A Potential Source of Bias in Group-Level EEG Microstate Analysis.

Area of Science:

• Neuroscience
• Cognitive Science
• Biomedical Engineering

Background:

• Microstate analysis is a valuable tool for high-density electroencephalographic (EEG) data.
• Methodological best practices for microstate analysis, particularly concerning group comparisons, remain unclear.
• Microstate characteristics (topography and temporal dynamics) can vary within and between individuals, posing analytical challenges.

Purpose of the Study:

• To compare different microstate analysis approaches for group differences using simulations.
• To investigate the impact of using separate versus uniform microstate maps on analytical outcomes.
• To propose a novel approach to mitigate biases in microstate group analysis.

Main Methods:

• Simulations based on real EEG data from healthy controls were used.
• Two primary approaches were compared: deriving separate microstate maps for subgroups versus a single set of maps for the entire dataset.
• A paired analysis of subgroup maps was proposed as a solution.

Main Results:

• Using separate subgroup maps led to inflated Type I error rates when no true group differences existed.
• Employing a single set of maps for differing groups confounded topographic effects with other metrics.
• Subtle topographic differences significantly impacted derived microstate metrics.

Conclusions:

• Current microstate analysis methods for group comparisons can introduce substantial bias.
• A uniform map approach can obscure true differences, while subgroup-specific maps can create false positives.
• A paired analysis of microstate maps is recommended to improve accuracy and mitigate bias in group studies.