Summary: Researchers validated a new MRI harmonization technique called the traveling-subject (TS) approach to address inconsistencies in brain imaging studies of attention-deficit/hyperactivity disorder (ADHD). By scanning the same healthy volunteers across multiple MRI scanners, the team measured and corrected scanner-related bias while preserving true biological differences. Using TS-corrected data, they found reduced gray matter volume in frontotemporal regions of children with ADHD, suggesting potential neuroimaging biomarkers for earlier diagnosis and more individualized treatment.
ADHD affects more than five percent of children and adolescents worldwide. It is characterized by age-inappropriate inattention, hyperactivity, and impulsivity, which can interfere with learning, social relationships, and daily life. Structural MRI studies have sought to identify brain differences linked to ADHD, but results have varied: some reports describe reduced gray matter volume (GMV) in children with ADHD, while others find no change or even increases. Part of this inconsistency stems from small sample sizes, differences between MRI scanners, and sampling variation across study sites.
Key Facts
- TS Method Advantage: The traveling-subject approach reduces scanner measurement bias without removing biological sampling differences, unlike some conventional harmonization tools.
- Brain Findings: TS-corrected analyses revealed smaller frontotemporal gray matter volumes in children with ADHD, regions implicated in cognition and emotional regulation.
- Clinical Potential: TS-harmonized multi-site MRI data may help develop neuroimaging biomarkers to support earlier diagnosis, monitor treatment response, and enable personalized interventions for ADHD.
To evaluate and correct scanner-related measurement bias, the research team used the TS method alongside a commonly used harmonization tool called ComBat. Fourteen healthy volunteers (the traveling subjects) were scanned on four different MRI machines over three months to quantify scanner-specific measurement bias. These scanner bias estimates were then applied to an independent child dataset drawn from the Child Developmental MRI (CDM) database, which was built through a collaboration between the University of Fukui, The University of Osaka, and Chiba University to support research on neurodevelopmental disorders.
The study analyzed MRI data from 14 traveling subjects, 178 typically developing (TD) children, and 116 children with ADHD. Gray matter volumes were estimated using FreeSurfer, and statistical comparisons between ADHD and TD groups used mixed-effect models that accounted for multi-site structure.
When comparing correction strategies, the TS method substantially reduced measurement bias across scanners while preserving sampling bias—meaning it corrected for device differences without erasing population-specific biological signals. By contrast, ComBat also reduced measurement bias but tended to diminish sampling bias, which can remove meaningful biological variation tied to the sample.
Using TS-harmonized data, the researchers observed decreased brain volumes in frontotemporal regions among children with ADHD compared with typically developing peers. The strongest localized reduction was in the right middle temporal gyrus (TS-corrected data: β = −0.255, FDR p = 0.001). These frontotemporal regions support functions such as information processing, attention control, and emotion regulation, processes commonly affected in ADHD.
The authors suggest that harmonizing multi-site MRI data with the TS method offers a reliable way to detect structural brain differences associated with ADHD. If replicated and extended, these TS-corrected patterns could serve as neuroimaging biomarkers to help clinicians identify ADHD earlier, tailor treatments to individual neuroanatomical profiles, and monitor therapeutic outcomes over time.
“By applying TS harmonization to correct site-related biases in multi-site MRI data, this approach helps isolate brain structure characteristics linked to ADHD,” the team notes. “These characteristics could support earlier diagnosis and more precise individualized interventions, ultimately improving quality of life and reducing the risk of comorbid psychiatric conditions.”
Publication and authorship
This collaborative study was led by Assistant Professor Qiulu Shou and Associate Professor Yoshifumi Mizuno (University of Fukui), with contributions from Professor Yoshiyuki Hirano (Chiba University) and Professor Kuriko Kagitani-Shimono (The University of Osaka). The findings were published in Molecular Psychiatry on August 8, 2025.
Funding information
This work was supported by grants and fellowships from the Japan Society for the Promotion of Science (JSPS KAKENHI; grant numbers: 24K16647, 24K21453, 21K02380, 23K12814, 23H00949, 22H01090, 23K02956, 23K07004, 24K21493), the Kawano Masanori Memorial Public Interest Incorporated Foundation for Promotion of Pediatrics, The Mother and Child Health Foundation, the Japan–U.S. Brain Research Cooperative Program, the University of Fukui, The Taiju Life Social Welfare Foundation, and the Collaborative Research Program of the Collaborative Research Network for Asian Children with Developmental Disorders (MEXT Policy Initiative).
About this neuroimaging and ADHD research news
Author: Yuuka Kawamoto
Source: University of Fukui
Contact: Yuuka Kawamoto – University of Fukui
Image credit: Neuroscience News
Original Research: Open access. “Brain structure characteristics in children with attention-deficit/hyperactivity disorder elucidated using traveling-subject harmonization” by Qiulu Shou et al., Molecular Psychiatry.
Abstract
Brain structure characteristics in children with attention-deficit/hyperactivity disorder elucidated using traveling-subject harmonization
Brain imaging studies of ADHD have produced inconsistent findings, in part because MRI scanners introduce measurement bias. This study applied a harmonization strategy—the traveling-subject (TS) method—to multi-site MRI data to reduce scanner-related measurement bias and clarify brain structural characteristics of children with ADHD.
MRI data from 14 traveling subjects, 178 typically developing children, and 116 children with ADHD were collected across multiple sites. The TS method and ComBat harmonization were applied to correct measurement bias. Gray matter volumes were estimated with FreeSurfer, and group differences were assessed using mixed-effect models. Compared with raw data, TS correction significantly reduced measurement bias while retaining sampling bias; ComBat reduced measurement bias but also markedly decreased sampling bias. TS-corrected results revealed frontotemporal volume reductions in the ADHD group, with the most notable effect observed in the right middle temporal gyrus (β = −0.255, FDR p = 0.001). These findings indicate that TS harmonization can improve the reliability of multi-site MRI studies and help identify structural brain features associated with ADHD.