Handling EEG artifacts and searching individually optimal experimental parameter in real time: a system development...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Neuroadaptive paradigms enhance event-related potential (ERP) generalizability and biomarker discovery.
• Real-time electroencephalography (EEG) artifact removal is crucial for robust ERPs in neuroadaptive research.
• Efficient artifact handling is needed to manage computational costs and prevent ERP distortion.

Purpose of the Study:

• To develop and validate a cost-efficient system for online artifact handling to support ERP-based neuroadaptive research.
• To improve the accuracy and reliability of ERPs obtained in real-time.
• To accelerate biomarker discovery by optimizing experimental conditions.

Main Methods:

• Developed a single trial PCA-based artifact removal (SPA) method using variance distribution dichotomies.
• Applied SPA in an ERP-based neuroadaptive paradigm utilizing Bayesian optimization.
• Optimized the inter-stimulus-interval (ISI) to maximize ERP signal-to-noise ratio.

Main Results:

• SPA demonstrated strong performance in computational efficiency and ERP pattern preservation compared to other algorithms.
• The Bayesian optimization procedure, powered by SPA, rapidly identified individually optimal ISIs.
• The developed system effectively extracts ERPs and preserves key neural effects.

Conclusions:

• SPA is a simple, cost-efficient method for real-time artifact removal in EEG.
• The validated system enhances ERP extraction and effect preservation.
• This approach significantly supports and advances ERP-based neuroadaptive paradigms.