Research map / Label-efficient and robust learning / Test-time adaptation
Test-time Adaptation for Foundation Medical Segmentation Model Without Parametric Updates
Kecheng Chen, Xinyu Luo, Tiexin Qin, Jie Liu, Hui Liu, Victor Ho Fun Lee, Hong Yan, Haoliang Li
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 |
|---|---|---|---|
| add DAL-CRF loss to latent refinement (ablation) | MedSAM latent refinement · Fundus OD segmentation, loose prompts | Dice: 85.95 -> 89.88 | Improved |
| add entropy minimization loss alone (ablation) | MedSAM direct inference · Fundus OD segmentation, loose prompts | Dice: 85.95 -> 86.01 | No clear effect |
| add entropy loss on top of DAL-CRF (ablation) | MedSAM latent refinement · Fundus OD segmentation, loose prompts | Dice: 89.88 -> 90.14 | Improved |
| use DenseCRF post-processing (ablation) | MedSAM direct inference · Fundus OD segmentation, loose prompts | Dice: 85.95 -> 79.18 | Hurt |
| swap DAL-CRF for InTEnt loss in latent framework (ablation) | latent refinement · Fundus OD segmentation, loose prompts | Dice: 89.88 -> 84.78 | Hurt |
| swap DAL-CRF for TENT loss in latent framework (ablation) | latent refinement · Fundus OD segmentation, loose prompts | Dice: 89.88 -> 86.28 | Hurt |
| swap DAL-CRF for CRF-SOD loss in latent framework (ablation) | latent refinement · Fundus OD segmentation, loose prompts | Dice: 89.88 -> 87.71 | Hurt |
| increase K from 2 to 4 (ablation) | latent refinement · Fundus OD segmentation, loose prompts | Dice: 89.75 -> 90.14 | Improved |
| increase K from 4 to 6 (ablation) | latent refinement · Fundus OD segmentation, loose prompts | Dice: 90.14 -> 90.37 | Improved |
| increase K from 6 to 8 (ablation) | latent refinement · Fundus OD segmentation, loose prompts | Dice: 90.37 -> 90.03 | Hurt |
| latent refinement instead of normalization update (TENT) | MedSAM · Fundus OC segmentation, loose prompts | Dice: 85.89 -> 89.02 | Improved |
| latent refinement instead of best baseline | MedSAM · Fundus OC segmentation, shrinkage prompts | Dice: 75.19 -> 77.77 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PASS — PASS: Part-Aware Self-Supervised Pre-training for Person Re-identification (ECCV 2022) · better on 3 of 3 shared results
- TENT — Tent: Fully Test-Time Adaptation by Entropy Minimization (ICLR 2021) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Memory-Supported Synergistic Adaptation for Training-Free Test-Time Medical Image Segmentation (ECCV 2026 · similarity 0.91)
- Each Test Image Deserves A Specific Prompt: Continual Test-Time Adaptation for 2D Medical Image Segmentation (CVPR 2024 · similarity 0.90)
- Gradient Alignment Improves Test-Time Adaptation for Medical Image Segmentation (AAAI 2025 · similarity 0.90)
- Steady Progress Beats Stagnation: Mutual Aid of Foundation and Conventional Models in Mixed Domain Semi-Supervised Medical Image Segmentation (CVPR 2025 · similarity 0.88)
- Efficient Deformable Convolutional Prompt for Continual Test-Time Adaptation in Medical Image Segmentation (AAAI 2025 · similarity 0.87)
- Tyche: Stochastic In-Context Learning for Medical Image Segmentation (CVPR 2024 · similarity 0.86)
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.
