A Multicentre Randomized Single Blind Placebo Controlled Study to Assess Efficacy of Hand Exoskeleton Controlled by Motor Imagery Based BCI for Post Stroke Patients Movement Rehabilitation

The BCI training is based on EEG activity patterns recording during hand motor imagery. A subject sit comfortably 1m from a computer screen which presents visual instructions during the session. The robotic exoskeleton is fixed to the paretic hand. Subject visually fixate on a circle presented in the center of the screen and receive instructions from three surrounding rhomboidal arrows. Subject is given three commands instructing them to relax or imagine slow hand opening movement with the right or left hand. The "Relax" command means that the subject have to sit still and look at the center of the screen. Commands are presented randomly, each of 10 sec duration.

A visual cue provides the subject with feedback regarding the mental task recognition: the central circle turned green if the classifier recognizes the task in agreement with the given command, or remains white if the signal is not recognized. In addition to visual feedback the subject is provided with kinesthetic feedback: the exoskeleton is opening the hand when the classifier recognizes the imagery of paretic hand movement.

The EEG is registered with 30 electrodes distributed over the head in accordance with the standard international 10-20 system. EEG signals are filtered from 5-30Hz. A Bayesian approach for EEG pattern classifying is used in the system. The activity sources most relevant for BCI functioning will be identified using an independent component analysis (ICA). Classification accuracy will by measured with Cohen's kappa (Kohavi and Provost, 1998). The procedure may consist from up to three sessions, the duration of 1 session is 10 min, there are 5 min time brake between session and total duration of procedure is up to 45 min. The number of procedures - at least 10. The maximal interval between procedures is 3 days.