---
title: "PathAgentBench：评估全切片病理图像中证据寻求型视觉语言模型的基准"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.19261v1"
permalink: "https://feed.astralor.com/items/itm_fc4e40d6531ca5e07f4d9264"
published: "2026-07-21T16:33:44.000Z"
score: 74
channel: "finance"
tags: ["PathAgentBench","全切片病理图像","多尺度推理","病理视觉语言模型","证据获取","诊断区域定位"]
---

# PathAgentBench：评估全切片病理图像中证据寻求型视觉语言模型的基准

## AI 点评

> 这篇论文最值得看的地方在于它没有重复“模型在裁剪好的图块上表现多好”的老故事，而是直面病理AI最核心的痛点：模型到底能不能像医生一样在整张切片中找到病变？答案让人清醒——定位和探索能力几乎为零。这对所有从事医学影像AI的研究者都是一个重要提醒：别光盯着已经切好的数据沾沾自喜，真正的挑战在真实的“寻找”过程中。对开发病理辅助诊断产品的团队来说，这个基准是评估自家模型能力的必备工具。

## 速览

PathAgentBench 是一个专门评估视觉语言模型在全切片病理图像（WSI）中证据获取能力的基准。它覆盖四项互补能力：跨模态匹配、诊断区域定位、多尺度推理和自主探索。基准包含 1,822 张 TCGA 切片和 17,135 条诊断路径，由十名病理学家标注。评估 20 个模型后发现，模型在多尺度推理上表现良好（>93% 准确率），但在诊断区域定位和自主探索中极差（定位交并比 <0.09，高倍探索命中率仅 0.020），揭示了推理与证据获取间的显著差距。

## 关键点

- PathAgentBench 构建了诊断树结构，将嵌套区域与不同倍率的病理发现关联，形成统一的评估框架。
- 现有最强模型在诊断区域定位任务上仍远逊于简单基线，交并比不足 0.09。
- 自主探索时高倍放大下的无条件命中率仅 0.020，模型难以从千兆像素图像中直接获取证据。
- 该基准公开了大规模标注数据和标准化评估协议，推动病理 AI 从“证据推理”走向“证据获取”。

## 全文

Whole-slide image (WSI) diagnosis requires identifying diagnostically relevant regions, examining them across magnifications, and integrating multi-scale evidence. However, most existing pathology benchmarks evaluate models on pre-cropped patches or pre-extracted slide features, leaving their ability to acquire evidence directly from gigapixel WSIs largely untested. We introduce PathAgentBench, a benchmark for evaluating evidence-seeking vision-language models (VLMs) across four complementary capabilities: image-to-text matching for evidence interpretation, text-to-image retrieval for evidence verification, diagnostic-region localization for evidence acquisition, and multi-scale reasoning for evidence integration. The benchmark is organized as a diagnostic tree that links nested regions across magnifications with scale-specific findings and path-level diagnoses. It contains 1,822 TCGA WSIs and 17,135 diagnostic paths annotated by ten board-certified pathologists. An additional private cohort of 190 breast cancer WSIs with detailed annotations is used to evaluate autonomous whole-slide exploration. We evaluate 20 general-purpose, medical, and pathology-specialized models. Leading open-weight models achieve over 93% accuracy in multi-scale reasoning and over 50% accuracy in both cross-modal matching tasks. In contrast, diagnostic-region localization remains challenging: the best text-guided mean intersection-over-union is below 0.09, underperforming a simple center-based heuristic. During autonomous exploration, the unconditional hit rate decreases from 0.522 at low magnification to 0.185 at intermediate magnification and 0.020 at high magnification. These results reveal a pronounced gap between reasoning over curated evidence and acquiring that evidence directly from WSIs. PathAgentBench provides a unified framework for measuring and improving evidence-seeking pathology models.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_fc4e40d6531ca5e07f4d9264