Almost free of calibration for SSVEP-based brain-computer interfaces

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) offer high information transfer rates (ITR) using algorithms like eTRCA and TDCA.
• Current SSVEP-BCI calibration is time-consuming, limiting practical applications.
• Existing data augmentation methods like SAME have limitations in exploiting cross-stimulus information.

Purpose of the Study:

• To develop an improved data augmentation technique for SSVEP-BCIs to reduce calibration time.
• To enhance the performance of SSVEP-BCIs with limited calibration data.
• To enable practical, plug-and-play SSVEP-BCI systems.

Main Methods:

• Proposed multi-stimulus SAME (msSAME), an extension of SAME that leverages frequency similarities for better data augmentation.
• Developed a semi-supervised approach using msSAME to further decrease calibration requirements.
• Evaluated msSAME on public datasets (Benchmark, BETA) and an online experiment.

Main Results:

• msSAME outperformed the original SAME method for both eTRCA and TDCA on benchmark datasets.
• The semi-supervised msSAME approach achieved performance comparable to fully calibrated methods.
• Online experiment demonstrated rapid calibration (24s for 40 targets) with high ITR (average 213.8 bits/min).

Conclusions:

• msSAME significantly reduces calibration effort for individual SSVEP-BCIs.
• The proposed methods pave the way for more practical and user-friendly SSVEP-BCI applications.
• This advancement addresses a key barrier to the widespread adoption of SSVEP-BCI technology.