High Classification Accuracy of a Motor Imagery Based Brain-Computer Interface for Stroke Rehabilitation Training

Area of Science:

• Neuroscience
• Rehabilitation Engineering
• Biomedical Engineering

Background:

• Motor imagery (MI) based brain-computer interfaces (BCI) offer real-time control for devices like functional electrical stimulation (FES).
• Stroke-induced motor area damage can impede MI-BCI control for rehabilitation.
• Evaluating MI-BCI control accuracy in stroke patients is crucial for therapeutic applications.

Purpose of the Study:

• To comparatively evaluate the MI-BCI control accuracy between stroke patients and healthy subjects.
• To assess the feasibility and effectiveness of MI-BCI for stroke rehabilitation.

Main Methods:

• Five stroke patients trained with the recoveriX system over 10-24 sessions.
• Electroencephalography (EEG) data classified during imagined left/right hand movements.
• Real-time feedback and functional electrical stimulation (FES) activated upon correct detection.

Main Results:

• Grand average mean accuracy was 87.4% across all patients and sessions.
• All patients achieved at least one session with maximum accuracy above 96%.
• Stroke patients exhibited high MI-BCI control accuracy, exceeding results from healthy controls in prior studies.

Conclusions:

• Stroke patients can effectively control MI-BCI systems with high accuracy, comparable or superior to healthy individuals.
• High motivation and the combined sensory feedback (visual, motor, tactile, proprioceptive) likely contribute to success.
• MI-BCI shows promise for stroke rehabilitation, with potential for motor function improvement even independent of classification accuracy gains.