Dual-supervised Asymmetric Co-training for Semi-supervised Medical Domain Generalization

2025-09-25

AI Summary

This paper addresses the more realistic cross-domain semi-supervised domain generalization (CD-SSDG) problem in medical image segmentation: during training, labeled and unlabeled data originate from distinct source domains, while the test domain remains entirely unknown—exacerbating both pseudo-label noise and domain shift. To tackle this, we propose a dual-supervised asymmetric co-training framework that jointly optimizes feature-level supervised signals with two divergent self-supervised tasks across parallel branches. This design explicitly mitigates pseudo-label errors and strengthens learning of domain-invariant features. The method requires no prior knowledge of the target domain and seamlessly integrates standard semi-supervised learning and domain generalization paradigms. Extensive experiments on multi-center medical datasets—including fundus, polyp, and spinal cord gray matter (SCGM)—demonstrate substantial improvements in cross-domain segmentation robustness and generalization performance. Our approach offers a promising pathway for clinical deployment under low-labeling-cost constraints.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Computer Vision: Segmentation
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Web applications in cross-disciplinary domains and verticals such as mixed reality, smart cities, and digital health
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Semi-supervised domain generalization (SSDG) in medical image segmentation offers a promising solution for generalizing to unseen domains during testing, addressing domain shift challenges and minimizing annotation costs. However, conventional SSDG methods assume labeled and unlabeled data are available for each source domain in the training set, a condition that is not always met in practice. The coexistence of limited annotation and domain shift in the training set is a prevalent issue. Thus, this paper explores a more practical and challenging scenario, cross-domain semi-supervised domain generalization (CD-SSDG), where domain shifts occur between labeled and unlabeled training data, in addition to shifts between training and testing sets. Existing SSDG methods exhibit sub-optimal performance under such domain shifts because of inaccurate pseudolabels. To address this issue, we propose a novel dual-supervised asymmetric co-training (DAC) framework tailored for CD-SSDG. Building upon the co-training paradigm with two sub-models offering cross pseudo supervision, our DAC framework integrates extra feature-level supervision and asymmetric auxiliary tasks for each sub-model. This feature-level supervision serves to address inaccurate pseudo supervision caused by domain shifts between labeled and unlabeled data, utilizing complementary supervision from the rich feature space. Additionally, two distinct auxiliary self-supervised tasks are integrated into each sub-model to enhance domain-invariant discriminative feature learning and prevent model collapse. Extensive experiments on real-world medical image segmentation datasets, i.e., Fundus, Polyp, and SCGM, demonstrate the robust generalizability of the proposed DAC framework.

Problem

• Addressing domain shifts between labeled and unlabeled training data
• Improving semi-supervised medical image segmentation generalization
• Solving inaccurate pseudolabels in cross-domain generalization scenarios

Innovation

• Dual-supervised asymmetric co-training framework for cross-domain generalization
• Feature-level supervision addresses inaccurate pseudo-labels from domain shifts
• Asymmetric auxiliary self-supervised tasks enhance domain-invariant feature learning