A Comprehensive Review of AI-Based Brain-Computer Interface with Prefrontal Cortex and Sensory-Motor Rhythms Systemization for Rehabilitation

Abstract

The paper examines brain-computer interfaces leveraging AI by analyzing four distinct states of the brain from EEG signals, including motor imagery of right-hand movements, motor imagery of left-hand movements, eyes open, and eyes closed states. The review categorizes signal processing techniques including preprocessing, feature extraction, and classification with tabulated results across datasets. The work addresses generalizability gaps in motor imagery state classification to advance rehabilitation applications.

Key Contributions

• Systematic categorization of existing EEG-based BCI techniques for motor imagery and eye state detection
• Comprehensive analysis of signal processing methodologies across preprocessing, feature extraction, and classification stages
• Comparative performance metrics across different AI classification approaches and public datasets
• Focus on addressing generalizability gaps in motor imagery state classification
• Actionable systemization of prefrontal cortex and sensory-motor rhythm signals for rehabilitation contexts

Methodology

Literature review analyzing published research on motor imagery and eye state categorization for EEG-based BCIs. The authors evaluate signal processing pipelines systematically and present statistical comparisons of classification results using various datasets. Four EEG brain states evaluated: right-hand MI, left-hand MI, eyes open, eyes closed.

Results

• Random forest classifier achieved accuracy up to 99.80% for eye state classification
• Support vector machine classification yielded 100% accuracy for motor imagery conditions under specific dataset/experimental conditions
• Multiple classification algorithms benchmarked across standardized datasets
• Prefrontal cortex + sensory-motor rhythm combination identified as effective signal source for rehabilitation BCIs

Limitations

Limited generalizability in the classification of motor imagery states or eye state conditions is identified as the key research gap. High accuracy figures (99-100%) reported under constrained experimental conditions may not transfer to real-world clinical settings with greater signal variability and user heterogeneity.

Source: A Comprehensive Review of AI-Based Brain-Computer Interface with Prefrontal Cortex and Sensory-Motor Rhythms Systemization for Rehabilitation by Anna Latha M, Ramesh R, Vellore Institute of Technology, Chennai