Development of a Brain-machine Interface for Stroke Rehabilitation Using Event-related Desynchronization and Proprioceptive Feedback

Abstract

We have developed a brain-machine interface (BMI) rehabilitation system for patients with stroke and motor paralysis, which provides proprioceptive feedback upon successful generation of motor-imagery (MI)-induced event-related desynchronization (ERD) and a decrease in mu band (8-13 Hz) activity derived from hand motor imagery. This system consists of an electroencephalogram (EEG) ampli er operated using the MATLAB Simulink software; a pneumatic robotic exoskeleton to provide proprioceptive feedback to the paralyzed hand; and a tablet computer placed over the paralyzed hand to display a hand-action movie to facilitate ERD generation. The EEG ampli er was connected and synchronized via the exoskeleton and tablet computer with an Arduino microcomputer. Nine patients in the subacute stage of recovery after stroke participated in a neurofeedback training experiment, which employed the aforementioned system. During the 4 weeks of this study, the participants received 2 weeks of BMI-based or control interventions in a random and counterbalanced order, in addition to their daily conventional physiotherapy. The control intervention consisted of the same MI training as the BMI-based intervention, but the exoskeleton always provided proprioceptive feedback regardless of the ERD strength. The ERD strength in the affected hemisphere showed a desirable increase with a signi cant improvement of nger joint spasticity, only after the DMB-based intervention period, and not after the control intervention period. The proposed neurofeedback training can help patients with stroke and movement disorders, because increased ERD strength may lead to recovery of motor function.