EEG Channel Selection Methods for Motor Imagery in Brain Computer Interface

EEG is widely applied in motor imagery because of its non-invasive and easy access. However, multi-channel EEG signals cause noise interference and redundant information. Channel selection can reduce the noisy signal to obtain more real signals. Therefore, many studies focus on the influence of channel selection algorithms on motion imagery decoding. This paper summarizes the characteristics of channel selection methods for decoding motion imagination tasks and compares their methods and results. One way to select the channel is based on neural network and takes the channel as the input parameter of the neural network; Another is an EEG classification method based on statistical information. Statistical information such as CSP rank, cross-correlation, and Pearson correlation coefficient is used to sort channels. In addition, we also discuss the relationship between the channel selection algorithm and high-density electrode design.