An open, analysis-ready, and quality controlled resource for pediatric brain white-matter research

Abstract

We created resources to facilitate research on the role of human brain microstructure in the development of mental health disorders, based on openly-available diffusion MRI (dMRI) data from the Healthy Brain Network (HBN) study. First, we curated the HBN dMRI data (N=2747) into the Brain Imaging Data Structure and preprocessed it according to best-practices, including denoising and correcting for motion effects, susceptibility-related distortions, and eddy currents. Preprocessed, analysis-ready data was made openly available. Data quality plays a key role in the analysis of dMRI, and we provide automated quality control (QC) scores for every scan, as part of the data release. To scale QC to this large dataset, we trained a neural network through the combination of a small data subset scored by experts and a larger set scored by community scientists. The network performs QC highly concordant with that of experts on a held out set (ROC-AUC=0.947). A further analysis of the neural network demonstrates that it relies on image features with relevance to QC. Altogether, this work both delivers a resource for transdiagnostic research in brain connectivity and pediatric mental health and serves as a novel tool for automated QC of large datasets.

White matter and literacy: a dynamic system in flux

“…Using the HBN dataset(32), one of the most diverse large-scale pediatric samples to date, we compared the white matter diffusion properties of struggling readers (again using a typical cutoff for dyslexia) and a control group. The HBN data was processed with QSIPrep(41), rigorous quality control was applied to each participant’s data(42), and tractometry was performed with pyAFQ(43). Group comparisons focused on the left arcuate, and left ILF, and posterior callosum, which have been most consistently implicated in past studies examining the relationship between the white matter and reading skill(26).…”

Fiber-Specific Structural Properties Relate to Reading Skills in Children and Adolescents

“…Phenotypic data were accessed in accordance with a data use agreement provided by the Child Mind Institute. Preprocessed DWI data were provided as part of the HBN Preprocessed Open Diffusion Derivatives (HBN-POD2) data set ( Richie-Halford et al, 2022 ). The Healthy Brain Network project was approved by the Chesapeake Institutional Review Board (now called Advarra, Inc.; https://www.advarra.com/).…”

“…We also considered multiple quality covariates, including mean framewise displacement, and neighbor correlation ( Yeh et al (2019 )). The machine-learning based quality score distribution from Richie-Halford et al (2022 ) was skewed towards 1 and not normally distributed, and thus was not a good candidate confound. Since gFD and gFC are calculated within fixels, and fixels are only segmented in white matter, differences in white matter volumetric proportions should not influence global fixel-metrics.…”

“…We downloaded preprocessed DWI (Richie-Halford et al, 2022) and phenotypic data from 2136 participants across the first eight data releases of the Healthy Brain Network (HBN) project (Alexander et al, 2017). HBN-POD2 distributes a quality metric accompanying each image that predicts the probability that the image would pass manual expert quality review ("xgb_qc_score", or "dl_qc_score" if the former score was not available) (Richie-Halford et al, 2022). It ranges from 0 (no chance of passing expert review) to 1 (image will definitely pass expert review).…”

“…We excluded any participants with a quality score of less than 0.5. Twenty different DWI acquisition parameters were present across participants (Covitz et al, 2022;Richie-Halford et al, 2022). We only included participants who had images acquired with the most common acquisition parameters in their site ("SITE_64dir_most_common").…”