Regularization SAME Method can Enhance the Performance of SSVEP-BCI with Very Weak Stimulation

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Steady-state visual evoked potential-based brain-computer interfaces (SSVEP-BCIs) offer high information transfer rates but can cause visual fatigue due to flickering stimuli.
• Reducing stimulus pixel density enhances SSVEP-BCI comfort but lowers the signal-to-noise ratio (SNR), challenging signal decoding.
• Effective decoding strategies are crucial for SSVEP-BCIs utilizing low-pixel density stimuli for improved user comfort.

Purpose of the Study:

• To develop and validate an optimized decoding strategy for SSVEP signals generated by low-pixel density stimuli.
• To enhance the signal-to-noise ratio (SNR) and classification accuracy of SSVEP-BCIs under reduced visual stimulation.
• To improve the overall performance and user comfort of SSVEP-BCI systems.

Main Methods:

• Employed the Source Aliasing Matrix Estimation (SAME) method to augment datasets and improve decoding accuracy for SSVEP signals.
• Optimized the SAME method with a regularization technique to further boost decoding performance.
• Conducted experiments using SSVEP stimuli with varying pixel densities (1% to 100%) and frequencies (7Hz to 39Hz).

Main Results:

• The SAME method significantly improved SSVEP classification accuracy compared to traditional methods, particularly for stimuli with pixel densities ≤ 50%.
• The maximum accuracy increase achieved with SAME reached 8.6% under very weak stimulation conditions.
• Regularization further enhanced SAME, yielding maximum improvements of 4.29% over the standard SAME method, demonstrating superior decoding performance.

Conclusions:

• The proposed regularization SAME method significantly enhances SSVEP decoding performance with low-pixel density stimuli.
• This advancement contributes to the development of more comfortable and effective SSVEP-BCI systems.
• The optimized SAME method addresses the challenge of decoding weak SSVEP signals, paving the way for practical BCI applications.