A Multivariate Approach for Patient Specific EEG Seizure Detection Using Empirical Wavelet Transform

Abstract

This paper investigates the multivariate oscillatory nature of electroencephalogram (EEG) signals in adaptive frequency scales for epileptic seizure detection. Methods: The empirical wavelet transform (EWT) has been explored for the multivariate signals in order to determine the joint instantaneous amplitudes and frequencies in signal adaptive frequency scales. The proposed multivariate extension of EWT has been studied on multivariate multi-component synthetic signal, as well as on multivariate EEG signals of CHB-MIT scalp EEG database. In a moving window based analysis, two seconds long multivariate EEG signal epochs containing five automatically selected channels have been decomposed and three features have been extracted from each one second part of the two-second-long joint instantaneous amplitudes of multivariate EEG signals. The extracted features from each oscillatory level have been processed using a proposed feature processing step and joint features have been computed in order to achieve better discrimination of seizure and seizure-free EEG signal epochs.

Existing System

Neural Network, Fourier Transform.

Proposed System

In this paper, a novel seizure detection algorithm has been proposed, which is able to analyze multivariate non-stationarity EEG signals. The EWT has been explored in a novel way for multivariate signals. The time-frequency representation for multivariate multi-component signals is presented using multichannel extension of EWT. The proposed seizure detection method makes use of five EEG channels selected automatically using a novel channel selection method. Three features have been extracted from different oscillatory levels of multivariate EEG signals followed by a proposed feature processing method. The proposed feature processing method is found very effective to distinguish seizure events in long EEG recordings of hours duration. The reduction of feature dimensionality, prior to classification could be considered as a crucial step for automated classification of seizure and seizure free EEG signals. Finally, the performance of the method has been evaluated using six well known classifiers, which have shown comparable performance among the other seizure detection methods. In future, the proposed method can also be tested in intracranial EEG database of long duration recordings.

Architecture

Proposed multivariate EWT

Feature Processing