Classification of Electroencephalography Bio Signals Based on Motor Imagery

Stroke is known to be a cause of multitude impairments, including abnormalities in physical, motor, and emotional capabilities in stroke patients. Studies have demonstrated that employing motor imagery tasks through Electroencephalogram (EEG) bio signals yields 10% higher accuracy compared to conventional Electroencephalogram (EEG) methods. Thus, the primary goal of this study is to employ machine learning algorithms to classify Electroencephalogram (EEG) bio signals based on motor imagery from healthy people and stroke patient’s open-source datasets. Methods such as Custom Spatial Filter (CSP) is used to extract pertinent features from the acquired data after thorough processing and filtering. It was found that SVM has yielded up to 98.86% of accuracy in classifying motor imagery tasks compared to Linear Discriminant Analysis (LDA), and Gaussian Naïve Bayes (GNB) algorithms.