Towards Enhancing Motor Imagery Based Brain-Computer Interface Performance by Integrating Speed of Imagined Movement

Left and right motor imagery tasks have commonly been utilized to construct a two-class Brain-computer Interface system, whilst the speed property of imagined movement has received less attention. In this study, we are trying to integrate the types and speed property of both imagined movement and real movement to further improve the performance of the two-class BCI system. Thus, real movement session and imagined movement session were carried out on the separated days. In real movement session, it has shown that 8 healthy volunteers have achieved an average accuracy of 67.62% with the same actual left and right hand clenching speed, and 78.62% with diverse speeds, which was a significant improvement (p=0.0176). Besides, only three subjects could pass the 70% accuracy threshold with same actual clenching speed, while six of them achieved to pass it with diverse speeds. In imagined movement session, all the subjects with diverse imagined clenching speed achieved a better control compared with same imagined speed. The proposed idea of integration of speed information has shown a promising benefit in two-class BCI construction in this preliminary study.