FedDOSE: Federated Learning Framework Decomposing Site Effects for Modeling Brain Dynamic Functional Connectivity

2026-08-07

AI Summary

This work addresses the challenge of site-induced statistical heterogeneity in multi-site fMRI data within federated learning, a problem exacerbated by the frequent neglect of the brain’s dynamic functional characteristics. To this end, the authors propose FedDOSE, a novel framework that jointly models dynamic functional connectivity (dFC) and site heterogeneity for the first time in a federated setting. FedDOSE efficiently encodes high-dimensional dFC tensors via module-guided Tucker decomposition and aligns cross-site class prototypes through an integration of optimal transport centroids and Procrustes analysis. Evaluated on the ABIDE-I, ABIDE-II, and ADHD-200 datasets, the method demonstrates superior performance over existing approaches, exhibiting enhanced generalization and robustness in diagnostic tasks for autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD).

Abstract

Functional Magnetic Resonance Imaging ( fMRI ) data are often pooled into collaborative multi-site consortia, as deep learning models for analyses require large datasets to generalize well. While Federated Learning (FL) offers a privacy-preserving paradigm for collaborative training, standard approaches continue to struggle with statistical heterogeneity. In particular, site differences pose a key challenge in multi-site data settings. Additionally, existing FL approaches for fMRI rely on static Functional Connectivity ( FC), omitting dynamic information in brain networks. To address this, we propose FedDOSE, a novel framework that explicitly decomposes site differences for analysis of dynamic FC (dFC). FedDOSE introduces a Modularity-Guided Tucker Decomposition block to encode high-dimensional dFC tensors and capture modular-level spatio-temporal patterns efficiently. Class-specific prototypes are generated across all sites and subsequently aligned at the global level by using a combination of Optimal Transport (OT) barycenter formulation and Procrustes analysis. Extensive experiments for diagnosing Autism Spectrum Disorder (ASD) and Attention-Deficit Hyperactivity Disorder (ADHD) on three multi-site resting-state fMRI datasets: ABIDE-I, ABIDE-II, and ADHD-200, demonstrate that FedDOSE outperforms state-of-the-art methods in ASD and ADHD detection. Our results highlight its effectiveness in learning robust representations from multi-site datasets for reliable analysis.

Problem

• Federated Learning
• site effects
• dynamic Functional Connectivity
• multi-site fMRI
• statistical heterogeneity

Innovation

• Federated Learning
• Dynamic Functional Connectivity
• Tucker Decomposition
• Optimal Transport
• Multi-site fMRI