Nonparametric Early Stopping Detection for c-VEP-based Brain-Computer Interfaces: A Pilot Study.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Code-modulated visual evoked potentials (c-VEP) offer high accuracy and short calibration times for brain-computer interfaces (BCI).
• Democratizing BCI use requires plug-and-play solutions, necessitating early stopping algorithms for real-time reliable selections.
• Existing early stopping techniques for c-VEP BCI under the circular shifting paradigm are limited.

Purpose of the Study:

• To propose a novel nonparametric early stopping method for c-VEP BCI systems.
• To enable real-time detection of the minimum code repetitions for reliable command selections.
• To reduce selection time in c-VEP BCI without compromising accuracy.

Main Methods:

• A novel nonparametric early stopping algorithm was developed.
• The algorithm approximates unattended command distributions to a normal distribution.
• Command selection is triggered when the command's correlation is identified as an outlier.

Main Results:

• Offline evaluation with 15 healthy users yielded an average accuracy of 97.08% and a speed of 1.37 s/command.
• Online evaluation with one user demonstrated technical feasibility, achieving 96.88% accuracy and 1.67 s/command.
• The proposed algorithm significantly reduces selection time while maintaining high accuracy.

Conclusions:

• The developed early stopping algorithm is feasible for real-time application in c-VEP BCI.
• This advancement can significantly decrease the time required for command selection.
• The method offers a practical step towards more accessible and user-friendly BCI systems.