Research map / Detection and segmentation / Open-set and few-shot detection
MediRound: Multi-Round Entity-Level Reasoning Segmentation in Medical Images
Qinyue Tong, Ziqian Lu, Jun Liu, Rui Zuo, Zheming Lu, Yueming Jin
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 Judgment & Correction Mechanism (JCM) | MediRound · MR-MedSeg val (all rounds) | Dice: 55.8 -> 58.4 | Improved |
| add JCM | MediRound · MR-MedSeg round 8 | Dice: 46.1 -> 54.8 | Improved |
| add JCM | MediRound · MR-MedSeg test hard | gIoU: 38.2 -> 40.5 | Improved |
| add JCM (vs READ-style variant) | MediRound · MR-MedSeg val (all rounds) | gIoU: 45.1 -> 49.0 | Improved |
| use MedSAM seg head instead of SAM (ablation) | MediRound · MR-MedSeg val (all rounds) | Dice: 46.7 -> 55.8 | Improved |
| use MedSAM seg head instead of SAM-Med2D (ablation) | MediRound · MR-MedSeg val (all rounds) | Dice: 53.9 -> 55.8 | Improved |
| use frozen MedSAM instead of LoRA-tuned MedSAM (ablation) | MediRound · MR-MedSeg val (all rounds) | Dice: 54.6 -> 55.8 | Improved |
| use LLaVA-Med instead of LLaVA-v1.5 (ablation) | MediRound · MR-MedSeg val (all rounds) | Dice: 53.4 -> 55.8 | Improved |
| LoRA tune LLaVA-Med instead of freezing it (ablation) | MediRound · MR-MedSeg val (all rounds) | Dice: 48.0 -> 55.8 | Improved |
| add cropped image of reference mask (ablation) | MediRound · MR-MedSeg val (all rounds) | Dice: 41.9 -> 55.4 | Improved |
| add bounding box on top of cropped image (ablation) | MediRound · MR-MedSeg val (all rounds) | Dice: 55.4 -> 55.8 | Improved |
| add bounding box on top of cropped image (ablation) | MediRound · MR-MedSeg val (all rounds) | cIoU: 56.0 -> 55.8 | No clear effect |
Closest papers by meaning
- Medical thinking with multiple images (ICLR 2026 · similarity 0.86)
- MedAtlas: Evaluating LLMs for Multi-Round, Multi-Task Medical Reasoning Across Diverse Imaging Modalities and Clinical Text (AAAI 2026 · similarity 0.86)
- Med-CMR: A Fine-Grained Benchmark Integrating Visual Evidence and Clinical Logic for Medical Complex Multimodal Reasoning (CVPR 2026 · similarity 0.86)
- DR$^2$Seg: Decomposed Two-Stage Rollouts for Efficient Reasoning Segmentation in Multimodal Large Language Models (ICML 2026 · similarity 0.85)
- MedXpertQA: Benchmarking Expert-Level Medical Reasoning and Understanding (ICML 2025 · similarity 0.85)
- M3CoTBench: Benchmark Chain-of-Thought of MLLMs in Medical Image Understanding (ICLR 2026 · 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.
