Non-negative data-driven mapping of structural connections with application to the neonatal brain

Area of Science:

• Neuroscience
• Developmental Biology
• Medical Imaging

Background:

• Mapping neonatal brain connections is vital for understanding early neurodevelopment, cognition, and behavior.
• Traditional atlas-based tractography faces challenges in the rapidly developing neonatal brain due to anatomical changes.
• Data-driven methods offer an alternative by not relying on predefined regions of interest.

Purpose of the Study:

• To develop and validate a novel data-driven framework for extracting white matter bundles and grey matter networks from neonatal diffusion MRI tractography data.
• To assess the accuracy and interpretability of the proposed framework compared to existing methods and traditional approaches.
• To generate a robust and reproducible connectivity-driven parcellation of the neonatal cortex.

Main Methods:

• Development of a novel data-driven framework using non-negative matrix factorization (NMF) for neonatal tractography.
• Implementation of a non-negative dual regression framework to map group-level components to individual subjects.
• Validation using in-silico simulations, comparison with independent component analysis (ICA), and application to the Developing Human Connectome Project dataset.

Main Results:

• The NMF framework successfully extracted white matter bundles and associated grey matter networks from neonatal data.
• The approach demonstrated accuracy in simulations and yielded interpretable components when compared to traditional methods and resting-state fMRI.
• A robust and reproducible connectivity-driven parcellation of the neonatal cortex was generated using data from 323 newborns.

Conclusions:

• The developed data-driven NMF framework provides a valid and robust method for analyzing neonatal brain connectivity.
• This approach overcomes limitations of atlas-based methods in the developing brain.
• The resulting connectivity-driven parcellation offers a valuable tool for studying neonatal brain organization and development.