Classification of EEG Signals for Hand Gripping Motor Imagery and Hardware Representation of Neural States Using Arduino-Based LED Sensors

Abstract

This study aims to classify the multi-frequency EEG signals associated with motor imagery while performing a carefully devised motor imagery task of gripping left/right hand for rehabilitation application. EEG-based brain–computer interface incorporates recording and classification of the transient EEG changes during different imagery tasks. For rehabilitation of gripping and release movement, ten healthy right-handed volunteers (seven males and three females with mean age: 21.3 years) participated in the EEG investigation using 32 Channel Brain product system. The volunteers performed the motor imagery cued tasks in random order. EEG data was processed to calculate relative alpha band power for each motor imagery trial block from channels C3 and C4 to be passed as feature vectors for the classification of the brain states. After a comparative analysis, SVM classification algorithm provided the highest accuracy of 75%, and the binary output was interfaced with Arduino Uno component to reciprocate left and right imaginary hand movement states using LED light bulbs.