Desynchronization Network Analysis for the Recognition of Imagined Movement

This paper reports on the use of electroencephalogram (EEG)-based phase desynchronization networks for the recognition of imagined movements. Features derived solely from these networks are classified using linear support vector machine. An average accuracy of 73% is achieved for the single-trial imagined hand versus foot movements. The results demonstrate that phase desynchronizations provide relevant information for the discrimination of mental tasks. This novel approach will potentially benefit the development of brain-computer interfaces