CluSem: Accurate clustering-based ensemble method to predict motor imagery tasks from multi-channel EEG data.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Motor imagery electroencephalogram (MI-EEG) classification is crucial for brain-computer interface (BCI) applications.
• Real-time MI-EEG signal classification presents significant challenges due to high dimensionality and dynamic data behaviors.
• Existing classification methods show limited prediction performance in real-time BCI.

Purpose of the Study:

• To introduce a novel clustering-based ensemble technique, CluSem, to improve real-time MI-EEG classification.
• To develop a brain game, CluGame, for evaluating the CluSem method in a practical BCI application.
• To demonstrate the effectiveness of CluSem in enhancing classification accuracy for motor imagery movements.

Main Methods:

• Developed CluSem, a clustering-based ensemble technique for MI-EEG signal classification.
• Created CluGame, a brain game utilizing CluSem for real-time classification and prediction.
• Implemented real-time EEG signal classification and prediction tabulation through animated balls controlled via threads within CluGame.

Main Results:

• CluSem demonstrated an improvement in classification accuracy ranging from 5% to 15% compared to existing methods.
• Performance enhancements were validated on both newly collected and publicly available EEG datasets.
• The source codes for CluSem and CluGame are publicly accessible for research reproducibility.

Conclusions:

• CluSem effectively enhances the classification performance of real-time MI-EEG signals in BCI applications.
• The developed CluGame provides a practical platform for evaluating and utilizing advanced BCI classification techniques.
• The proposed method offers a promising solution for overcoming the limitations of current real-time BCI systems.