Towards reliable morphological subtyping and cancer detection on Barrett's esophagus pathology through ensembling foundation models.
16 affiliations
- 1Institute of Pathology, School of Medicine and Health, Technical University of Munich, Munich, Germany. [email protected].
- 2Department of Medical Informatics, School of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran. [email protected].
- 3Munich Center for Machine Learning, Munich, Germany. [email protected].
- 4TUM School of Computation, Information and Technology, Technical University of Munich, Munich, Germany.
- 5AI for Image-Guided Diagnosis and Therapy, Technical University of Munich, Munich, Germany.
- 6Institute of Pathology, School of Medicine and Health, Technical University of Munich, Munich, Germany.
- 7Munich Center for Machine Learning, Munich, Germany.
- 8Computer Aided Medical Procedures, Technical University of Munich, Munich, Germany.
- 9Klinikum Rechts Der Isar, Medical Clinic and Polyclinic II, Technical University of Munich, Munich, Germany.
- 10Department of Medical Informatics, School of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran.
- 11Pharmaceutical Sciences Research Center, Institute of Pharmaceutical Technology, Mashhad University of Medical Sciences, Mashhad, Iran.
- 12Department of Medical Informatics, University of Amsterdam, Amsterdam, the Netherlands.
- 13Department of Medicine II, University Medical Center Freiburg, Freiburg, Germany.
- 14Department of Nephrology, School of Medicine, Technical University of Munich, Munich, Germany.
- 15Pathologie Starnberg MVZ GmbH, Starnberg, Bayern, Germany.
- 16Munich Data Science Institute, Munich, Germany.
Abstract
The reliability of emergent histopathology foundation models (Histo-FMs) for real-world diagnostic tasks remains unproven. Given the complexity of diagnostic tasks and the challenges inherent in real-world data, we utilized Histo-FMs to investigate their utility for morphological subtyping of Barrett's esophagus (BE) and detecting esophageal adenocarcinoma (EAC) across the BE progression spectrum. We benchmarked Histo-FMs on the BarrettNet cohort, spanning diagnostic categories ranging from normal to EAC (N2EAC). The dataset comprised 3528 hematoxylin and eosin (H&E)-stained whole-slide images (WSIs) from 790 patients. A multi-magnification, multi-backbone aggregation of the five most expert-consistent single-scale models yielded incremental performance gains (AUROC of 0.907, F1-score of 0.696, accuracy of 0.795, and κ of 0.651 for morphological subtyping; AUROC of 0.909, F1-score of 0.836, accuracy of 0.959, and κ of 0.673 for EAC detection), with non-uniform gains across single-model comparisons and limited benefit for dysplasia. The findings support the capability of Histo-FMs for BE morphological subtyping and EAC detection in a multicenter clinical cohort, while highlighting persistent challenges in dysplasia assessment and the need for external validation. This study provides a foundation for future clinical validation.
Related
- Detection of high-grade dysplasia and adenocarcinoma in Barrett's esophagus using high-resolution virtual chromoendoscopy versus the Seattle protocol: the CONVERSE study
- Pathology foundation-model embeddings reveal a morphological landscape of breast lesions beyond diagnostic categories
- Establishment and performance evaluation of quadruplex RT-qPCR method for universal avian influenza virus detection and H5/H7/H9 subtyping
- Noninvasive Detection and Subtyping of Esophageal Cancer by Circulating Cell-Free mtDNA Fragmentomics: A Multi-Center Validation Study.
- Preoperative 3D morphological phenotyping of hepatocellular carcinoma via a foundation model: a non-invasive biomarker for recurrence risk stratification.
BibTeX
@article{kazemi2026towards,
title = {{Towards reliable morphological subtyping and cancer detection on Barrett's esophagus pathology through ensembling foundation models.}},
author = {Azar Kazemi and Camillo Saueressig and Jingsong Liu and Han Li and Julius Shakhtour and Bianca Grosser and Fabian Stögbauer and Julia Horstmann and Nassir Navab and Saeid Eslami and Michael Quante and Julia Slotta-Huspenina and Peter J Schüffler},
journal = {NPJ digital medicine},
year = {2026},
month = oct,
doi = {10.1038/s41746-026-03329-3},
url = {https://europepmc.org/article/MED/42919834},
note = {CC BY},
}
RIS
TY - JOUR
TI - Towards reliable morphological subtyping and cancer detection on Barrett's esophagus pathology through ensembling foundation models.
AU - Azar Kazemi
AU - Camillo Saueressig
AU - Jingsong Liu
AU - Han Li
AU - Julius Shakhtour
AU - Bianca Grosser
AU - Fabian Stögbauer
AU - Julia Horstmann
AU - Nassir Navab
AU - Saeid Eslami
AU - Michael Quante
AU - Julia Slotta-Huspenina
AU - Peter J Schüffler
PY - 2026
DA - 2026/10/21
JO - NPJ digital medicine
DO - 10.1038/s41746-026-03329-3
UR - https://europepmc.org/article/MED/42919834
AB - The reliability of emergent histopathology foundation models (Histo-FMs) for real-world diagnostic tasks remains unproven. Given the complexity of diagnostic tasks and the challenges inherent in real-world data, we utilized Histo-FMs to investigate their utility for morphological subtyping of Barrett's esophagus (BE) and detecting esophageal adenocarcinoma (EAC) across the BE progression spectrum. We benchmarked Histo-FMs on the BarrettNet cohort, spanning diagnostic categories ranging from normal to EAC (N2EAC). The dataset comprised 3528 hematoxylin and eosin (H&E)-stained whole-slide images (WSIs) from 790 patients. A multi-magnification, multi-backbone aggregation of the five most expert-consistent single-scale models yielded incremental performance gains (AUROC of 0.907, F1-score of 0.696, accuracy of 0.795, and κ of 0.651 for morphological subtyping; AUROC of 0.909, F1-score of 0.836, accuracy of 0.959, and κ of 0.673 for EAC detection), with non-uniform gains across single-model comparisons and limited benefit for dysplasia. The findings support the capability of Histo-FMs for BE morphological subtyping and EAC detection in a multicenter clinical cohort, while highlighting persistent challenges in dysplasia assessment and the need for external validation. This study provides a foundation for future clinical validation.
SN - 2398-6352
ER -