Better Brain Imaging Method Sharpens ADHD Insights

Researchers in Japan have shown that a harmonization approach called the traveling-subject (TS) method can reduce measurement bias in brain imaging studies, helping address inconsistencies in magnetic resonance imaging (MRI) results for children with attention deficit/hyperactivity disorder (ADHD). The method could improve the comparability of imaging data collected across multiple research sites.

ADHD affects more than 5% of children and adolescents worldwide. The condition is marked by inattention, impulsivity or hyperactivity that is not age-appropriate. Although MRI is widely used to explore the neurobiological basis of ADHD, previous research has yielded conflicting results about structural differences in the brains of children with the condition.

Some studies have reported lower gray matter volume (GMV) in children with ADHD, while others have found no significant differences or even increased GMV when compared with typically developing (TD) children. The inconsistencies have been attributed to small sample sizes and differences in MRI scanners or image acquisition protocols.

Aiming to correct for scanner-related bias

To address these limitations, researchers used the TS harmonization method, which involves scanning the same individuals at multiple sites to quantify scanner-related differences. In this study, 14 healthy volunteers were scanned at 4 different institutions over 3 months to identify measurement bias. These estimates were then applied to a larger dataset of MRI scans collected from the Child Developmental MRI (CDM) database, which included data from 116 children with ADHD and 178 TD children.

The CDM database is a collaborative initiative by the University of Fukui, The University of Osaka and Chiba University, aimed at gathering neuroimaging data from more than 1,000 children to support research into neurodevelopmental conditions such as ADHD.

Researchers compared the TS method with ComBat harmonization, a statistical technique that controls for differences between sites and scanners in large datasets. While both methods reduced measurement bias, ComBat also removed sampling bias – potentially eliminating meaningful biological variability. In contrast, the TS method preserved these between-group differences.

Distinct volumetric changes in brain regions

After applying the TS method, researchers found that children with ADHD had smaller brain volumes in frontotemporal regions compared to their TD peers. These areas are associated with functions such as emotional regulation and cognitive processing, which are often altered in individuals with ADHD.

The study indicates that TS harmonization can retain biologically relevant differences between groups while controlling for technical variability. This suggests the method could help researchers more accurately identify structural biomarkers associated with ADHD.

Towards more consistent multi-site imaging studies

The findings support the use of the TS method in future multi-site imaging studies of ADHD and other neurodevelopmental disorders. While the approach does not eliminate all variability, it offers a potential solution to a long-standing problem in neuroimaging research: distinguishing true biological signals from scanner-induced noise.

As imaging datasets grow and collaborations across institutions become more common, harmonization methods like TS may play an increasingly important role in ensuring data comparability across studies.

Reference: Shou Q, Yamashita M, Hirano Y, et al. Brain structure characteristics in children with attention-deficit/hyperactivity disorder elucidated using traveling-subject harmonization. Mol Psychiatry. 2025. doi: 10.1038/s41380-025-03142-6

This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source. Our press release publishing policy can be accessed here.

This content includes text that has been generated with the assistance of AI. Technology Networks' AI policy can be found here.