Test-Time Domain Generalization via Universe Learning: A Multi-Graph Matching Approach for Medical Image Segmentation

2025-03-17

AI Summary

In medical image segmentation, existing test-time adaptation (TTA) methods neglect morphological priors, limiting domain generalization performance. To address this, we propose a Multi-Graph Matching Universal Learning framework for Test-Time Domain Generalization (TTDG), enabling morphology-aware adaptive segmentation on unlabeled test data. Our key contributions are: (1) a learnable universal embedding representation that captures cross-domain structural commonalities; (2) a morphology-guided multi-graph matching mechanism that explicitly incorporates organ shape and topological priors; and (3) a cycle-consistency constraint to ensure stability during adaptation. Evaluated on two benchmark medical segmentation datasets, TTDG consistently outperforms state-of-the-art TTA and domain generalization methods under both single-source and multi-source settings. The framework is robust, label-free at test time, and publicly available.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Segmentation
• Application Domains: Transportation

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Despite domain generalization (DG) has significantly addressed the performance degradation of pre-trained models caused by domain shifts, it often falls short in real-world deployment. Test-time adaptation (TTA), which adjusts a learned model using unlabeled test data, presents a promising solution. However, most existing TTA methods struggle to deliver strong performance in medical image segmentation, primarily because they overlook the crucial prior knowledge inherent to medical images. To address this challenge, we incorporate morphological information and propose a framework based on multi-graph matching. Specifically, we introduce learnable universe embeddings that integrate morphological priors during multi-source training, along with novel unsupervised test-time paradigms for domain adaptation. This approach guarantees cycle-consistency in multi-matching while enabling the model to more effectively capture the invariant priors of unseen data, significantly mitigating the effects of domain shifts. Extensive experiments demonstrate that our method outperforms other state-of-the-art approaches on two medical image segmentation benchmarks for both multi-source and single-source domain generalization tasks. The source code is available at https://github.com/Yore0/TTDG-MGM.

Problem

• Addresses domain shift in medical image segmentation.
• Incorporates morphological information via multi-graph matching.
• Improves test-time adaptation using learnable universe embeddings.

Innovation

• Multi-graph matching for medical image segmentation
• Learnable universe embeddings with morphological priors
• Unsupervised test-time adaptation for domain generalization