Research map / Multi-agent systems and games / LLM agents
PathFinder: A Multi-Modal Multi-Agent System for Medical Diagnostic Decision-Making Applied to Histopathology
Fatemeh Ghezloo, Mehmet Saygin Seyfioglu, Rustin Soraki, Wisdom O. Ikezogwo, Beibin Li, Tejoram Vivekanandan, Joann G. Elmore, Ranjay Krishna, Linda Shapiro
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 T5 text conditioning to the navigator (ablation) | PathFinder navigator · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.64 -> 0.74 | Improved |
| use T5 instead of CLIP text encoder (ablation) | text-conditioned navigator · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.62 -> 0.74 | Improved |
| use iterative navigation instead of ABMIL top-10 patches (ablation) | PathFinder · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.54 -> 0.74 | Improved |
| use selective navigation instead of exhaustive search (ablation) | PathFinder · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.68 -> 0.74 | Improved |
| sample patches by navigator instead of imitated pathologist sampling (ablation) | PathFinder · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.63 -> 0.74 | Improved |
| use fine-tuned Quilt-LLaVA description agent instead of LLaVA-Med (ablation) | PathFinder · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.56 -> 0.74 | Improved |
| swap CLIP for T5 encoder when descriptions are from LLaVA-Med (ablation) | text-conditioned navigator · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.60 -> 0.56 | Hurt |
| add Triage Agent (ablation) | PathFinder · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.58 -> 0.74 | Improved |
| use trained GPT-2 Diagnosis Agent instead of prompting GPT-4o (ablation) | PathFinder · M-Path skin biopsy WSI diagnosis (majority voting) | accuracy: 0.49 -> 0.74 | Improved |
| use Triage Agent transformer instead of TransMIL | triage classifier · M-Path Class 1 vs non-Class 1 triage | overall accuracy: 0.83 -> 0.91 | Improved |
Closest papers by meaning
- CPathAgent: An Agent-based Foundation Model for Interpretable High-Resolution Pathology Image Analysis Mimicking Pathologists' Diagnostic Logic (NeurIPS 2025 · similarity 0.91)
- Patho-R1: A Multimodal Reinforcement Learning-Based Pathology Expert Reasoner (AAAI 2026 · similarity 0.87)
- Patho-AgenticRAG: Towards Multimodal Agentic Retrieval-Augmented Generation for Pathology VLMs via Reinforcement Learning (AAAI 2026 · similarity 0.87)
- PathAsst: A Generative Foundation AI Assistant towards Artificial General Intelligence of Pathology (AAAI 2024 · similarity 0.86)
- PathView-Bench: Can Multimodal Large Language Models Achieve Fine-grained Multiscale Understanding of Pathology Images? (NeurIPS 2026 · similarity 0.86)
- PathVQ: Reforming Computational Pathology Foundation Model for Whole Slide Image Analysis via Vector Quantization (NeurIPS 2025 · 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.
