An iterative optimization technique for robust channel selection in motor imagery based Brain Computer Interface

Brain-Computer Interface (BCI) provides a direct communication pathway between brain and computer/machine bypassing the conventional pathway of nerves and muscles. Electroencephalography (EEG) is the most commonly used brain signal acquisition technique in BCI systems. The use of motor imagery (MI) patterns in EEG-based BCI has been proven as an effective method to translate the user's movement intention to commands for controlling external devices. To obtain high classification accuracy of MI, conventional EEG based BCI employ a large number of scalp electrodes. However, this is inconvenient in the clinical scenarios where preparation time is of paramount importance. This paper proposes a channel selection method which utilizes a priori information of the MI task and iteratively optimizes the number of relevant channels, thereby improving the classification accuracy. The proposed method is employed in BCI Competition III dataset IVa and BCI Competition IV, dataset 2a to classify hand and foot MI tasks. The proposed method results in better accuracy than state-of-the-art methods with a significant reduction in the number of channels.