Brain Connectivity Modelling Through Joint Estimation of Parcels and Gradients

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Network Science

Background:

• Understanding whole-brain connectivity topography is crucial for neuroscience.
• Existing models struggle to disentangle functional segregation from connectivity gradients.
• Resting-state functional MRI (fMRI) data offers a window into intrinsic brain organization.

Purpose of the Study:

• To develop a novel framework for modeling brain connectivity topography in resting-state fMRI.
• To differentiate between functional segregation (abrupt connectivity changes) and connectivity gradients (smooth variations).
• To provide insights into the organizational principles of poorly characterized brain regions.

Main Methods:

• Proposed a framework to model whole-brain connectivity topography using resting-state fMRI.
• Assumed functional segregation corresponds to low-rank structure and gradients to sparse, non-low-rank structure in the connectome.
• Decomposed the connectome into low-rank and sparse components, integrating local-nonlinear and global-linear embedding strategies.

Main Results:

• The hybrid model (low-rank + sparse) approximated the empirical dense connectome better than purely low-rank or gradient approaches.
• Derived connectivity gradients showed strong correspondence with task-based topographic maps.
• The framework successfully disentangled functional segregation from connectivity gradients.

Conclusions:

• The proposed hybrid modeling framework effectively captures brain connectivity topography.
• This approach offers a promising method for analyzing brain organization and understanding functional segregation and gradients.
• The findings contribute to characterizing organizational principles in complex brain networks.