CLASSIFICATION OF EEG SIGNAL USING ENSEMBLE SYNCHRONIZATION VIA WAVELET TRANSFORM

Abstract

In this research work classification of the subjects using their respective single trial EEG signals is done based on the ensemble synchronization and the phase based on the wavelet transform. Phase synchronization matrix is formed based on the instantaneous parameters of the wavelet-based phase estimation between the EEG channels. Frobenius norm is used for the normalization of the ensemble synchronization so that it can be compared with the other subjects on the 0 to 1 scale. Lower gamma band intrahemispheric area of the brain was studied as from the other research work this band is mainly responsible for the long-range coordination of the cognitive process and thus is an important factor for the classification of the subject into healthy or the schizophrenic one. It was observed that this process classified the subject into their respective category with an accuracy of 75% as compared to the other models using the Hilbert transform based which gave an accuracy of about 70%. This is due to the time frequency localization property of the wavelet transform.