INVESTIGATING DIFFERENCES IN BRAIN FUNCTIONAL NETWORKS USING HIERARCHICAL COVARIATE-ADJUSTED INDEPENDENT COMPONENT...

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Statistical Modeling

Background:

• Human brain function relies on complex networks of distinct brain regions.
• Independent Component Analysis (ICA) is widely used in fMRI to identify these functional networks.
• Current ICA methods struggle to directly account for clinical and demographic variabilities in network analysis.

Purpose of the Study:

• To develop a formal statistical framework, hierarchical covariate-adjusted ICA (hc-ICA), for analyzing covariate effects in brain functional networks.
• To enable reliable and powerful statistical testing of group differences in brain networks while controlling for confounding factors.
• To investigate functional network differences in individuals with post-traumatic stress disorder (PTSD).

Main Methods:

• Proposed a novel hierarchical covariate-adjusted ICA (hc-ICA) model.
• Developed an analytically tractable Expectation-Maximization (EM) algorithm for maximum likelihood estimation.
• Introduced a fast, subspace-based approximate EM algorithm and a voxel-wise approximate inference procedure for efficient group difference testing.

Main Results:

• Simulation studies demonstrated the advantages of hc-ICA over existing methods.
• The method provides a more reliable and powerful tool for evaluating group differences in brain functional networks.
• An application to an fMRI study revealed differences in brain functional networks associated with PTSD.

Conclusions:

• The hc-ICA model offers a statistically rigorous approach to incorporate covariate effects into ICA decomposition for fMRI data.
• The proposed algorithms enhance computational efficiency and accuracy in analyzing group differences.
• This method advances the understanding of how clinical conditions like PTSD affect brain functional networks.