Functional covariance networks: obtaining resting-state networks from intersubject variability

Area of Science:

• Neuroscience
• Brain Imaging
• Functional Connectivity

Background:

• Resting-state networks (RSNs) are crucial for understanding brain function.
• Current methods often analyze time series data, but parameter-based approaches offer new insights.
• Investigating group-level RSN separation requires robust analytical techniques.

Purpose of the Study:

• To develop and validate a novel approach for group-level RSN separation using resting-state parameters.
• To determine if resting-state functional connectivity (RSFC) parameters, not non-RSFC parameters, are key to elucidating RSNs.
• To explore the neurophysiological origins of low-frequency fluctuations (LFFs) and RSNs.

Main Methods:

• Utilized spatial independent component analysis (ICA) on resting-state parameters: amplitude of low-frequency fluctuation (ALFF), fractional ALFF (fALFF), Hurst exponent, and signal standard deviation.
• Analyzed covariance patterns of these parameters across subjects, excluding time series data.
• Compared results with non-RSFC parameters like BOLD signal mean and gray matter volume.

Main Results:

• Identified "functional covariance networks" by analyzing resting-state parameter covariance, which strongly correlated with known RSN maps.
• Demonstrated that RSFC parameters, particularly LFF properties, are primary drivers for revealing RSNs.
• Confirmed the presence of a common influence underlying individual RSFC networks and suggested neurophysiological origins.

Conclusions:

• The proposed method effectively identifies RSNs through functional covariance networks derived from resting-state parameters.
• RSFC parameters are superior to non-RSFC parameters in defining meaningful RSNs.
• Findings support the neurophysiological basis of LFFs and their role in RSN organization.