Analysis and Classification for EEG Patterns of Force Motor Imagery Using Movement Related Cortical Potentials

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Brain-computer interfaces (BCIs) offer natural human-computer interaction.
• Decoding kinetic information from motor imagery is a key research area.
• Motor-related cortical potentials (MRCPs) reflect motor planning processes.

Purpose of the Study:

• To analyze and discriminate electroencephalography (EEG) patterns of different hand force motor imagery levels.
• To evaluate the effectiveness of MRCPs in classifying motor imagery tasks.
• To compare classification accuracy using MRCPs, common spatial patterns (CSP), and their combination.

Main Methods:

• Nine healthy subjects performed hand force motor imagery tasks at 30% and 10% maximal voluntary contraction (MVC).
• EEG data were analyzed to extract MRCPs and CSP features.
• Machine learning algorithms were used to classify different force levels of motor imagery.

Main Results:

• MRCPs revealed significant discrimination between different force levels, primarily through motor planning manifestations.
• The combined use of MRCP and CSP features achieved an average classification accuracy of 78.3%.
• This combined approach showed a significant improvement over CSP-only (8.5% higher) and MRCP-only (2% higher) features.

Conclusions:

• MRCPs are a feasible and effective feature for classifying hand force motor imagery.
• Combining MRCPs with CSP enhances classification performance in motor imagery-based BCIs.
• This study demonstrates the potential of MRCPs for developing more sophisticated brain-computer interfaces.