Enhancing Detection of SSVEP-based BCIs Using Adjacent Frequencies Fusion Method

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) facilitate direct communication between the brain and external devices.
• Steady-state visual evoked potentials (SSVEP) are a key modality for BCI applications.
• Existing SSVEP decoding algorithms often overlook correlations between adjacent frequency signals.

Purpose of the Study:

• To introduce and evaluate a novel decoding algorithm, adjacent frequencies fusion filter bank canonical correlation analysis (AFF-FBCCA), for SSVEP-based BCIs.
• To enhance the accuracy and robustness of SSVEP signal decoding by leveraging information from adjacent frequencies.
• To provide a training-free and user-friendly solution for SSVEP BCIs.

Main Methods:

• Development of the adjacent frequencies fusion filter bank canonical correlation analysis (AFF-FBCCA) algorithm.
• Incorporation of weighted fusion of adjacent frequency information to exploit signal similarities.
• Dynamic adjustment of weight coefficients based on the time window for adaptive performance.
• Validation using public benchmark datasets for SSVEP-based BCIs.

Main Results:

• AFF-FBCCA consistently outperformed standard filter bank canonical correlation analysis (FBCCA) across all tested time windows.
• Significant improvements in classification accuracy were observed with the proposed AFF-FBCCA method.
• Enhanced information transfer rates (ITR) were achieved, indicating more efficient BCI communication.
• The method demonstrated robustness and maintained its advantage without requiring prior training.

Conclusions:

• AFF-FBCCA offers a superior approach for decoding SSVEP signals in BCIs compared to traditional methods.
• The algorithm's ability to utilize adjacent frequency information enhances BCI performance and user experience.
• AFF-FBCCA presents a promising, accurate, and user-friendly solution for advancing SSVEP-based BCI technology.