Overview
- The study validated a traveling-subject strategy by scanning 14 healthy volunteers on four machines to estimate measurement bias and applied those corrections to a multi-site pediatric dataset.
- Across 178 typically developing children and 116 with ADHD from the CDM database, the method significantly lowered measurement bias while preserving sampling variance.
- ComBat also reduced measurement bias but markedly diminished sampling variance, which can mask real group differences in multi-site data.
- TS-corrected analyses identified smaller gray matter volumes in frontotemporal regions in the ADHD group, including a significant reduction in the right middle temporal gyrus (FDR p=0.001).
- The findings were published August 8, 2025, in Molecular Psychiatry, and researchers plan to expand CDM to more than 1,000 participants and test TS-derived patterns as candidate biomarkers pending further validation.