Detecting and harmonizing scanner differences in the ABCD study - annual release 1.0
Dylan M. NielsoniD, Francisco PereiraiD, Charles Y. ZhengiD, Nino MigineishviliiD, John Anthony LeeiD, Adam G. ThomasiD, Peter A. BandettiniiD
Abstract
In order to obtain the sample sizes needed for robustly reproducible effects, it is often necessary to acquire data at multiple sites using different MRI scanners. This poses a challenge for investigators to account for the variance due to scanner, as balanced sampling is often not an option. Similarly, longitudinal studies must deal with known and unknown changes to scanner hardware and software over time. In this manuscript, we have explored scanner-related differences in the dataset recently released by the Adolescent Brain Cognitive Development (ABCD) project, a multi-site, longitudinal study of children age 9-10. We demonstrate that scanner manufacturer, model, as well as the individual scanner itself, are detectable in the resting and task-based fMRI results of the ABCD dataset. We further demonstrate that these differences can be harmonized using an empirical Bayes approach known as ComBat. We argue that accounting for scanner variance, including even minor differences in scanner hardware or software, is crucial for any analysis.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 51
- Assay
- Research Focus
- Approach
Abstract summary
Claude Opus 5.5 · xhigh effort · 18 Oct 2026 · from the abstract
Scanner manufacturer, model and even the individual machine leave detectable signatures in ABCD resting-state and task fMRI results; ComBat harmonization is reported to remove them.
Multi-site imaging consortia rarely balance scanner assignment against the variables of interest, so differences between machines can be mistaken for biology, and longitudinal designs add drift as hardware and software change mid-study. The authors ask whether such signatures are actually present in a multi-site, longitudinal paediatric fMRI dataset, and whether an existing empirical Bayes correction can be applied to remove them.
Key results
SuggestedWhich scanner acquired a scan can be recovered from the fMRI results themselves: not only the vendor and the model, but the particular device, in both resting-state and task data.
Reported for the ABCD release examined; the abstract gives no classification accuracy, effect size or statistical test for this detectability. · Abstract
SuggestedApplying ComBat, an empirical Bayes harmonization method, removed the scanner-linked differences the authors detected.
Stated as a demonstration on the same ABCD fMRI measures; no residual-effect numbers, comparison with other harmonization methods, or check that biological signal survived are reported in the abstract. · Abstract
ClaimedThe authors conclude that any analysis of these data should model scanner variance, down to small hardware or software differences between otherwise identical machines.
Presented as an argument in the abstract rather than a tested claim; no example of an analysis whose conclusions change with and without correction is given. · Abstract
Methods, data and reproducibility
Resting-state and task-based fMRI results from the Adolescent Brain Cognitive Development (ABCD) multi-site longitudinal cohort were examined for signatures of the acquiring scanner at three levels — manufacturer, model and individual device — and the data were then adjusted with ComBat, an empirical Bayes harmonization method. The abstract does not state which fMRI measures or tasks were analysed.
- Resting-state and task-based fMRI from the ABCD study, a multi-site longitudinal cohort of children aged 9-10 · human
- Methods & tools
- ComBat
Limitations (2)
- No quantities are given: neither how strongly scanner identity is detectable nor how much of it ComBat removes can be gauged from the text available. · noted in this summary
- Only one correction method is reported, on one data release, and the abstract does not say whether harmonization leaves age, task or individual-difference effects intact. · noted in this summary
Who should read:Read this if you analyse multi-site or longitudinal MRI such as ABCD and need to know whether scanner identity shows up in fMRI measures and whether ComBat is an adequate fix.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Harmonization of cortical thickness measurements across scanners and sites
Jean‐Philippe Fortin, Nicholas C. Cullen, … Russell Taki Shinohara·NeuroImagePeer-reviewed·Nov 20, 2017
- A site-wise reliability analysis of the ABCD diffusion fractional anisotropy and cortical thickness: impact of scanner platforms
Yezhi Pan, L. Elliot Hong, … Shuo Chen·Human Brain MappingPeer-reviewed·Nov 1, 2024
- Longitudinal ComBat: A Method for Harmonizing Longitudinal Multi-scanner Imaging Data
Joanne C. Beer, Nicholas James Tustison, … Kristin A. Linn·NeuroImagePeer-reviewed·Jul 5, 2020
- ComBat Harmonization: Empirical Bayes versus Fully Bayes Approaches
Maxwell J. Reynolds, Tigmanshu Chaudhary, … the Alzheimer’s Disease Neuroimaging Initiative·NeuroImage ClinicalPeer-reviewed·Jan 1, 2023
- Harmonizing Functional Connectivity Reduces Scanner Effects in Community Detection
Andrew A. Chen, Dhivya Srinivasan, … Haochang Shou·NeuroImagePeer-reviewed·Apr 11, 2022
- ComCat: Combating Covariate Effects in Brain Analysis
Christian Gaser B, Robert Dahnke, … Thomas E. Nichols·bioRxiv·Jun 23, 2026