Classification of EEG for Upper Limb Motor Imagery: An Approach for Rehabilitation

Abstract

Patients suffering from severe motor neuron diseases (MND) experience motor disability and their rehabilitation has always remained a challenge. Electroencephalogram (EEG) based brain computer interface (BCI) is a system that can be used for the rehabilitation of patients suffering from amputation or from severe disease like MND, stroke, locked in syndrome (LIS); where EEG signal is acquired from brain scalp while performing certain mental task such as motor imagery, cognitive imagery etc. In the present paper brain signals i.e. EEG for 10 motor imagery movements of upper limb acquired from 4 subjects were classified. Filter Bank Common Spatial Pattern (FBCSP) algorithm was used for extracting features of EEG signal captured from 5 electrodes placed over motor cortex and mutual information is used for feature selection. Classification algorithm followed was linear Support Vector Machine (SVM) in MATLAB 2015a.