Wrist movement discrimination in single-trial EEG for Brain–Computer Interface using band powers

This paper proposes the use of Wavelet Transform (WT)-based descriptive features for analysing the Electroencephalogram (EEG) data for wrist movements. The interest is in analysing the EEG data to determine the left and right wrist movements using a small set of features. The methodology uses wavelet decomposition up to a specified decomposition level and calculates Power Spectral Density (PSD) in each subband. Features extracted from these bands are input to the Radial Basis Function (RBF) classifier to test classification accuracy. Results reveal the existence of significant change in power distribution in the two types of wrist movements. The classification rate was up to 91% with an average of 86% in the four subjects. These results show further improvement in recognition rate when compared with the groups’ effort earlier on same database.