Single Versus Multiple Events Error Potential Detection in a BCI-Controlled Car Game With Continuous and Discrete...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Continuous brain-computer interface (BCI) applications require accurate error detection for optimal user experience.
• Traditional single-trial error detection methods in BCI can be limited in accuracy.
• Motor imagery (MI) is a common BCI control paradigm.

Purpose of the Study:

• To develop and evaluate a novel method for detecting errors in continuous BCI applications.
• To enhance error detection accuracy by analyzing multiple events (MEs) instead of single trials.
• To assess the efficacy of the new method in a BCI-driven game.

Main Methods:

• A new method termed the ME method was developed, combining and averaging classification results of single events (SEs).
• The ME method determined the correctness of MI trials by analyzing event sequences.
• Offline simulations and an online experiment using a BCI-driven car game were conducted to evaluate the method.

Main Results:

• The ME method achieved feasible accuracy in distinguishing erroneous from correct MI trials, even with low single-event error potential (ErrP) detection rates.
• In the online experiment, subjects using the error detection method achieved higher scores compared to those without.
• The improved performance came at the cost of slightly increased game completion times.

Conclusions:

• Combining multiple events (MEs) analysis offers a reliable approach for monitoring continuous states in BCI applications.
• The developed ME method demonstrates a novel and promising technique for online error detection in continuous BCI.
• Error potential (ErrP) detection, when utilizing ME analysis, can be a valuable tool for improving BCI performance.