When Repository Labels Are Not Image-Level Truth: A Supervision Auditing Framework for Chest Radiograph AI
当仓库标签并非图像级真值:面向胸部X光AI的监督审计框架
专题命中 仓库级理解 :repository(title,abstract)
AI总结 针对胸部X光仓库标签非图像级真值的问题,提出RSA框架,以MIMIC-CXR心脏肥大数据为例发现其标签与专家评估一致性极低,用专家精选队列训练的DenseNet121模型达0.853的ROC-AUC,强调监督审计的重要性。
Comments Accepted (oral) at the 3rd MICCAI Student Board (MSB) EMERGE Workshop, MICCAI 2026. This is the authors' version; the final authenticated version will appear in Springer Lecture Notes in Computer Science (LNCS). 10 pages, 3 figures