Motor Imagery-Based BCI for Individual Finger Classification by EEG Current Source Estimation: A Feasibility Study

The brain-computer interface (BCI) using electroencephalography (EEG) can detect the activity of the sensorimotor cortex elicited by the motor imagery task of left and right movements with high accuracy. However, discrimination of more detailed body parts, such as individual fingers, remains challenging due to the low spatial resolution of EEG. In this study, as preliminary research for the development of BCI that can detect responses to motor imagery of each finer, estimated cortical currents were investigated. Cortical current is expected to be able to detect motor imagery task of detailed body parts. In this study, the spatial distribution and classification of brain responses corresponding to the index and little finger motor imagery task were analyzed using estimated cortical currents from EEG. The results showed that all five subjects showed differences in the spatial distribution of cortical current during imagery task of each finger. In addition, on four subjects, the classification accuracy was greater than 70%. These results indicate that classification accuracy can be improved in discriminating brain responses to motor imagery of fingers, which has been considered as a challenging task with the method using conventional EEG signals only.