P. D. Bobrov, D. Husek, A. V. Korshakov, A. A. Frolov
The paper examines the sources of brain activity, contributing to EEG patterns, corresponding to motor imagination. The accuracy of their classifying specifies the efficiency of brain-computer interface (BCI), allowing to control external technical devices directly by signals of brain activity instead of natural muscle forces. The sources of brain activity corresponding to motor imagination are revealed by Independent Component Analysis (ICA). EEG during motor imagination is shown to be a superposition of components dependent and not dependent of the type of imaginary movement. Components dependent on the type of imaginary movement are relevant for EEG pattern classifying. They are localized in central brain areas. Components not dependent of the type of imaginary movement are localized in occipital areas. They relate to the attention to task performance independently of the imaginary movement and not relevant for BCI efficiency. Excluding of these components from EEG patterns improves essentially the BCI performance.