Multimodal Motor Imagery BCI Based on EEG and NIRS

Brain-computer interface comprises technologies for brain activity identification used in many application fields such as motor imagery, disease or mental state detection. Multimodal approach that utilizes hybrid data can be improve motor imagery classification. The paper explores utilization of several classification techniques for multimodal electroencephalography (EEG) and near-infrared spectroscopy (NIRS) data classification in motor imagery BCI. Five classifiers used in the evaluation are Logistic Regression, K-Nearest Neighbours, Support Vector Machines, Linear Regression, SVC Radial Basis Regression and their performance is compared on EEG and EEG+NIRS datasets for motor imagery tasks classification.