EEG Signal Complexity Measurements to Enhance BCI-Based Stroke Patients' Rehabilitation

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Science

Background:

• Stroke is a leading cause of death and disability worldwide.
• Brain-computer interface (BCI) techniques show potential for improving stroke patient rehabilitation outcomes.
• Motor imagery (MI) based BCIs offer a non-invasive approach to assist in recovery.

Purpose of the Study:

• To enhance motor imagery (MI)-based brain-computer interface (BCI) systems for stroke patient rehabilitation.
• To analyze electroencephalogram (EEG) data using a novel framework incorporating advanced signal processing and feature extraction techniques.
• To evaluate the classification performance of the proposed framework for four distinct motor imagery tasks.

Main Methods:

• Preprocessing of EEG data included conventional filtering and Independent Component Analysis (ICA) for denoising.
• Feature extraction involved calculating complexity (Fractal Dimension, Hurst Exponent) and irregularity (Tsallis Entropy, Dispersion Entropy) parameters.
• Dimensionality reduction was performed using Laplacian Eigenmap (LE), followed by classification using k-nearest neighbors (KNN), Support Vector Machine (SVM), and Random Forest (RF).

Main Results:

• The proposed MI-based BCI framework demonstrated effective feature extraction and classification.
• Laplacian Eigenmap (LE) combined with Random Forest (RF) achieved 74.48% accuracy, and with KNN achieved 73.20% accuracy.
• The integrated approach, including ICA denoising and proposed features, accurately described the MI framework for classifying four motor imagery tasks.

Conclusions:

• The developed MI-based BCI framework, utilizing ICA denoising and advanced features, is effective for stroke rehabilitation.
• The framework's ability to accurately classify motor imagery tasks provides a valuable tool for clinicians and researchers.
• This study contributes to the development of improved rehabilitation programs for individuals affected by stroke.