Algorithmic Analysis on Automated Channel Selection Framework for Motor Imagery BCI

Recently, motor imagery (MI) classification in electroencephalography (EEG) based brain-computer interface (BCI) systems is a significant field of brain science study. Since EEG includes multi-channel recordings with massive data, it causes a great challenge for extracting more representative data for proficient classification of MI tasks. Therefore, it is essential to reduce the redundant data from the original EEG signal by electing suitable channels and also to lessen the computation cost. In this work, it is planned to select the most suitable channel using a hybrid model called Geometric Mean based Moth flame hybridized with Firefly Update (GM-MFU) model. Along with this, the suitable samples related to the channels are also chosen via the GM-MFU model. Further, the selected channels and features extracted from the original signal with Common Spatial Pattern (CSP) are classified using Bayesian Linear Discriminate Analysis (BLDA). Eventually, algorithmic analysis is carried out on the proposed GM-MFU algorithm by varying its parameters.