Design and comparative evaluation of a BCI-based upper extremity robotic rehabilitation protocol

Abstract

We advocate online modification of robot-assisted task speed, based on continuously inferred motor imagery as an effective rehabilitation protocol for increasing the involvement levels of the patients in physical rehabilitation exercises. To study efficacy of such Brain-Computer Interface (BCI) based physical rehabilitation protocols, we conduct human subject experiments on healthy volunteers, comparing several BCI-based protocols with haptic and visual feedback with each other and with conventional robot-assisted rehabilitation protocols, in terms of intensity and sustainability of motor imagery. Our results provide evidence that the online adjusted BCI-based robotic protocol helps subjects produce stronger and more sustained motor imagery throughout the motor task, compared to other BCI-based protocols. We also show that BCI-assisted robotic therapy can enable a level of motor cortical activity that is similar to a scenario in which the subjects could actually execute the motion. These results suggest that BCI-assisted rehabilitation methods that provide online modification of the task speed based on continuously inferred motor imagery have potential in increasing the level of involvement of patients during exercises and may lead to more effective recovery.