Classification of motor imagery movements using multivariate empirical mode decomposition and short time Fourier transform based hybrid method

Abstract

Effective online processing of electroencephalogram (EEG) signals is a prerequisite of brain computer interfacing (BCI). In this paper, we propose a hybrid method consisting of multivariate empirical mode decomposition (MEMD) and short time Fourier transform (STFT) to identify left and right hand imaginary movements from EEG signals. Experiments are carried out using the publicly available benchmark BCI competition II Graz motor imagery data base. The EEG epochs are decomposed into multiple intrinsic mode functions (IMFs) by applying MEMD. The most significant mode is subjected to the short time Fourier transform; the peak of the magnitude spectrum is used as feature representing the corresponding epoch. The efficacy of the proposed feature extraction scheme is demonstrated by intuitive, statistical and graphical analyses. The performance of the proposed feature extraction scheme is investigated for various choices of classifiers. Our findings suggest that k-Nearest Neighbor (kNN) emerges as the best classification model yielding 90.71% accuracy. The performance of our method is also compared to that of existing works in the literature. Experimental outcomes backed by statistical validation manifest that the performance of the proposed method is comparable or better than many of the state-of-the-art algorithms.