Revealing Developmental Regionalization of Infant Cerebral Cortex Based on Multiple Cortical Properties

Area of Science:

• Neuroscience
• Developmental Biology
• Computational Biology

Background:

• The human brain exhibits dynamic and heterogeneous cortical development in the first two postnatal years.
• Understanding early cortical regionalization is crucial for localizing brain regions and advancing knowledge of brain development.
• Existing knowledge is limited by the lack of computational tools and large-scale datasets for analyzing early cortical development.

Purpose of the Study:

• To explore and delineate early cortical developmental regionalization.
• To address methodological and knowledge gaps in understanding infant brain development.
• To establish neurobiologically meaningful units for precise region localization.

Main Methods:

• Developed a novel multi-view nonnegative matrix factorization (MV-NMF) method to analyze complex, high-dimensional neuroimaging data.
• Incorporated multiple complementary cortical properties (views) for a comprehensive analysis of regionalization.
• Applied orthogonal and graph-induced constraints to ensure sparsity and compactness of discovered cortical regions.

Main Results:

• Utilized an unprecedented dataset of 1,560 longitudinal MRI scans from 887 infants.
• Delineated the first neurobiologically meaningful representation of early cortical regionalization.
• Identified distinct and complementary cortical properties contributing to the regionalization process.

Conclusions:

• The novel MV-NMF method effectively captures early cortical developmental regionalization.
• The study provides a valuable reference map for understanding infant brain development.
• This work advances the precise localization of brain regions during a critical developmental period.