Automated identification system for focal EEG signals using fractal dimension of FAWT-based sub-bands signals

Abstract

The classification of focal and non-focal electroencephalogram (EEG) signals for diagnosis of epilepsy at an early stage is one of the most difficult problems. There have been many attempts to develop automated detection algorithms to assist clinical research for presurgical analysis of epilepsy. In this paper, a novel approach for studying EEG signals has been proposed using flexible analytic wavelet transform (FAWT) which is a nonstationary signal processing technique. In this study, EEG signals are decomposed into the desired number of sub-bands (SBs). Fractal dimension (FD) is used as a feature and then computed it for all SB signals which are obtained from FAWT. The significant features obtained from the Kruskal–Wallis statistical test and are classified using robust energy-based least square twin support vector machine (RELS-TSVM). In order to show the effectiveness of the proposed method for classification of focal (F) and non-focal (NF) EEG signals, publicly available database termed as Bern-Barcelona EEG dataset is used for the study. © Springer Nature Singapore Pte Ltd 2019.