Research map / Detection and segmentation / Open-set and few-shot detection
I-MedSAM: Implicit Medical Image Segmentation with Segment Anything
Xiaobao Wei, Jiajun Cao, Yizhu Jin, Ming Lu, Guangyu Wang, Shanghang Zhang
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 frequency adapter (amplitude spectrum) to LoRA + INR model (ablation) | I-MedSAM · Kvasir-Sessile | Dice (%): 88.83 -> 91.49 | Improved |
| use amplitude rather than phase spectrum in frequency adapter (ablation) | I-MedSAM · Kvasir-Sessile | Dice (%): 90.60 -> 91.49 | Improved |
| add INR decoder to LoRA + FA model (ablation) | I-MedSAM · Kvasir-Sessile | Dice (%): 88.74 -> 91.49 | Improved |
| add INR decoder to LoRA + FA model (ablation) | I-MedSAM · Kvasir-Sessile 384->128 | Dice (%): 75.69 -> 91.45 | Improved |
| add frequency adapter to LoRA + INR model (ablation) | I-MedSAM · Kvasir-Sessile -> CVC | Dice (%): 83.40 -> 88.83 | Improved |
| add frequency adapter to LoRA-only baseline (ablation) | I-MedSAM · Kvasir-Sessile -> CVC | Dice (%): 82.57 -> 82.61 | No clear effect |
| add uncertainty-guided sampling in coarse-to-fine INR (ablation) | I-MedSAM · Kvasir-Sessile | Dice (%): 87.77 -> 91.49 | Improved |
| sample more points (Top-50%) instead of 12.5% for UGS (ablation) | I-MedSAM · Kvasir-Sessile | Dice (%): 91.49 -> 90.27 | Hurt |
| sample fewer points (Top-3.125%) instead of 12.5% for UGS (ablation) | I-MedSAM · Kvasir-Sessile | Dice (%): 91.49 -> 90.48 | Hurt |
| use I-MedSAM instead of MedSAM | SAM-based segmentation · BCV (multi-organ) | Dice (%): 85.85 -> 89.91 | Improved |
| use implicit (INR) output instead of discrete output at changed resolution | I-MedSAM vs MedSAM · Kvasir-Sessile 384->896 | Dice (%): 83.19 -> 91.33 | Improved |
Later papers that compared with this method
1 later accepted paper reports I-MedSAM as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- From Minimal Clinical Prompts to 3D: Spacing-Aware Prompt Propagation for Multimodal Prostate Lesion Segmentation in bpMRI (ECCV 2026 · better on 2 of 2 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
- Tyche: Stochastic In-Context Learning for Medical Image Segmentation (CVPR 2024 · similarity 0.88)
- Enhancing SAM with Efficient Prompting and Preference Optimization for Semi-supervised Medical Image Segmentation (CVPR 2025 · similarity 0.86)
- Interactive Medical Image Segmentation: A Benchmark Dataset and Baseline (CVPR 2025 · similarity 0.86)
- Unleashing the Potential of SAM for Medical Adaptation via Hierarchical Decoding (CVPR 2024 · similarity 0.86)
- Multi-view Evidential Learning-based Medical Image Segmentation (AAAI 2025 · similarity 0.86)
- BLO-SAM: Bi-level Optimization Based Finetuning of the Segment Anything Model for Overfitting-Preventing Semantic Segmentation (ICML 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.
