Brain network re-estimation improves accuracy across imaging sites
Beyond Site Agreement: Re-estimation for Brain Network Generalization
Machine Learning
Summary
Brain scans from different hospitals or machines often show differences that make it hard to use one brain analysis method everywhere. The authors found that simply agreeing on brain connections across sites isn’t enough because the measurements can change even within one scan. They created a method called BRIO that checks how stable these brain connections are within each scan to better align data from different sites. This approach improved the accuracy of brain disorder classification across multiple datasets.
What this means in practice
- •For clinical neuroimaging teams: Enhance diagnostic models by improving brain scan data consistency across hospitals and scanners for psychiatric disorder classification.
- •For medical imaging software developers: Incorporate within-scan brain connectivity stability checks to improve machine learning models that analyze resting-state fMRI data from multiple sources.
Authors
Yingxu Wang, Kunyu Zhang, Yanwu Yang3, Thomas Wolfers, Yujie Wu, Siyang Gao, Nan Yin
Abstract
Cross-site out-of-distribution (OOD) generalization in resting-state functional magnetic resonance imaging (rs-fMRI) often relies on learning task-discriminative representations from full-scan functional connectivity (FC) graphs and promoting invariance across source sites. However, FC graphs are estimated from finite, temporally correlated blood-oxygen-level-dependent (BOLD) sequences. Cross-site agreement therefore does not necessarily imply that predictive evidence remains supported under FC re-estimation within the same scan. In this paper, we propose Brain Network Re-estimation-Informed OOD Learning (BRIO), a framework that uses within-scan FC re-estimation to guide cross-site alignment. BRIO maps fullscan graphs and their re-estimates into consistently indexed connectome factors, enabling comparisons of their predictive contributions. It assesses re-estimation support from changes in these contributions relative to within-class subject variability and class separation. For each source-site pair and class, this task-calibrated support from both sites is combined with predictive relevance to form pairwise qualifications, which determine relative factor weights and overall alignment strength. Leave-one-site-out experiments on four real-world datasets (ABIDE, REST-metaMDD, SRPBS, and ABCD) show that BRIO consistently outperforms competitive baselines, with relative improvements of up to 3.8% in accuracy. These gains also persist under an alternative brain parcellation on ABIDE.