowadays. This necessitates the application of different computer aided diagnosis methods to interpret, distinguish and analyze various signals and images. In this paper, we have proposed a novel method to identify the Epilepsy from EEG signals. RBF Kernel based Support Vector Machine (SVM) is employed for automatic classification of normal (with closed eyes) and epilepsy patients from their Electroencephalography or EEG signals. Six features are extracted from EEG signals using cross-wavelet transform. Cross-wavelet Transform has not been used before for EEG signal classification. These features are used to train SVM performing binary classification. The average accuracy of SVM based binary classifier is obtained as high as 84.90% in 10-fold cross-validation.