Adaptive Detection of Idle State in Motor Imagery based Brain Computer Interface

摘要

Asynchronous control is an important issue for brain-computer interfaces (BCIs) working in real life. To date, most of asynchronous BCI systems need predefined decision thresholds to tell when the user is idle. In this paper, we proposed an adaptive method for off-line idle state detection in motor imagery (MI) based BCIs. This method can automatically adjust the decision thresholds according to the separation and compactness ratio of event-related desynchro nization (ERD) features in 2-class MI tasks. And it treats the prediction labels by a fuzzy way. Experimental results of BCI competition Ⅲ dataset IVc show that the proposed method can decrease the prediction mean square error (MSE) to a level below 0.3 and improve the detection rate of idle state to 50％.