Does Distance Between Electrodes Affect the Accuracy of Decoding the Motor Imagery Using EEG?

The selection of channels in electroencephalography (EEG) setups for motor imagery tasks has long been a subject of research interest. However, the role of the distance between electrodes in this process remains relatively unexplored. In this letter, we address this gap by integrating a distance measure into the channel selection process and investigating its impact on decoding motor imagery tasks. Our methodology involves incorporating a distance measure in conjunction with normalized mutual information for subject-independent optimal EEG electrode selection. To validate our approach, we utilized the brain-computer interface (BCI) Competition III IVa dataset, a publicly available benchmark for EEG-based brain–computer interface research. The findings of our study reveal the significance of considering electrode distance in channel selection for motor imagery tasks. Our results indicate that the inclusion of the distance measure leads to improvements in decoding accuracy. Empirical analysis demonstrates that the distance measure-based method exhibits 9% enhancement in decoding accuracy compared to the 3Cs (C3, C4, and Cz), even when employing as few as 15 channels selected out of the original 118. This highlights the effectiveness and efficiency of utilizing electrode distance in EEG channel selection for motor imagery tasks.