Research map / Detection and segmentation / Open-set and few-shot detection
PathChat-SegR1: Reasoning Segmentation in Pathology via SO-GRPO
Zelin Liu, Dongdong Chen, Yusong Sun, Yuqi Hu, Huang Jie, Sicheng Dong, Xu Han, Hongmei Yi, Qiyuan Bao, Lichi 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 RuiPath pathology encoder (ablation) | PathChat-SegR1 · PMBT | Dice: 0.42 -> 0.58 | Improved |
| add Seg-Adapter (ablation) | PathChat-SegR1 · PMBT | Dice: 0.52 -> 0.58 | Improved |
| add autonomous SEG token emission (ablation) | PathChat-SegR1 · PMBT | Dice: 0.51 -> 0.58 | Improved |
| add stain-invariant MedSAM encoder pre-training (ablation) | PathChat-SegR1 · PMBT | Dice: 0.51 -> 0.58 | Improved |
| add supervised fine-tuning (ablation) | PathChat-SegR1 · PMBT | Dice: 0.44 -> 0.58 | Improved |
| add reinforcement learning (SO-GRPO) (ablation) | PathChat-SegR1 · PMBT | Dice: 0.40 -> 0.58 | Improved |
| add length reward (ablation) | PathChat-SegR1 · PMBT | Dice: 0.56 -> 0.58 | Improved |
| add spatial reward (ablation) | PathChat-SegR1 · PMBT | Dice: 0.53 -> 0.58 | Improved |
| use SO-GRPO instead of standard GRPO (ablation) | PathChat-SegR1 · PMBT | Dice: 0.53 -> 0.58 | Improved |
| add standard GRPO over SFT only (ablation) | PathChat-SegR1 · PMBT | Dice: 0.40 -> 0.53 | Improved |
| add GAE credit assignment (ablation) | PathChat-SegR1 · PMBT | Dice: 0.55 -> 0.58 | Improved |
| add differentiable segmentation reward (ablation) | PathChat-SegR1 · PMBT | Dice: 0.56 -> 0.58 | Improved |
Closest papers by meaning
- Patho-R1: A Multimodal Reinforcement Learning-Based Pathology Expert Reasoner (AAAI 2026 · similarity 0.90)
- PathGen-1.6M: 1.6 Million Pathology Image-text Pairs Generation through Multi-agent Collaboration (ICLR 2025 · similarity 0.86)
- PathView-Bench: Can Multimodal Large Language Models Achieve Fine-grained Multiscale Understanding of Pathology Images? (NeurIPS 2026 · similarity 0.85)
- PathAsst: A Generative Foundation AI Assistant towards Artificial General Intelligence of Pathology (AAAI 2024 · similarity 0.85)
- PathVQ: Reforming Computational Pathology Foundation Model for Whole Slide Image Analysis via Vector Quantization (NeurIPS 2025 · similarity 0.85)
- Advancing Generalizable Tumor Segmentation with Anomaly-Aware Open-Vocabulary Attention Maps and Frozen Foundation Diffusion Models (CVPR 2025 · similarity 0.84)
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.
