Classification of Electroencephalographic Patterns of Imaginary One-hand Finger Movements for Brain-Computer Interface Development

The results of kinesthetic motor imagery EEG-pattern classification of one hand fingers and wrist movements executed in a given rhythm are presented in this study. The classifiers were based on the support vector machine method and on the developed neural network committee. It was shown that the accuracy of pairwise EEG-pattern classification of imaginary movements by means of the neural network committee was higher on average than the accuracy of the support vector machine classifier. The possibility of improving the accuracy of fine motor imaginary classification was revealed with the help of individual approach implementation for selection of EEG-pattern classification parameters.