Dynamic mode decomposition of resting-state and task fMRI.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Data Analysis

Background:

• Functional connectivity (FC) analysis of fMRI data typically uses static component analysis methods.
• Static approaches overlook temporal dynamics in fMRI time series, limiting characterization of brain functional organization.
• Recent research indicates that FC dynamics encode crucial information about brain function.

Purpose of the Study:

• To introduce and validate a dynamic extension of component analysis for identifying dynamic modes (DMs) in fMRI time series.
• To characterize the spatio-temporal organization of brain activity by incorporating temporal information.
• To explore the relationship between dynamic brain network properties and behavioral/demographic measures.

Main Methods:

• Applied a novel dynamic component analysis framework to resting-state and motor-task fMRI data from 730 Human Connectome Project (HCP) subjects.
• Identified dominant dynamic modes (DMs) and characterized their temporal properties (oscillatory periods, damping times).
• Correlated temporal properties of resting-state DMs with 158 behavioral and demographic HCP measures.

Main Results:

• Dominant DMs in resting-state fMRI showed resemblance to known resting-state networks, with added temporal characterization.
• Motor-task DMs revealed novel spatio-temporal interactions in brain areas like the posterior parietal cortex and motor cortex, not evident in activation maps.
• Two canonical components linked resting-state DM temporal dynamics to behavioral and demographic data.

Conclusions:

• Dynamic component analysis is a feasible and relevant approach for analyzing fMRI data.
• DCA provides a richer spatio-temporal characterization of brain activity compared to static methods.
• The framework offers a promising tool for understanding brain functional organization and its relation to behavior.