Research map / Detection and segmentation / CNN features
Retriever Encoder Selection Matters for In-Context Learning-based Medical Segmentation
Fan Wang, Zhongyi Han, Yongshun Gong, Yilong Yin
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 |
|---|---|---|---|
| select RE per query with S3 (IRES) instead of GBC task-level selection | UniverSeg · FUNDUS-OD avg | DSC: 72.2 -> 79.4 | Improved |
| per-query S3 selection vs oracle per-dataset best encoder (ablation) | ICL-MedSAM · FUNDUS-OD avg | DSC: 88.4 -> 91.4 | Improved |
| per-query S3 selection vs best fixed encoder (ablation) | ICL-SAM · FUNDUS-OC avg | DSC: IRES 56.9 vs Best-S 53.3 on OC avg; UniverSeg 57.5 vs 59.1 | Hurt |
| GBC feature-based encoder selection vs random retrieval (ablation) | ICL-SAM · BraTS2020 avg | DSC: 17.5 -> 19.2 | No clear effect |
| add IRES encoder selection | ICL-SAM · BraTS2020 avg | DSC: 17.5 -> 32.4 | Improved |
| add IRES encoder selection (ablation) | ICL-MedSAM · BraTS2020 LGG | DSC: 33.1 -> 31.4 | Hurt |
| add parallel prediction to per-query S3 selection (ablation) | ICL-MedSAM · REFUGE | time (min): 15.1m -> 7.7m | Improved |
| add reciprocal neighbor reuse on top of parallel prediction (ablation) | ICL-MedSAM · REFUGE | time (min): 7.7m -> 5.7m | Improved |
| increase reciprocal neighbors m (ablation) | IRES · RIM-ONE-r3 | DSC: up to 5% drop | Hurt |
| increase reciprocal neighbors m (ablation) | IRES · REFUGE, Drishti-GS | DSC: within 1% | No clear effect |
| permute order of top-3 demonstrations (ablation) | IRES · RIM-ONE-r3 optic cup | DSC: <0.5% variation | No clear effect |
| add more ICL demonstrations (ablation) | ICL-MedSAM · RIM-ONE-r3 optic cup | DSC: diminishing returns | Improved |
Closest papers by meaning
- Enhancing SAM with Efficient Prompting and Preference Optimization for Semi-supervised Medical Image Segmentation (CVPR 2025 · similarity 0.86)
- Unified Medical Lesion Segmentation via Self-referring Indicator (CVPR 2025 · similarity 0.85)
- Tyche: Stochastic In-Context Learning for Medical Image Segmentation (CVPR 2024 · similarity 0.85)
- Unleashing the Potential of SAM for Medical Adaptation via Hierarchical Decoding (CVPR 2024 · similarity 0.85)
- Rethinking Decoder Design: Improving Biomarker Segmentation Using Depth-to-Space Restoration and Residual Linear Attention (CVPR 2025 · similarity 0.85)
- Training Like a Medical Resident: Context-Prior Learning Toward Universal Medical Image Segmentation (CVPR 2024 · similarity 0.85)
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.
