EEG-Rhythm Specific Taylor–Fourier Filter Bank Implemented With O-Splines for the Detection of Epilepsy Using EEG Signals

Abstract

The neurological disorder which is associated with the abnormal electrical activity generated from the brain causing seizures is typically termed as epilepsy. The automated detection and classification of epilepsy based on the analysis of the electroencephalogram (EEG) signal are highly required for its early diagnosis. In this paper, we have developed an EEG-rhythm specific Taylor-Fourier filter-bank implemented with O-splines for the detection and classification of epilepsy from the EEG signal. The energy features are evaluated from the Taylor-Fourier sub-band signals of the EEG signal. The classifiers such as K-nearest neighbor (KNN) and least square support vector machine (SVM) are employed for the classification of normal, seizure-free and seizure from the Taylor-Fourier EEG-band energy (TFEBE) features. The experimental results demonstrate that, for the classification of normal, seizure-free, and seizure classes, the least square SVM classifier has an overall accuracy value of 94.88% using the EEG signals from the Bonn university database. The proposed EEG rhythm specific Taylor-Fourier filter-bank with O-splines can be implemented in real-time for the detection of epileptic seizures from EEG signals.