Five-class motor imagery BCI classification and its application to brain-controlled wheelchairs

Area of Science:

• Neuroscience and Biomedical Engineering
• Brain-Computer Interfaces (BCIs)
• Rehabilitation Technology

Background:

• Brain-controlled wheelchair (BCW) technology offers mobility for individuals with physical impairments.
• Motor imagery-based BCIs (MI-BCIs) provide non-invasive neural control but face challenges in classification accuracy.
• Reliable multi-class classification is crucial for advanced BCW functionality.

Purpose of the Study:

• To enhance the performance of MI-BCI systems for BCW applications.
• To develop an advanced feature extraction algorithm for improved classification accuracy.
• To validate the proposed system in real-world BCW navigation tasks.

Main Methods:

• Proposed an EHT-CSP algorithm integrating Ensemble Empirical Mode Decomposition Hilbert-Huang Transform (EEMD-HHT) with Time-Frequency Common Spatial Pattern (TFCSP).
• Extracted marginal spectrum entropy and energy spectrum entropy using EEMD-HHT, combined with TFCSP features.
• Utilized Light Gradient Boosting Machine (LightGBM) for classification on a custom five-class MI-EEG dataset.

Main Results:

• Achieved an average classification accuracy of 78.45% for the proposed MI-BCI system.
• All participants successfully navigated BCW obstacle avoidance tasks.
• The EHT-CSP algorithm demonstrated superior performance compared to existing methods.

Conclusions:

• The proposed EHT-CSP algorithm significantly improves MI-BCI performance for BCW applications.
• The developed system enables reliable and intuitive control for wheelchair navigation.
• This advancement holds promise for enhancing assistive technologies for individuals with motor disabilities.