A Complete Survey on Common Spatial Pattern Techniques in Motor Imagery BCI

Abstract

Background: Brain-computer interfaces that use motor imagery hold promise for direct communication and control through brain signals. Common Spatial Pattern (CSP) techniques have emerged as powerful tools for extracting discriminative features from electroencephalogram (EEG) signals in tasks requiring motor imagery. Objective: This survey paper aims to provide a comprehensive analysis of different CSP techniques employed in motor imagery BCIs, highlighting their strengths and limitations. Methods: We reviewed the literature and identified various CSP techniques, including Riemannian CSP, deep learning-based CSP, multiway CSP, and temporally weighted CSP etc. For each technique, we examined their underlying principles, algorithmic implementation, advantages, disadvantages, filtering technique used, classification accuracy, dataset used and relevant comments. Conclusion: Understanding and comparing different CSP techniques are crucial for enhancing the performance of motor imagery-based BCIs. Each technique has its own advantages and considerations, such as computational complexity and adaptability to different BCI scenarios. This survey serves as a valuable resource for researchers and practitioners in selecting appropriate CSP techniques to advance the area towards successful brain-controlled systems by enhancing the reliability and accuracy of motor imagery-based BCIs.