In Neuron ; h5-index 148.0
Cerebral white matter undergoes a rapid and complex maturation during the early postnatal period. Prior magnetic resonance imaging (MRI) studies of early postnatal development have often been limited by small sample size, single-modality imaging, and univariate analytics. Here, we applied nonnegative matrix factorization, an unsupervised multivariate pattern analysis technique, to T2w/T1w signal ratio maps from the Developing Human Connectome Project (n = 342 newborns) revealing patterns of coordinated white matter maturation. These patterns showed divergent age-related maturational trajectories, which were replicated in another independent cohort (n = 239). Furthermore, we showed that T2w/T1w signal variations in these maturational patterns are explained by differential contributions of white matter microstructural indices derived from diffusion-weighted MRI. Finally, we demonstrated how white matter maturation patterns relate to distinct histological features by comparing our findings with postmortem late fetal/early postnatal brain tissue staining. Together, these results delineate concise and effective representation of early postnatal white matter reorganization.
Nazeri Arash, Krsnik Željka, Kostović Ivica, Ha Sung Min, Kopić Janja, Alexopoulos Dimitrios, Kaplan Sydney, Meyer Dominique, Luby Joan L, Warner Barbara B, Rogers Cynthia E, Barch Deanna M, Shimony Joshua S, McKinstry Robert C, Neil Jeffrey J, Smyser Christopher D, Sotiras Aristeidis
2022-Oct-13
MRI, Neurodevelopment, data-driven parcellation, histology, newborn, subplate remnant, tissue microstructure, topography, unsupervised machine learning, white matter maturation
