Related Experiment Video
Updated: Oct 7, 2026
Predictive Immune Modeling of Solid Tumors
Published on: February 25, 2020
Foundation Model-Based Computational Pathology Predicts Breast Cancer Biomarker Status from Intraoperative Frozen
Shun Liu1, Jun Li2, Yixiang Lian3
1Department of Ultrasound, Department of Medical Imaging, the Affiliated Changsha Central Hospital, Hengyang Medical School, University of South China, Changsha, China; Key Laboratory of Medical Imaging Precision Theranostics and Radiation Protection, College of Hunan Province, The Affiliated Changsha Central Hospital, Hengyang Medical School, University of South China, Changsha, China; Institute of Medical Imaging, Hengyang Medical School, University of South China, Hengyang, China.
Abstract:
Prediction of breast cancer biomarkers from routine formalin-fixed, paraffin-embedded (FFPE) hematoxylin and eosin (H&E) images has been explored previously, but whether biomarker-associated morphologic signals can be reliably captured from artifact-prone intraoperative frozen-section (FS) specimens remains insufficiently studied. This retrospective proof-of-concept study developed a foundation model-based multiple-instance learning (MIL) framework to predict estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2), and Ki-67 status directly from H&E-stained intraoperative FS whole-slide images (WSIs) of 248 breast cancer patients, using postoperative immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) as ground truth. Exploratory feature evaluation identified UNI2-h as a suitable encoder for downstream modeling. In five-fold cross-validation, areas under the receiver operating characteristic (ROC) curve were 0.847 for ER, 0.906 for PR, 0.893 for HER2, and 0.799 for Ki-67, with moderate agreement for ER, PR, and HER2 and fair agreement for Ki-67. An exploratory within-cohort cross-preparation analysis showed limited FFPE-to-FS transferability, whereas the FS-trained model retained relatively favorable performance on FFPE images. Attention heatmaps highlighted tumor regions with biologically plausible biomarker-associated morphologic patterns. These findings provide proof-of-concept that intraoperative FS H&E images retain learnable biomarker-associated morphologic signals despite preparation-related alterations.
