Abstract

Classifying single-trial electroencephalogram (EEG)-based motor imagery (MI) tasks is extensively used to control brain-computer interface (BCI) applications, as a communication bridge between humans and computers. However, the low signal-noise ratio and individual differences of EEG can affect the classification results negatively. In this paper, we propose an improved common spatial pattern (B-CSP) method to extract features for alleviating these adverse effects. Firstly, for different subjects, the method of Bhattacharyya distance is utilized to select the optimal frequency band of each electrode including strong event-related desynchronization (ERD) and event-related synchronization (ERS) patterns; then, the signals of optimal frequency band are decomposed into spatial patterns, and the features that can describe the maximum differences of two classes of MI are extracted from the EEG data. The proposed method is applied in the public data set and experimental data set to extract features which are input into a back propagation neural network (BPNN) classifier to classify single-trial MI EEG. Furthermore, the other two conventional feature extraction methods (original CSP and AR) are used to compare with our proposed method. An improved classification performance in both data sets (public data set: 91.25±1.77% for left hand vs. foot and 84.50±5.42% for left hand vs. right hand, experimental data set: 90.43±4.26% for left hand vs. foot) verify the advantages of B-CSP method over conventional methods. The results demonstrate that our proposed B-CSP method can classify EEG-based MI tasks effectively, and this study provides practical and theoretical approaches to the BCI applications.