Automatic feature selection based motor imagery movements detection scheme from EEG signals in the Dual Tree Complex Wavelet Transform domain

An automatic feature selection based classification scheme in the Dual Tree Complex Wavelet Transform (DTCWT) domain from electroencephalogram (EEG) signals has been presented in this study to identify left and right hand imagery movements. First, the EEG epochs are decomposed into different real and imaginary coefficient bands and then, some statistical features like norm entropy and mean absolute deviation (MAD) have been calculated. These features are combined into a single feature space and after that optimal features have been selected automatically imposing some feature selection criteria from this combined feature space. The selected features have been justified as suitable to classify different kinds of motor imagery EEG signals by statistical hypothesis testing (e.g., one way ANOVA) and graphical analyses (e.g., scatter plots). Finally, K-nearest neighbor (kNN) based classifiers are developed using these selected features for classifying 2 types of imagery hand movements. 90.36% overall accuracy is achieved in publicly available BCI competition II Graz data set which is shown to be superior than several existing methods.