Harmonized diffusion MRI data and white matter measures from the Adolescent Brain Cognitive Development Study

Abstract

The Adolescent Brain Cognitive Development (ABCD) study is a groundbreaking effort aimed at providing a comprehensive understanding of adolescent brain development. With data collected from over 10,000 children across 21 sites, this study promises to unlock key insights into the cognitive, behavioral, and neuroimaging data that underpins this critical period of development. However, the potential use of diffusion MRI (dMRI) data in this study is hindered by site-specific scanner and protocol variability, which precludes naive pooling of the data from across sites for analysis. In response to this challenge, we have developed an aggregated database of harmonized dMRI data, comprising imaging data from 9345 subjects and their white matter tracts, as well as tract-specific microstructural measures. This resource has been made publicly available via the NIMH Data Archive (NDA). The generation of this database required enormous computational effort, taking ~50,000 CPU hours to effectively harmonize the dMRI data, run whole brain tractography, and perform white matter parcellation. The data available through the NDA includes: harmonized dMRI data, several dMRI derived measures (e.g., fractional anisotropy) for 73 anatomically well-defined white matter fiber bundles in all 9345 subjects and streamlines corresponding to each fiber bundle for visualization. In this work, we further demonstrate the efficacy of our dMRI data harmonization algorithm on the ABCD dataset using multiple experiments, allowing researchers to perform their own analyses on the harmonized data. The data resources made available through this work will allow for new structural connectivity studies of neurodevelopment in children and adolescents at an unprecedented scale.

Large‐scale investigation of white matter structural differences in bilingual and monolingual children: An adolescent brain cognitive development data study

“…The ABCD Study dataset provides the advantages of a largescale data set, but since it has been collected across multiple sites with site-specific scanner variability, there are challenges to using and processing the raw dMRI data. To overcome this challenge, the ABCD team has harmonized data across sites, performed the white matter parcellations, and ran the whole-brain tractography resulting in quality-controlled neuroimaging data for over 9000 of their participants (Cetin-Karayumak et al, 2023). The present study utilizes the harmonized data provided by the ABCD Study researchers to avoid potential processing issues.…”

Executive functioning, behavior, and white matter microstructure in the chronic phase after pediatric mild traumatic brain injury: results from the adolescent brain cognitive development study

“…To this end, the minimally processed dMRI data (3.0 release) were harmonized with our well-validated approach based on rotation invariant spherical harmonics (RISH, https://github.com/pnlbwh/dMRIharmonization; Cetin- Karayumak et al, 2019Karayumak et al, , 2020. Details on the specifics of the harmonization used in the current study are reported by Cetin- Karayumak et al (2023b). We note that diffusion data acquired on MRI scanners manufactured by Philips did not pass quality control criteria during harmonization (Cetin- Karayumak et al, 2023b), which led to the exclusion of n = 38 children with mTBI, n = 75 TDC and n = 80 OI controls from imaging analyses.…”

“…Details on the specifics of the harmonization used in the current study are reported by Cetin- Karayumak et al (2023b). We note that diffusion data acquired on MRI scanners manufactured by Philips did not pass quality control criteria during harmonization (Cetin- Karayumak et al, 2023b), which led to the exclusion of n = 38 children with mTBI, n = 75 TDC and n = 80 OI controls from imaging analyses. The final sample size for dMRI analyses used in this study was n = 321 children with mTBI (n = 636 TDC, n = 656 OI).…”

“…To investigate the neurophysiological mechanisms underlying the behavioral differences reported above, we examined white matter microstructure in several tracts that have previously been associated with mTBI, executive functioning, and the related behavior. To account for non-linear scanner effects, we used the newly available diffusion MRI data that were harmonized across study sites (Cetin-Karayumak et al, 2023b). The effectiveness of this harmonization approach has been demonstrated in several neuroimaging studies (e.g., Cetin Karayumak, Kubicki, & Rathi, 2018, Cetin-Karayumak et al, 2020, 2023aDe Luca et al, 2022;Di Biase et al, 2021;Elad et al, 2021;Lv et al, 2021;Seitz et al, 2021;Ye et al, 2021).…”