Hierarchical Extraction of Functional Connectivity Components in Human Brain Using Resting-State fMRI

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Network Science

Background:

• Functional brain networks are crucial in cognitive neuroscience, but traditional methods often analyze them at a single scale.
• Evidence suggests a hierarchical organization of brain function, necessitating multi-scale analytical approaches.
• High dimensionality of functional connectivity data poses a challenge for accurate analysis.

Purpose of the Study:

• To develop a novel method for extracting hierarchical functional connectivity components from resting-state fMRI data.
• To introduce a multi-scale decomposition approach building on Sparse Connectivity Patterns (SCPs).
• To capture interpretable, hierarchically organized, and heterogeneous brain patterns.

Main Methods:

• Utilized resting-state fMRI data to generate correlation matrices.
• Developed a cascaded factorization method to extract hierarchical Sparse Connectivity Patterns (SCPs).
• Formulated the decomposition as a non-convex optimization problem solved with adaptive gradient descent.

Main Results:

• Validated the novel method using simulated data and two real-world fMRI datasets.
• Demonstrated that multi-scale hierarchical SCPs are reproducible across data sub-samples.
• Showed superior reproducibility of hierarchical patterns compared to single-scale patterns.

Conclusions:

• The proposed method effectively extracts reproducible, multi-scale hierarchical functional brain networks.
• Hierarchical decomposition offers a more comprehensive understanding of brain functional organization.
• The approach advances the analysis of high-dimensional neuroimaging data.