Saved in:
Bibliographic Details
Main Authors: Lv, Xingguo, Dong, Xingbo, Wang, Liwen, Yang, Jiewen, Zhao, Lei, Pu, Bin, Jin, Zhe, Li, Xuejun
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.13012
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910878665801728
author Lv, Xingguo
Dong, Xingbo
Wang, Liwen
Yang, Jiewen
Zhao, Lei
Pu, Bin
Jin, Zhe
Li, Xuejun
author_facet Lv, Xingguo
Dong, Xingbo
Wang, Liwen
Yang, Jiewen
Zhao, Lei
Pu, Bin
Jin, Zhe
Li, Xuejun
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2503_13012
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Test-Time Domain Generalization via Universe Learning: A Multi-Graph Matching Approach for Medical Image Segmentation
Lv, Xingguo
Dong, Xingbo
Wang, Liwen
Yang, Jiewen
Zhao, Lei
Pu, Bin
Jin, Zhe
Li, Xuejun
Computer Vision and Pattern Recognition
Artificial Intelligence
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.
title Test-Time Domain Generalization via Universe Learning: A Multi-Graph Matching Approach for Medical Image Segmentation
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2503.13012