An automatic subject specific channel selection method for enhancing motor imagery classification in EEG-BCI using correlation

A motor imagery (MI) based brain–computer interface (BCI) decodes the intention from electroencephalogram (EEG) of a subject and translates this into control signal. These intentions are hence classified as different cognitive tasks, e.g. left right hand movements. challenge in developing BCI is handling high dimensionality data recorded multichannel EEG signals which highly subject-specific. Designing portable whilst minimizing channel number challenge. To end, paper presents method to reduce count with goal reducing computational complexity maintaining sufficient level accuracy, by utilising an automatic subject-specific selection created using Pearson correlation coefficient. This computes between helps select correlated channels for particular without compromising classification accuracy (CA). Common spatial patterns (CSP) used analyse imagined movements evaluated on both Competition III Dataset IIIa foot tasks IVa. For datasets, minimum identified average reduction 65.45% demonstrating increase >5% CA Cz reference.