Single-Option P300-BCI Performance Is Affected by Visual Stimulation Conditions.

Area of Science:

• Neuroscience
• Computer Science
• Biomedical Engineering

Background:

• The P300 paradigm is a key Brain-Computer Interface (BCI) technique, valued for its reliability.
• However, P300 BCI applications face challenges with single-trial classification accuracy and training efficiency.
• Optimizing stimulus presentation is crucial for improving BCI performance.

Purpose of the Study:

• To evaluate single-trial classification effectiveness in P300 BCIs under varying visual stimulation conditions.
• To investigate the impact of stimulus type (color highlighting vs. cartoon face) and number of options on target/non-target classification.
• To analyze how training and testing with different datasets affect classification performance.

Main Methods:

• A P300 experimental protocol was designed with 19 healthy subjects over 3 sessions.
• Two visual stimulation conditions were tested: color highlighting and superimposing a cartoon face.
• The number of stimulus options varied from four to nine symbols.

Main Results:

• Event-Related Potentials (ERP) and classification accuracy were significantly stronger with cartoon faces.
• Classification accuracy remained similar regardless of the number of stimulus options presented.
• Performance decreased when training and testing with different stimulus types but was consistent across different numbers of symbols.

Conclusions:

• Cartoon faces are a more effective stimulus for eliciting stronger P300 responses and improving BCI classification.
• The number of stimulus options has a minimal impact on classification accuracy, suggesting flexibility in interface design.
• These findings aid in designing BCI systems that enhance evoked potentials while optimizing training duration.