Discrimination of Four Classes in Brain-Computer Interface Based on Motor Imagery

Abstract

Brain–computer interface provides a voluntarily, non-manual control for artificial limb or device by translating brain activity patterns into control commands. The research investigated the classification of multiclass motor imagery for electroencephalogram (EEG)-based Brain-Computer Interface (BCI) using independent component analysis (ICA), principle component analysis (PCA) and support vector machine (SVM) techniques. The proposed techniques were evaluated by Cohen's kappa coefficient and gave average accuracy around (97+2%) in session one and (31+4%) in session two in classifying four different motor imageries (MI) from EEG measurements for nine subjects under investigating.