An EEG Study on Hand Force Imagery for Brain-Computer Interfaces

Motor imagery based BCIs are one of the most important BCI paradigms. Although it has been studied for a long time, the EEG features for kinetic information of motor imagery are still less known. In this paper, we explored EEG patterns of hand force motor imagery. Six subjects participated in this study, who were required to imagine clenching their hands with two different levels of force during the experiment. Time-frequency analyses showed a significant decrease of EEG power at alpha and beta band when subjects performed the motor imagery task, compared to the rest state. Furthermore, the power decrease of the high force imagery was significantly larger than that of the low force imagery. Support vector machines were used to classify the three different EEG patterns ( rest vs. high force vs. low force) and achieved an accuracy of 71% on average. It suggests that the force level of motor imagery plays a critical role in the patterns of event-related desynchronization, and could be used to control BCIs.