Constructing fine-grained spatiotemporal neonatal functional atlases with spectral functional network learning.

Area of Science:

• Neuroscience
• Developmental Biology
• Medical Imaging

Background:

• Early human brain development is critical for cognitive and behavioral outcomes.
• Spatiotemporal (4D) brain functional atlases are essential for understanding brain development.
• Existing 4D atlases are limited for early life due to fMRI noise and rapid brain changes.

Purpose of the Study:

• To develop a novel, data-driven framework for generating high-quality 4D functional brain atlases in early life.
• To address challenges of noise and temporal inconsistency in functional magnetic resonance imaging (fMRI) data for atlas creation.
• To produce the first neonatal 4D functional atlases with fine-grained spatiotemporal resolution.

Main Methods:

• Integrated low-rank tensor learning with spectral embedding for a Spectral Functional Network Learning (SFNL) framework.
• Utilized low-rank tensor learning to capture common functional connectivity (FC) patterns across ages, optimizing FCs for temporal consistency.
• Employed spectral embedding to mitigate fMRI noise by reconstructing FC networks in spectral space.

Main Results:

• Successfully generated the first neonatal 4D functional atlases using the developing Human Connectome Project (dHCP) dataset.
• Demonstrated superior performance in functional homogeneity, reliability, and temporal consistency compared to existing methods.
• Validated the framework's efficacy and robustness through network analysis, including individual identification and functional system development.

Conclusions:

• The SFNL framework provides a robust method for creating consistent, high-quality spatiotemporal functional atlases for early human brain development.
• The generated neonatal 4D atlases offer unprecedented spatiotemporal resolution, advancing developmental neuroscience research.
• The 4D atlases and associated code are publicly available to facilitate further research.