Brain–Machine Interface Using Functional Electrical Stimulation and Motion-Related Cortical Potentials Identified by a Support Vector Machine

This article presents a novel brain-machine interface system that uses electroencephalogram (EEG) and functional electric stimulation (FES) to support neurorehabilitation. Although repetitive facilitation exercise is a remarkable neurorehabilitation method, improving its success rate has proven difficult. This is because therapists are unable to detect patient motor commands and are thus, unable to stimulate the affected part of the body at the same time the motor commands are traveling along neuronal pathways. Here, we investigated the feasibility of both detecting the motor signal and, within milliseconds, stimulating the affected limb via a brain-machine interface system that combines the EEG and FES. The system extracts the timing of the motor commands from the EEG data using a support vector machine algorithm. Three healthy individuals participated in three experiments to: 1) decide the appropriate electric current value for stimulation; 2) determine the appropriate discrimination timing for the motor command detection algorithm; and 3) validate the effect using the system. The validation experiments were carried out under three experimental conditions-no system, fixed FES, and EEG-based FES. The results showed that for all three participants, wrist accelerations were significantly higher for EEG-based FES than for no system or fixed FES. We conclude that the EEG-based FES system detects the timing of motor commands and simultaneously stimulates a patient's affected limb in response.