Feature Extraction and Classification of EEG Signals for Seizure Detection

Abstract

Epileptic seizure is a neurological disorder characterized by abnormal synchronous discharge of the neuronal activities in the brain structures. These abnormal electrical activities can be recorded via multi-channel electroencephalography (EEG) signals placed on the scalp of the brain. Usually, these signals, recorded from this EEG device, are interpreted by the neurologist which require their availability and it is very time consuming especially for long duration signals. This study presents a fully automatic system for the detection of seizure from non-seizure signals. Firstly, it pre-processes the signal to remove noise and artefacts from the raw-EEG signals and then extracts features. Features are extracted from time-domain, spectral domain, wavelet domain. In addition, connectivity and entropy based feature have also been extracted. After that, prominent features have been selected from this large feature set by a multi-objective evolutionary algorithm and finally, Support Vector Machine (SVM) classifier has been used for classification. A Bayesian optimization algorithm has been used to optimize the hyper-parameters of SVM. Linear Discriminant Analysis (LDA) and Quadratic Linear Discriminant Analysis (QLDA) have also been used for comparison. The proposed system is tested on a publicly available CHB-MIT database and results show the significance of the proposed system. The distinguished accuracy of the classifier is 76.41%, 80.79% and 97.05% in LDA, QLDA and SVM, respectively.