Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SHAPE: Structure-aware Hierarchical Unsupervised Domain Adaptation with Plausibility Evaluation for Medical Image Segmentation

Linkuan Zhou, Yinghao Xia, Yufei Shen, Xiangyu Li, Wenjie Du, Cong Cong, Leyi Wei, Ran Su, Qiangguo Jin

Abstract

Unsupervised Domain Adaptation (UDA) is essential for deploying medical segmentation models across diverse clinical environments. Existing methods are fundamentally limited, suffering from semantically unaware feature alignment that results in poor distributional fidelity and from pseudo-label validation that disregards global anatomical constraints, thus failing to prevent the formation of globally implausible structures. To address these issues, we propose SHAPE (Structure-aware Hierarchical Unsupervised Domain Adaptation with Plausibility Evaluation), a framework that reframes adaptation towards global anatomical plausibility. Built on a DINOv3 foundation, its Hierarchical Feature Modulation (HFM) module first generates features with both high fidelity and class-awareness. This shifts the core challenge to robustly validating pseudo-labels. To augment conventional pixel-level validation, we introduce Hypergraph Plausibility Estimation (HPE), which leverages hypergraphs to assess the global anatomical plausibility that standard graphs cannot capture. This is complemented by Structural Anomaly Pruning (SAP) to purge remaining artifacts via cross-view stability. SHAPE significantly outperforms prior methods on cardiac and abdominal cross-modality benchmarks, achieving state-of-the-art average Dice scores of 90.08% (MRI->CT) and 78.51% (CT->MRI) on cardiac data, and 87.48% (MRI->CT) and 86.89% (CT->MRI) on abdominal data. The code is available at https://github.com/BioMedIA-repo/SHAPE.

SHAPE: Structure-aware Hierarchical Unsupervised Domain Adaptation with Plausibility Evaluation for Medical Image Segmentation

Abstract

Unsupervised Domain Adaptation (UDA) is essential for deploying medical segmentation models across diverse clinical environments. Existing methods are fundamentally limited, suffering from semantically unaware feature alignment that results in poor distributional fidelity and from pseudo-label validation that disregards global anatomical constraints, thus failing to prevent the formation of globally implausible structures. To address these issues, we propose SHAPE (Structure-aware Hierarchical Unsupervised Domain Adaptation with Plausibility Evaluation), a framework that reframes adaptation towards global anatomical plausibility. Built on a DINOv3 foundation, its Hierarchical Feature Modulation (HFM) module first generates features with both high fidelity and class-awareness. This shifts the core challenge to robustly validating pseudo-labels. To augment conventional pixel-level validation, we introduce Hypergraph Plausibility Estimation (HPE), which leverages hypergraphs to assess the global anatomical plausibility that standard graphs cannot capture. This is complemented by Structural Anomaly Pruning (SAP) to purge remaining artifacts via cross-view stability. SHAPE significantly outperforms prior methods on cardiac and abdominal cross-modality benchmarks, achieving state-of-the-art average Dice scores of 90.08% (MRI->CT) and 78.51% (CT->MRI) on cardiac data, and 87.48% (MRI->CT) and 86.89% (CT->MRI) on abdominal data. The code is available at https://github.com/BioMedIA-repo/SHAPE.
Paper Structure (15 sections, 12 equations, 5 figures, 3 tables)

This paper contains 15 sections, 12 equations, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Self-Supervised Foundation Models
  4. Feature Alignment
  5. Structural Plausibility
  6. Method
  7. Hierarchical Feature Modulation
  8. Hypergraph Plausibility Estimation
  9. Structural Anomaly Pruning
  10. Overall Learning Objective
  11. Experiments and Results
  12. Experimental details
  13. Comparison with state-of-the-art methods
  14. Ablation study
  15. Conclusions

Figures (5)

  • Figure 1: The pipeline of SHAPE. (a) Hierarchical Feature Modulation (HFM). (b) Hypergraph Plausibility Estimation (HPE). (c) Structural Anomaly Pruning (SAP).
  • Figure 2: Qualitative results of SHAPE and typical methods. Yellow arrows indicate areas where SHAPE outperforms competing methods.
  • Figure 3: (a) Feature distribution before adaptation. (b) Feature distribution after adaptation by AdaIN. (c) Feature distribution after adaptation by HFM.
  • Figure 4: Comparison of the effectiveness of different feature modulation methods. Take the class inside the red dotted line as an example.
  • Figure 5: Hyperparametric sensitivity analysis of all adaptation segmentation tasks. The vertical axis is the mean Dice scores (%) of all classes.