Progressive Self-Supervised Learning with Individualized Community Assignment for Brain Network Analysis

2026-06-28

AI Summary

Existing self-supervised learning approaches in brain network analysis often overlook the heterogeneity of functional community structures across individuals, limiting their ability to capture subject-specific organizational patterns. To address this, this work proposes BrainPICM, a novel framework that explicitly incorporates individualized functional communities into masked modeling for the first time. BrainPICM employs progressive unbalanced optimal transport to model soft assignments of regions of interest (ROIs) to communities, combined with a confidence-guided curriculum masking strategy and a bias-aware aggregation module to enable community-aware self-supervised learning. Evaluated on three fMRI datasets—ABIDE-I, ADHD-200, and ADNI—the method significantly outperforms state-of-the-art approaches, achieving notable improvements in both diagnostic accuracy and interpretability of functional reorganization.

Abstract

Brain networks exhibit a modular community structure that varies across individuals and neurological conditions. However, existing self-supervised learning (SSL) methods often overlook this heterogeneity, relying on generic masking strategies that fail to capture subject-specific functional organization. We propose BrainPICM, a self-supervised framework for brain network analysis via progressive individualized community aware masking. BrainPICM formulates ROI-to-community mapping as a progressive unbalanced optimal transport process, yielding soft assignments and per-ROI confidence scores. Guided by these confidence estimates, a curriculum-style masking strategy gradually incorporates low-confidence, potentially pathological regions into training, enabling the model to learn both stable modular structures and individual variations. Additionally, a deviation-aware aggregation module quantifies functional reorganization by measuring mass redistribution relative to a population template, enhancing interpretability and downstream prediction. Experiments on three fMRI datasets (ABIDE-I, ADHD-200, ADNI) show that BrainPICM consistently outperforms state-of-the-art supervised and SSL methods in diagnostic accuracy, indicating that explicitly injecting modular community structure into masked modeling yields more functionally consistent and generalizable representations. The source code for this approach will be released at https://github.com/Hrychen7/BrainPICM.

Problem

• brain networks
• modular community structure
• individual heterogeneity
• self-supervised learning
• functional organization

Innovation

• progressive self-supervised learning
• individualized community assignment
• optimal transport
• curriculum masking
• functional reorganization