Research map / Vision-language and multimodal / Visual question answering
Act Like a Pathologist: Tissue-Aware Whole Slide Image Reasoning
Wentao Huang, Weimin Lyu, Peiliang Lou, Qingqiao Hu, Xiaoling Hu, Shahira Abousamra, Wenchao Han, Ruifeng Guo, Jiawei Zhou, Chao Chen, Chen Wang
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 patches with learned question-guided selection instead of random sampling (ablation) | HistoSelect · WSI-Bench Close Diagnosis | accuracy: 78.02 -> 85.79 | Improved |
| select question-relevant patches instead of diversity-based (DivPrune) (ablation) | HistoSelect · WSI-Bench Close Diagnosis | accuracy: 80.99 -> 85.79 | Improved |
| use learned selectors with IB regularization instead of non-learnable similarity (ablation) | HistoSelect · WSI-Bench Close Diagnosis | accuracy: 81.98 -> 85.79 | Improved |
| add group sampler (coarse-to-fine) (ablation) | HistoSelect · WSI-Bench Close Diagnosis | accuracy: 81.82 -> 85.79 | Improved |
| add patch selector (ablation) | HistoSelect · WSI-Bench Close Diagnosis | accuracy: 81.32 -> 85.79 | Improved |
| increase token budget from 1k to 5k (ablation) | HistoSelect · WSI-Bench Close Diagnosis | accuracy: 82.15 -> 85.79 | Improved |
| increase token budget from 5k to 10k (ablation) | HistoSelect · WSI-Bench Close Diagnosis | accuracy: 85.79 -> 85.12 | No clear effect |
| select 30% of patches (5k tokens) instead of all | HistoSelect · WSI-Bench Close | token computation: 70% reduction | Improved |
| use HistoSelect instead of SlideChat | HistoSelect · Close-ended VQA average | accuracy: 80.88 -> 83.80 | Improved |
| use HistoSelect instead of Quilt-LLaVA on treatment questions (open-ended) | HistoSelect · WSI-Bench VQA Treatment | WSI-R: 0.807 -> 0.801 | Hurt |
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.
- SlideChat — SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding (CVPR 2025) · better on 11 of 11 shared results
- Quilt-LLaVA — Quilt-LLaVA: Visual Instruction Tuning by Extracting Localized Narratives from Open-Source Histopathology Videos (CVPR 2024) · better on 4 of 6 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
- SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding (CVPR 2025 · similarity 0.90)
- CPath-Omni: A Unified Multimodal Foundation Model for Patch and Whole Slide Image Analysis in Computational Pathology (CVPR 2025 · similarity 0.90)
- MLLM-HWSI: A Multimodal Large Language Model for Hierarchical Whole Slide Image Understanding (CVPR 2026 · similarity 0.90)
- PathVQ: Reforming Computational Pathology Foundation Model for Whole Slide Image Analysis via Vector Quantization (NeurIPS 2025 · similarity 0.89)
- PathView-Bench: Can Multimodal Large Language Models Achieve Fine-grained Multiscale Understanding of Pathology Images? (NeurIPS 2026 · similarity 0.89)
- Patho-R1: A Multimodal Reinforcement Learning-Based Pathology Expert Reasoner (AAAI 2026 · similarity 0.89)
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.
