A Motor Imagery Based Brain-Computer Interface for Stroke Rehabilitation

Brain-Computer Interfaces (BCIs) have been used to assist people with impairments since many years. In most of these applications the BCI is intended to substitute functions the user is no longer able to perform without help. For example BCIs could be used for communication and for control of devices like robotic arms, wheelchairs or also orthoses and prostheses. Another approach is not to replace the motor function itself by controlling a BCI, but to utilize a BCI for rehabilitation that enables the user to restore normal or "more normal" motor function. Motor imagery (MI) itself is a common strategy for motor rehabilitation in stroke patients. The idea of this paper is it to assist the MI by presenting online feedback about the imagination to the user. A BCI is presented that classifies MI of the left hand versus the right hand. Feedback is given to the user with two different strategies. One time by an abstract bar feedback, and the second time by a 3-D virtual reality environment: The left and right hand of an avatar in the 1st person's perspective in presented to him/her. If a motor imagery is detected, the according hand of the avatar moves. Preliminary tests were done on three healthy subjects. Offline analysis was then performed to (1) demonstrate the feasibility of the new, immersive, 3-D feedback strategy, (2) to compare it with the quite common bar feedback strategy and (3) to optimize the classification algorithm that detects the MI.