Improving brain-computer interface performance with optimized frequency interaction and enhancement techniques:...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Motor imagery-based brain-computer interfaces (MI-BCI) are crucial for assistive technologies.
• Improving MI-BCI classification accuracy and robustness is an ongoing challenge.
• Spontaneous electroencephalography (EEG) signals offer a rich source for feature extraction.

Purpose of the Study:

• To enhance MI-BCI classification accuracy using Cross-Frequency Coupling (CFC) features.
• To leverage spontaneous EEG signals for increased system robustness.
• To develop an integrated pipeline for efficient MI-BCI classification.

Main Methods:

• EEG data from 25 participants performing motor imagery tasks were analyzed.
• Phase-Amplitude Coupling (PAC) was used to extract CFC features.
• Particle Swarm Optimization (PSO) selected optimal EEG channels.
• XGBoost classifier with 10-fold cross-validation was employed, integrated as the CFC-PSO-XGBoost (CPX) pipeline.

Main Results:

• The CPX pipeline achieved an average classification accuracy of 76.7% using only eight EEG channels, outperforming existing methods.
• CPX demonstrated robustness and scalability on the BCI Competition IV-2a dataset, achieving 78.3% multi-class accuracy.
• The results highlight the effectiveness of CFC features and PSO-based channel selection for MI-BCI.

Conclusions:

• The CPX method significantly improves MI-BCI classification accuracy by utilizing spontaneous EEG and CFC features.
• This approach offers a robust and practical solution for BCI applications.
• CPX enables effective brain-to-device communication with reduced channel requirements and high performance.