Single-Trial EEG Classification Using Spatio-Temporal Weighting and Correlation Analysis for RSVP-Based Collaborative...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Single brain-computer interfaces (BCI) face performance limitations.
• Collaborative BCI (cBCI) systems integrating multi-user electroencephalogram (EEG) data offer improved performance.
• Challenges in cBCI include feature extraction, data fusion, and reducing calibration time.

Purpose of the Study:

• To propose and evaluate a novel spatio-temporal weighting and correlation analysis (STC) algorithm for cBCI systems.
• To enhance discriminant feature extraction for multi-user EEG data.
• To improve classification performance and reduce calibration time in cBCI.

Main Methods:

• Developed a spatio-temporal weighting and correlation analysis (STC) algorithm for event-related potential (ERP) feature extraction and classification.
• Employed source extraction and interval modeling to address inter-trial variability.
• Utilized spatio-temporal weighting, temporal projection, and correlation analysis for feature extraction, fusion, and classification.

Main Results:

• Evaluated the STC algorithm using collaborative cross-session EEG datasets from 14 subjects (RSVP paradigm).
• STC demonstrated significantly higher performance in single-user and collaborative EEG classification compared to state-of-the-art algorithms.
• Achieved superior performance for both within-session and cross-session datasets.

Conclusions:

• The STC algorithm effectively enhances classification performance in multi-user collaboration and cross-session transfer for RSVP-based BCI systems.
• The proposed method contributes to reducing system calibration time.
• STC shows promise for advancing cBCI technology by improving accuracy and efficiency.