Improved filter bank common spatial pattern algorithm based on the sparrow search algorithm

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Motor imagery (MI) electroencephalography (EEG) decoding is crucial for human-computer interaction and rehabilitation.
• Traditional EEG decoding methods struggle with individual brain rhythm variability due to fixed frequency-band segmentation.
• Advances in brain-computer interface (BCI) technology highlight the need for personalized decoding strategies.

Purpose of the Study:

• To develop an adaptive method for optimizing frequency-band segmentation in motor imagery EEG decoding.
• To enhance the performance of BCIs by accounting for individual differences in brain activity.
• To integrate the Sparrow Search Algorithm (SSA) with Filter Bank Common Spatial Pattern (FBCSP) for adaptive sub-band selection.

Main Methods:

• An adaptive approach was developed, integrating the Sparrow Search Algorithm (SSA) with Filter Bank Common Spatial Pattern (FBCSP).
• SSA was employed to adaptively search for optimal sub-band boundaries, enabling individualized frequency-band selection for MI EEG decoding.
• The proposed SSA-FBCSP method was evaluated using the BCI Competition IV 2a dataset and combined with various classifiers (SVM, LDA, KNN).

Main Results:

• The SSA-FBCSP method demonstrated improved frequency-band adaptability in cross-session evaluations.
• The SSA-FBCSP-LDA combination achieved the highest performance, with an average accuracy of 89.92%, surpassing the conventional approach by 21.76%.
• Adaptively selected sub-bands correlated with Event-Related Desynchronization/Synchronization (ERD/ERS) patterns, validating the optimization effectiveness.

Conclusions:

• The proposed adaptive SSA-FBCSP method significantly enhances motor imagery EEG decoding accuracy and personalization.
• The approach offers a favorable balance of accuracy, interpretability, and computational efficiency compared to deep learning models.
• This technique presents a promising direction for developing personalized brain-computer interface systems.