Research map / Label-efficient and robust learning / Test-time adaptation
Single-Domain Generalization in Medical Image Segmentation via Test-Time Adaptation from Shape Dictionary
Quande Liu, Cheng Chen, Qi Dou, Pheng-Ann Heng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Integrate a learned shape dictionary and dual-consistency test-time adaptation | Single-domain prostate MRI segmentation · Prostate MRI, unseen sites B–F average | Dice: 78.2 -> 81.1 | Improved |
| Integrate a learned shape dictionary and dual-consistency test-time adaptation | Single-domain prostate MRI segmentation · Prostate MRI, unseen sites B–F average | Hausdorff Distance: 41.0 -> 32.3 | Improved |
| Integrate a learned shape dictionary and dual-consistency test-time adaptation | Single-domain fundus segmentation · Fundus segmentation, unseen sites B–D average | Dice: 84.6 -> 85.8 | Improved |
| Integrate a learned shape dictionary and dual-consistency test-time adaptation | Single-domain fundus segmentation · Fundus segmentation, unseen sites B–D average | Hausdorff Distance: 22.3 -> 20.5 | Improved |
| Add integration of the shape dictionary to the baseline (ablation) | Single-domain prostate MRI segmentation · Prostate MRI, unseen sites B–F average | Dice: 74.8 -> 78.4 | Improved |
| Add integration of the shape dictionary to the baseline (ablation) | Single-domain fundus segmentation · Fundus segmentation, unseen sites B–D average | Dice: 82.7 -> 84.4 | Improved |
| Add test-time adaptation to the model with shape dictionary integration (ablation) | Baseline + ISD · Prostate MRI, unseen sites B–F average | Dice: +2.7% | Improved |
| Add test-time adaptation to the model with shape dictionary integration (ablation) | Baseline + ISD · Fundus segmentation, unseen sites B–D average | Dice: +1.4% | Improved |
| Use middle-sized dictionary (K=32 or 48) instead of K=16 or K=64 (ablation) | TASD · Prostate MRI and fundus segmentation | Dice | Improved |
| Use fewer ground-truth masks to learn the shape dictionary (ablation) | TASD · Prostate MRI segmentation | Dice: 0.7% and 1.6% | Hurt |
| Use fewer ground-truth masks to learn the shape dictionary (ablation) | TASD · Fundus segmentation | Dice: 0.4% and 0.9% | Hurt |
| Add consistency regularization on shape coefficients (ablation) | TASD test-time adaptation · Medical image segmentation | Segmentation performance | Improved |
Closest papers by meaning
- Why Invariance is Not Enough for Biomedical Domain Generalization and How to Fix It (NeurIPS 2026 · similarity 0.89)
- Learning Generalized Medical Image Segmentation from Decoupled Feature Queries (AAAI 2024 · similarity 0.89)
- Domain Generalized Medical Landmark Detection via Robust Boundary-Aware Pre-Training (AAAI 2025 · similarity 0.88)
- LangDAug: Langevin Data Augmentation for Multi-Source Domain Generalization in Medical Image Segmentation (ICML 2025 · similarity 0.88)
- S2S2: Semantic Stacking for Robust Semantic Segmentation in Medical Imaging (AAAI 2025 · similarity 0.87)
- Open-Set Domain Adaptation for Semantic Segmentation (CVPR 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
