Robust time-varying multivariate coherence estimation: application to electroencephalogram recordings during general...

Area of Science:

• Neuroscience
• Signal Processing
• Data Analysis

Background:

• Coherence analysis measures frequency-dependent covariance in multivariate oscillatory data.
• Global coherence summarizes this by analyzing the first mode of spectral matrix eigenvalue decomposition.
• Current methods are sensitive to noise, potentially confounding neural population activity.

Purpose of the Study:

• To enhance the global coherence procedure for increased robustness against noise.
• To enable characterization of how power within specific coherent modes changes over time.

Main Methods:

• Methodological enhancements to the global coherence procedure.
• Utilizing eigenvalue decomposition of the cross-spectral matrix.
• Time-series analysis of power within coherent modes.

Main Results:

• Improved robustness of global coherence analysis to noise.
• Ability to track temporal dynamics of power within distinct coherent modes.
• Reduced confounding of activity from disparate neural populations.

Conclusions:

• The enhanced global coherence method provides a more reliable tool for analyzing complex neural data.
• This advancement facilitates deeper understanding of dynamic functional connectivity in the brain.