Phase synchronization analysis of EEG signals: an evaluation based on surrogate tests

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Phase synchronization (PS) analysis is crucial for inferring functional connectivity from neural signals like electroencephalography (EEG).
• Methodological challenges persist in quantifying functional connectivity using PS, specifically regarding optimal EEG segment duration and significance testing methods.

Purpose of the Study:

• To address critical methodological issues in phase synchronization (PS) analysis for electroencephalography (EEG).
• To determine optimal EEG signal duration for PS index estimation.
• To identify suitable methods for inferring the significance level of estimated PS indices.

Main Methods:

• Conducted intensive computational studies using surrogate tests on artificial EEG data (rank order, phase spectra, instantaneous frequency shuffling).
• Employed intersubject EEG pairs, assuming independence between subjects, for validation.
• Evaluated the efficacy of phase-shuffled surrogate data and intersubject surrogate tests for significance testing.

Main Results:

• The phase-shuffled surrogate method proved effective for significance testing of PS indices, yielding results comparable to intersubject EEG surrogate tests.
• An EEG signal duration encompassing approximately 3 to 16 cycles is generally suitable for accurate PS analysis.
• Mean phase coherence-based PS index is recommended for analyzing EEG data sampled at lower rates.

Conclusions:

• Phase-shuffled surrogate testing is a viable and reliable method for assessing the statistical significance of PS indices in EEG.
• The optimal duration for EEG signal segments in PS analysis is relatively short, around 3-16 cycles.
• Mean phase coherence offers a robust approach for PS analysis, especially when dealing with lower sampling rate EEG data, enhancing the reliability of functional connectivity estimations.