Multiclass epileptic seizure classification using time-frequency analysis of EEG signals

Seizure is a transient abnormal behavior of neurons within one or several neural networks, which limits the patients physical and mental activities. Since conventional time or frequency domain analysis is found inadequate to describe the characteristics of a non-stationary signal, such as electroen-cephalography (EEG), in this paper, we propose to transform the EEG data using twelve Cohen class kernel functions in order to facilitate the time-frequency analysis. The transformed data thus obtained is exploited to formulate a feature vector consists of modular energy and modular entropy that can better model the time-frequency behavior of the EEG data. The feature vector is fed to an Artificial Neural Network (ANN) classifier in order to classify epileptic seizure data originating from different parts and state of the brain. A number of simulations is carried out using a benchmark EEG dataset. It is shown that the proposed method is capable of producing greater accuracy in comparison to that obtained by using a state-of-the-art method of epileptic seizure classification using the same EEG dataset and classifier.