Tensor Based Temporal and Multilayer Community Detection for Studying Brain Dynamics During Resting State fMRI

Area of Science:

• Neuroscience
• Network Science
• Data Science

Background:

• Resting-state functional magnetic resonance imaging (fMRI) is crucial for understanding brain organization in health and disease.
• Emerging evidence highlights the dynamic nature of functional connectivity during resting states.
• Analyzing temporal dynamics offers superior insights compared to static network analysis.

Purpose of the Study:

• To introduce a novel tensor-based algorithm for temporal and multi-layer community detection in brain networks.
• To identify and track the evolving community structure of brain networks across time and subjects.
• To investigate the temporal dynamics of functional connectivity using fMRI data.

Main Methods:

• Developed a tensor-based algorithm incorporating multi-layer community detection.
• Constructed fMRI connectivity networks across various regions of interest.
• Utilized Tucker decomposition to determine the subspace representing community structure over time.

Main Results:

• Brain dynamics were characterized by recurring functional connectivity states across time and subjects.
• Evaluated dynamic brain behavior by assessing subnetwork consistency during resting state.
• Identified specific networks (default mode, cognitive control, limbic) exhibiting low temporal consistency, indicating dynamic behavior.

Conclusions:

• Functional brain connectivity is inherently dynamic.
• The developed community detection method reveals smooth temporal evolution of brain network structures.
• This approach enhances understanding of subnetwork behavior during resting-state fMRI.