BCI Channel Reduction and Optimization for Stroke Rehabilitation

Brain computer interface (BCI)-based system has become an alternative treatment for stroke rehabilitation. Multichannel EEG signal extraction is commonly used in BCIs, whereby reducing channels number has significant role in reducing the model complexity and the computational time and hence reducing the patient rehabilitation session time. Thus, in this paper, a BCI model for post-stroke neurorehabilitation with reduced number of channels is presented. The presented BCI model balances between model complexity/rehabilitation time and accuracy by optimizing the channels number. The model was evaluated using five classification algorithms at different number of channels using a dataset of 50 poststroke patients. The results showed that using the presented model with the selected 8 channels reduced the computational time by 60% and achieved accuracy of 88%.