Visuomotor Transformation with a P300 Brain-Computer Interface Combined with Robotics and Virtual Reality: A Device for Post-Stroke Rehabilitation

Abstract

Brain-computer interfaces (BCIs) are a promising approach to post-stroke rehabilitation. While such treatments have relied mostly on motor imagery, therapies are needed which engage visuomotor transformations. We hypothesized that a P300-based BCI, where flickering stimuli represent motor targets, could provide an effective therapy. To this end, we engineered a rehabilitation system composed of a virtual reality setup, a P300-BCI for decoding targets from EEG, and a robot for moving the stroke-affected arm. This system operates reliably in healthy subjects and stroke patients. Evidence of clinical benefits has been found. Three groups of patients were examined: the BCI group (N=8) where the BCI triggered robotic-driven arm movements, the NoBCI group (N=7) with the same task instructions but no BCI, and the Control group (N=7) without training. On FMA-UE scale, neurological improvements were 1.4±1.6, 23.1±11.8, and 27.6±13.7 points in the Control, NoBCI and BCI groups, respectively; on ARAT scale they were 0.0±0.0, 14.7±20.6, and 20.6±19.3. Our results support the idea that decoding cortical responses to visual targets and using this information to control an orthotic robot could aid post-stroke rehabilitation. We suggest that this method facilitates a visuomotor transformation that normally engages multiple cortical areas and need to be repaired by neuroplasticity.

Keywords: Stroke, Rehabilitation, Brain-Computer Interface (BCI), Robotics, Virtual Reality, Biofeedback, Neurorehabilitation