Improving Individual-Specific SSVEP-BCI with Adaptive Channel and Subspace Selection in TRCA

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Individual-specific steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) show promise but are limited by generalized channel and subspace selection.
• Existing methods often lead to suboptimal performance by not fully exploiting individual-specific calibration data.

Purpose of the Study:

• To propose AS-TRCA, a novel method for developing purely individual-specific SSVEP-BCIs by fully leveraging subject-specific information.
• To enhance BCI performance through optimal channel learning and selection (OCLS) and optimal subspace selection (OSS).

Main Methods:

• AS-TRCA employs OCLS using sparse learning with spatial distance constraints to select optimal subject-specific channels.
• AS-TRCA utilizes OSS to adaptively determine the number of optimal subject-specific task-related subspaces by maximizing profile likelihood.

Main Results:

• AS-TRCA successfully identifies meaningful channels and the appropriate number of task-related subspaces for individual subjects.
• AS-TRCA significantly improved decoding accuracy across various SSVEP-BCI decoding methods, including deep learning approaches.

Conclusions:

• AS-TRCA offers a promising approach for advancing SSVEP-BCI performance by enabling purely individual-specific BCI development.
• The method enhances existing SSVEP-BCI decoding techniques, paving the way for broader practical applications.