Advanced diffusion MRI dataset offers new view of adolescent brain development

Researchers have created a large, openly available neuroimaging resource from more than 24,000 MRI scans of over 10,000 adolescents, providing an unprecedented view of how the brain’s wiring develops during adolescence. Using advanced, open-source image processing methods, the team identified highly consistent patterns of development across 21 study sites, despite differences in scanner technology and image quality.

The study, led by researchers at the Perelman School of Medicine at the University of Pennsylvania with collaborators across the country, was published today in Neuron.

“We showed that advanced state-of-the-art approaches to diffusion MRI can reveal aspects of adolescent brain development that conventional measures may miss,” said senior author Ted Satterthwaite, MD, the McLure II Professor of Psychiatry and Behavioral Research and director of the Penn Lifespan Informatics and Neuroimaging Center (PennLINC) at Penn Medicine. “Despite substantial differences in image quality across tens of thousands of scans acquired by different scanners, we found remarkably consistent patterns of white matter development.”

Building a large, open resource

The research used data from the National Institutes of Health-funded Adolescent Brain Cognitive Development (ABCD) Study, the largest long-term study of brain development and child health in the United States. The researchers processed more than 24,000 scans collected from over 10,000 adolescents at multiple points in time.

The team focused on diffusion MRI, a type of imaging that measures how water moves through brain tissue and can provide information about the microscopic organization of white matter—the connections that allow different parts of the brain to communicate.

The resulting dataset, the ABCD-BIDS Community Collection, makes these extensively processed imaging data available as a resource for other researchers studying adolescent brain development.

“Large studies like ABCD give us enormous statistical power, but they also introduce major technical challenges,” said first author Steven Meisler, PhD, a PennLINC postdoctoral fellow. “When scans are collected on different scanners, with different software and at different sites, those technical differences can be larger than the biological changes we are trying to measure.”

Finding consistent patterns across sites

To address this challenge, the researchers used advanced open-source image-processing and statistical methods created by Penn researchers designed for large, multisite neuroimaging data. After accounting for technical differences between scans, they examined how multiple features of white matter changed across adolescence.

The analyses revealed widespread and highly consistent developmental patterns. Importantly, more advanced diffusion MRI measures that capture more complex features of white matter showed developmental changes that were less apparent using traditional measures. The researchers also found that harmonizing data across scanners and sites substantially improved the consistency of developmental patterns observed across the 21 study locations.

A resource designed to be reused

Beyond these findings, the researchers say a central goal of the project was to make large-scale neuroimaging analyses easier to reproduce and extend.

“Large research consortia generate extraordinary datasets, but the processing needed to make those data scientifically useful is often difficult for individual researchers to reproduce,” Meisler said. “By releasing both the processed data and the open-source tools used to create them, we hope other researchers can build on this work rather than having to start from scratch.”

Researchers in the PennLINC laboratory are already using the resource to investigate how white matter development relates to environmental exposures and other aspects of adolescent health and development.