A Statistical Characterization of Dynamic Brain Functional Connectivity

Area of Science:

• Neuroscience
• Statistics
• Psychiatry

Background:

• Dynamic functional connectivity (dFC) in mental disorders remains debated due to a lack of empirical distribution analysis.
• Previous research has not sufficiently demonstrated the non-stationarity of functional connectivity distributions.

Purpose of the Study:

• To provide statistical evidence for the existence of dFC by analyzing the empirical probability distribution of functional connectivity.
• To investigate dFC patterns in individuals with ADHD, bipolar disorder, and schizophrenia compared to healthy controls.

Main Methods:

• Utilized resting-state fMRI data from 186 subjects from the UCLA Consortium for Neuropsychiatric Phenomics dataset.
• Employed individual voxel signals and Pearson's correlation to analyze temporal variations and non-stationarity in functional connectivity.
• Applied beta distribution to model the empirical distribution of correlation coefficients across 49 regions of interest.

Main Results:

• The empirical distribution of the correlation coefficient demonstrated non-stationarity.
• The beta distribution accurately approximated the exact correlation coefficient distribution.
• Distinct dFC patterns, derived from fitted beta distributions, were identified and showed potential as biomarkers for mental disorders.

Conclusions:

• The study presents the first comprehensive empirical distribution analysis of dFC, providing strong evidence for its existence.
• Identified unique dFC patterns across different mental disorders, suggesting their potential utility as biomarkers.
• Introduced an innovative statistical approach to understand the dynamic nature of functional connectivity in resting-state fMRI data.