Research map / Vision-language and multimodal / Multimodal LLMs
WSI-LLaVA: A Multimodal Large Language Model for Whole Slide Image
Yuci Liang, Xinheng Lyu, Wenting Chen, Meidan Ding, Jipeng Zhang, Xiangjian He, Song Wu, Xiaohan Xing, Sen Yang, Xiyue Wang, Linlin Shen
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 WSI encoder (ablation) | WSI-LLaVA · WSI-Bench test | average score: 0.632 -> 0.680 | Improved |
| add WSI encoder (ablation) | WSI-LLaVA · WSI-Bench morphological analysis open | WSI-Relevance: 0.447 to 0.610 | Improved |
| add WSI encoder (ablation) | WSI-LLaVA · WSI-Bench treatment planning open | WSI-Relevance: 0.816 to 0.845 | Improved |
| add stage 1 WSI-text alignment on top of stages 2 and 3 (ablation) | WSI-LLaVA · WSI-Bench test | average score: 0.654 -> 0.680 | Improved |
| add stage 2 feature space alignment after stage 1 removed too (stage 3 only to full) (ablation) | WSI-LLaVA · WSI-Bench test | average score: 0.643 -> 0.654 | Improved |
| add all three training stages versus task tuning only (ablation) | WSI-LLaVA · WSI-Bench test | average score: 0.643 -> 0.680 | Improved |
| train/use WSI-LLaVA with full-WSI input instead of thumbnail MLLMs | WSI-LLaVA vs Quilt-LLaVA · WSI-Bench test | average score: 0.721 -> 0.754 | Improved |
| compare WSI-LLaVA against general-purpose GPT-4o | WSI-LLaVA · WSI-Bench test | average score: 0.507 -> 0.754 | Improved |
| WSI-LLaVA versus second-best on zero-shot histological grading | WSI-LLaVA · SlideBench-VQA BCNB | histological grading: 46.3% vs. second-best 30.0% | Improved |
| WSI-LLaVA versus second-best on zero-shot molecular subtyping | WSI-LLaVA · SlideBench-VQA BCNB | molecular subtyping: 29.2% vs. second-best 18.1% | Improved |
| WSI-LLaVA versus Quilt-LLaVA on molecular subtyping | WSI-LLaVA · WSI-Bench diagnosis | WSI-Precision: 0.630 vs 0.675 | Hurt |
| use WSI metrics rather than NLU metrics for report evaluation | Hist-Gen vs WSI-LLaVA · WSI-Bench report case | WSI-Precision: BLEU-1 0.531 vs. 0.521; WSI-Precision 0.233 vs 0.8 | Mixed |
Later papers that compared with this method
2 later accepted papers report WSI-LLaVA as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- MLLM-HWSI: A Multimodal Large Language Model for Hierarchical Whole Slide Image Understanding (CVPR 2026 · better on 7 of 7 shared results)
- Toward Interpretable Analysis of Whole-slide Pathology Images via Large Language Model-based Agentic Reasoning (ECCV 2026 · better on 4 of 4 shared results)
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.
- Quilt-LLaVA — Quilt-LLaVA: Visual Instruction Tuning by Extracting Localized Narratives from Open-Source Histopathology Videos (CVPR 2024) · better on 6 of 6 shared results
- SlideChat — SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding (CVPR 2025) · better on 1 of 1 shared result
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
- MLLM-HWSI: A Multimodal Large Language Model for Hierarchical Whole Slide Image Understanding (CVPR 2026 · similarity 0.92)
- PathVQ: Reforming Computational Pathology Foundation Model for Whole Slide Image Analysis via Vector Quantization (NeurIPS 2025 · similarity 0.90)
- SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding (CVPR 2025 · similarity 0.90)
- PathFLIP: Fine-grained Language-Image Pretraining for Versatile Computational Pathology (AAAI 2026 · similarity 0.89)
- PathView-Bench: Can Multimodal Large Language Models Achieve Fine-grained Multiscale Understanding of Pathology Images? (NeurIPS 2026 · similarity 0.89)
- CPath-Omni: A Unified Multimodal Foundation Model for Patch and Whole Slide Image Analysis in Computational Pathology (CVPR 2025 · similarity 0.88)
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.
