病理学基础模型辅助 Barrett 食管异型增生分级,AUC 达 0.96
Multi-institutional grading of Barrett's esophagus dysplasia without stain normalization using a pathology foundation model and multiple instance learning
病理医生和做医疗 AI 的团队都该看看这篇。Barrett 食管低级别异型增生历来判读差异很大,这个系统用基础模型加多示例学习,130 例测试集 AUC 到 0.96,而且不用染色归一化就能跨 4 家机构保持表现,省掉了以往系统最麻烦的站点级预处理。
一项多机构回顾性诊断准确性研究测试了基于病理学基础模型的深度学习系统,用于 Barrett 食管异型增生分级。研究使用来自 4 家机构的 969 张 H&E 染色全切片图像,模型先通过专家标注训练 patch 级分类器定位诊断相关区域,再用属性感知多示例学习在切片级别区分非异型增生(NDBE)、低级别和高级别异型增生。
在 130 例保留测试集上,宏平均 AUC 为 0.96(95% CI 0.93–0.98);二分类异型增生检测的敏感度为 98%(80/82),特异度为 81%(39/48),阴性似然比 0.030,与专家共识的二次加权 kappa 为 0.83。测试集中没有高级别异型增生病例被误判为 NDBE。
该系统在 4 家机构间表现一致且无需染色归一化,说明基础模型具有染色鲁棒性,可能免除以往系统依赖的机构级预处理,使跨机构部署成为可能。这对低级别异型增生判读差异大、影响临床决策的场景有直接参考价值。
Multi-institutional grading of Barrett's esophagus dysplasia without stain normalization using a pathology foundation model and multiple instance learning
Barrett's esophagus dysplasia grading suffers substantial inter-observer variability, particularly for low-grade dysplasia, complicating clinical decision-making. We conducted a retrospective multi-institutional diagnostic accuracy study of a foundation model-based deep learning system for dysplasia grading, using 969 H\&E-stained whole slide images from four institutions. Patch-level classifiers trained on expert annotations identified diagnostically relevant regions, followed by attribute-aware multiple instance learning for slide-level classification as non-dysplastic Barrett's esophagus (NDBE), low-grade dysplasia, or high-grade dysplasia. On a held-out test set (n = 130), macro-averaged AUC was 0.96 (95\% CI 0.93--0.98). For binary dysplasia detection, sensitivity was 98\% (80/82) and specificity 81\% (39/48), with a negative likelihood ratio of 0.030 and quadratic-weighted \kap{} of 0.83 against expert consensus. No high-grade dysplasia case was misclassified as NDBE. Performance was consistent across institutions without stain normalization, indicating that stain-robust foundation-model representations may remove the site-specific preprocessing that has constrained prior systems to institution-specific deployment.
This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows.
Source attribution: curated from arXiv cs.AI (@Amirali Khosravi, Jaidip M. Jagtap, D. Chamil Codipilly, Catherine E. Hagen, Jason T. Lewis, Rickey E. Carter, Shahriar Faghani, Vani J. A. Konda, Fouad Otaki, Shajan Peter, Nicholas Shaheen, Sachin Wani, Prasad G. Iyer, Bradley J. Erickson).