Computer-aided diagnosis of epilepsy using bispectrum of EEG signals

Abstract

This chapter aims to analyze the dynamics of brain activity from the electroencephalogram (EEG) signals and classify the seizures that are responsible for epilepsy. Hence, seizure classification is the primary task of this article. In this study, a nonlinear higher-order spectral method is proposed that significantly explore the underlying dynamics of nonstationary EEG signals. Various statistical parameters are measured from the principal region of the higher-order spectra that are subjected to the data reduction technique of the locality sensitive discriminant analysis (LSDA). The LSDA maps the measured features at higher dimensional space and ranks them according to the probability of discrimination. The ranked features are then used as inputs to the support vector machine (SVM) classifier with radial basis function kernel. The proposed algorithm is simulated on the web-available Bonn university database that achieved excellent seizures classification accuracy.