Channel and Feature Selection for a Motor Imagery-Based BCI System Using Multilevel Particle Swarm Optimization

Area of Science:

• Neuroscience
• Computer Science
• Signal Processing

Background:

• Brain-computer interfaces (BCI) enable communication and control by linking the brain to external devices.
• Classification performance in BCIs is often limited by irrelevant channels and misleading features.
• Efficient signal processing is crucial for real-time BCI applications.

Purpose of the Study:

• To propose an efficient signal processing framework for channel and feature selection in BCIs.
• To enhance classification performance and reduce computational load.
• To validate the proposed framework using the BCI Competition III dataset I.

Main Methods:

• Utilized particle swarm optimization (PSO) for channel and feature selection.
• Employed modified Stockwell transforms for feature extraction.
• Applied multilevel hybrid PSO-Bayesian linear discriminant analysis for optimization and classification.

Main Results:

• Achieved a classification accuracy of 99%, a significant improvement over the non-optimized method (89%).
• Reduced the number of features used to less than 10.5% of the original.
• Decreased test time by over 90%, with Kappa values of 0.98 and F-score of 98.99%.

Conclusions:

• The proposed PSO-based channel and feature selection framework effectively improves BCI classification performance.
• The framework accelerates convergence and reduces training time, making it suitable for real-time applications.
• This approach offers a valuable reference for the development of real-time BCI systems.