Network analysis of whole-brain fMRI dynamics: A new framework based on dynamic communicability

Area of Science:

• Neuroimaging
• Network Neuroscience
• Computational Psychiatry

Background:

• Structural connectivity (SC) and functional connectivity (FC) map brain organization using network theory.
• Current network analyses of FC are often static, failing to capture the dynamic nature of fMRI time series.
• Understanding brain communication dynamics is crucial for interpreting brain function.

Purpose of the Study:

• To introduce a novel network-oriented analysis for dynamic whole-brain effective connectivity (EC).
• To develop a model-based approach that accounts for the temporal dynamics of fMRI signals.
• To provide time-dependent metrics for characterizing brain network activity and communication.

Main Methods:

• Tuning a multivariate Ornstein-Uhlenbeck (MOU) process to resting-state fMRI data for MOU-EC and input property estimation.
• Utilizing Green function analysis for time-dependent network impulse response.
• Calculating a time-dependent graph descriptor, 'communicability', to assess node and connection roles in activity propagation.

Main Results:

• The developed framework provides dynamic, time-dependent graph-like descriptors of brain activity.
• Communicability metrics reveal the roles of nodes and connections in propagating activity over time.
• Analysis showed a temporal shift from segregated to globally integrated network activity, with merging functional communities.

Conclusions:

• The proposed model-based approach effectively captures the temporal dynamics of fMRI signals for connectivity analysis.
• This dynamic framework offers a more accurate representation of brain communication compared to static methods.
• The findings provide a robust foundation for analyzing and interpreting fMRI data, including task-evoked activity.