A Synchronous and Closed-Loop Architecture of BCI-Based Rehabilitation System for Stroke with Robot and Virtual Reality

摘要

In this paper, we proposed a novel architecture of a scalp electroencephalography (EEG) based rehabilitatoin system via brain computer interface with robotic and virtual reality devices. More and more attentions had been taken to the brain computer interface, because it is the only way for some special patients to communicate with the outside world.EEG has show significant potential as a non-invasive way for BCI. But it is very hard to decode useful information from EEG signals in a reliable and efficient way, and also very difficult to get desiable rehabilitation result by current rehabilitation system. To this end, we propose an integrated EEG-driven rehabilitation system with assistive robotic and virtual reality feedback for some stroke patients. The framework of our method contains follow steps: signals accquire, signals preprocess; feature extract; classificatoin and feedback to human. Many advance algorithm has been surveryed and combined to accquire a better result. We use common spatial pattern as the spatial feature of signals, and convert signals to spectral via fast fourier transform. With the spatial and spectral features, we use support vector machine to classify these features and reduce final classification result by a fusion algorithm. At last, the classification result has been used to control robot and virtual reality devices. This synchronous and closed-loop rehabilitation system can achieve benefits for some peoples who suffer from stroke.