Applying SSVEP BCI on Dynamic Background.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Science

Background:

• Steady-state visual evoked potential (SSVEP) based brain-computer interfaces (BCIs) are highly efficient and accurate.
• Existing SSVEP paradigms face challenges in dynamic background environments.
• Advanced decoding methods are crucial for improving BCI performance.

Purpose of the Study:

• To propose novel modulation methods for the SSVEP paradigm to enhance performance in dynamic backgrounds.
• To introduce an end-to-end neural network, MTSGNN, for decoding SSVEP signals.
• To evaluate the effectiveness of the proposed modulation and decoding techniques.

Main Methods:

• Developed two SSVEP modulation methods: color inversion and brightness compression.
• Proposed the Multi-scale Temporal-Spatial Global average pooling Neural Network (MTSGNN) for SSVEP signal decoding.
• Conducted experiments to compare modulation techniques and evaluate MTSGNN against competitive methods.

Main Results:

• Brightness compression modulation demonstrated superior performance compared to color inversion and no modulation.
• MTSGNN achieved significant performance improvements over the best competitive decoding method (11.98% under color inversion, 3.9% under brightness compression, and 5.15% with no modulation).
• The proposed methods showed effectiveness and robustness, particularly in dynamic background conditions.

Conclusions:

• Brightness compression modulation and MTSGNN effectively enhance SSVEP BCI performance.
• The developed methods improve SSVEP signal decoding accuracy, especially in challenging dynamic environments.
• This research expands the practical applications of SSVEPs in BCIs.