Non-stationary Group-Level Connectivity Analysis for Enhanced Interpretability of Oddball Tasks.

Area of Science:

• Neuroscience
• Signal Processing
• Biomarkers

Background:

• Neural responses in oddball tasks serve as biomarkers for information processing.
• Non-stationarity in neural activity complicates accurate estimation of brain network interactions.
• Existing methods struggle with the dynamic, time-varying nature of brain networks.

Purpose of the Study:

• To develop a robust method for analyzing non-stationary functional connectivity in electroencephalogram (EEG) data.
• To improve the assessment of brain potential and information processing during oddball tasks.
• To accurately evaluate the contribution of neural network links in discriminating between oddball responses.

Main Methods:

• Developed a piecewise multi-subject analysis applied over time intervals assuming local stationarity.
• Experimented with fixed and variable time-window segmentation approaches for EEG recordings.
• Utilized the weighted Phase Lock Index (wPLI) as a functional connectivity metric.
• Validated the approach on real-world EEG data using a supervised thresholding method.

Main Results:

• Demonstrated the effectiveness of variable time segmentation for extracting functional connectivity.
• Showcased the ability of the piecewise analysis to handle non-stationary neural activity.
• Confirmed improved discrimination of oddball responses through piecewise group-level analysis.

Conclusions:

• Piecewise analysis, particularly with variable time segmentation, enhances the study of non-stationary functional connectivity in EEG.
• This method provides a more accurate physiological biomarker for evaluating brain information processing.
• The approach offers a refined way to analyze neural dynamics in response to deviant stimuli.