Online motor-imagery based BCI

This contribution presents a study on feedback training using a motor-imagery based brain-computer interface (BCI). In contrast to majority of BCI papers focusing on development of complex signal processing methods, we used the simplest possible available methods to focus on influence of the feedback itself and development of methodology how to conduct the experiments to support user training. We have shown that simple signal processing methods are not only sufficient to achieve high-speed control but their usage is desired in online applications as the simple methods can be more easily adapted. We have introduced a novel adaptive threshold classification approach. Classification scores up to 100 % (75 % in average over all subjects and sessions) were achieved with classification of left and right arm motor imagery and visual feedback in experiments performed in synchronous manner and asynchronous self-paced operation was reached with few proficient subjects.