Determination of Dynamic Brain Connectivity via Spectral Analysis

Area of Science:

• Neuroscience
• Computational Neuroscience
• Physics

Background:

• Traditional functional connectivity (FC) analysis often suffers from temporal averaging, windowing artifacts, and noise.
• Existing methods struggle to capture the dynamic nature of brain connectivity at fine spatial scales.

Purpose of the Study:

• To introduce a novel spectral analysis method based on neural field theory and physical eigenmodes for analyzing dynamic functional connectivity (FC).
• To demonstrate how this approach overcomes limitations of conventional FC analysis, providing more accurate and interpretable insights into brain dynamics.

Main Methods:

• Utilizing spectral analysis grounded in neural field theory and physical eigenmodes of brain dynamics.
• Integrating analysis over space rather than averaging over time to mitigate artifacts.
• Applying the method to simple test cases to validate its efficacy.

Main Results:

• The spectral analysis effectively reduces or eliminates temporal averaging effects, windowing artifacts, and noise.
• Demonstrated that functional connectivity (FC) is inherently dynamic, even with fixed brain structure and effective connectivity.
• Identified that observed FC patterns are dominated by a limited number of dominant eigenmodes.
• Eigenmodes are shown to overlap across the entire brain, unlike artificially discretized networks, offering a more holistic representation.

Conclusions:

• Spectral analysis using eigenmodes provides a robust alternative to covariance-based FC, offering direct biophysical interpretability.
• This method avoids common artifacts introduced by statistical analyses lacking physical grounding.
• Tracking eigenmode coefficients over time is proposed as a superior representation for understanding brain dynamics.