Deep learning model for automated detection of Helicobacter pylori and intestinal metaplasia on gastric biopsy digital whole slide images
American Journal of Clinical Pathology · 19.11.2025 · 164 (5) · 830-839
DOI: 10.1093/ajcp/aqaf110 ↗Kurz gefasst
Fachartikel in der Zeitschrift American Journal of Clinical Pathology, online erschienen am 25.09.2025. Verfasst von Li Y Khor und 17 weiteren Autor:innen. Beteiligte Einrichtung: Department of Anatomical Pathology, Singapore General Hospital, Singapore. Hauptthemen laut MEDLINE: Deep Learning, Helicobacter pylori, Helicobacter Infections, Gastritis. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 28 Literaturangaben. Gefördert von: National Research Foundation, Singapore.
Abstract (Originalsprache)
Objective
To develop an automated detection tool for Helicobacter pylori (HP) microorganisms (HPOrg) and intestinal metaplasia (IM) identification on gastric biopsy specimens on hematoxylin and eosin (H&E) whole-slide images (WSIs), incorporating background histopathologic features.
Methods
A total of 180 H&E gastric biopsy WSIs, archived at the Department of Anatomical Pathology, Singapore General Hospital, were used to train, validate, and test (60:20:20) a decision support tool. Eighty WSIs displayed non-HP inflammation; 100 were annotated for HP-associated gastritis, HPOrg, and IM. A 2-stage model was employed—a Vision Transformer–based model filtered artifacts after stain normalization, and then a Graph Attention Network component aggregated patch-level features, giving a prediction for each of 6 tissue sections within each WSI, with a majority vote determining the final WSI prediction.
Results
A total of 776 636 patches were used for training/validation and testing. The optimized model showed HPOrg classification (precision: 0.604, F1-score: 0.617, and top 10 micro F1-score: 0.714) and IM classification (precision: 0.905, F1-score: 0.861, and top 10 micro F1-score: 1.0). The macro average F1-score was 0.739, section-level precision was 0.981, and the F1-score was 0.945. The WSI-level precision achieved was 1.0, with a F1-score of 0.96.
Conclusions
We demonstrate a 2-stage model to detect HP and IM in gastric biopsy specimens, considering background inflammation, which more closely reflects real-world clinical diagnosis.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Anatomical Pathology, Singapore General Hospital, Singapore.
- Pathology Academic Clinical Programme, Duke-NUS Medical School, Singapore.
- AI Singapore, Singapore.
- Department of Pathology and Laboratory Medicine, Kandang Kerbau Women's and Children's Hospital, Singapore.
ORCID-Profile: Li Y Khor, Chee L Cheng
Förderung
- National Research Foundation, Singapore
Zitieren
APA
Khor, L. Y., Neo, C. C., Prathaban, K., Choa, E., Quah, W. K., Lum, E. N., Chen, R., Heng, S. Y., Koh, V. C., Seow, J. X., Jegannathan, N., Shi, R., Loong, S., Song, L. H., Natarajan, A., Ravi, S., Oh, K. S., & Cheng, C. L. (2025). Deep learning model for automated detection of Helicobacter pylori and intestinal metaplasia on gastric biopsy digital whole slide images. American Journal of Clinical Pathology, 164(5), 830-839. https://doi.org/10.1093/ajcp/aqaf110
Vancouver
Khor LY, Neo CC, Prathaban K, Choa E, Quah WK, Lum EN, et al. Deep learning model for automated detection of Helicobacter pylori and intestinal metaplasia on gastric biopsy digital whole slide images. American Journal of Clinical Pathology. 2025;164(5):830-839. doi:10.1093/ajcp/aqaf110
BibTeX
@article{Khor2025,
author = {Khor, Li Y and Neo, Calvin C and Prathaban, Karthik and Choa, Esther and Quah, Wai K and Lum, Eunice N and Chen, Raphael and Heng, Seow Y and Koh, Valerie C and Seow, Jia X and Jegannathan, Nagalakshmi and Shi, Ruoyu and Loong, Shihleone and Song, Lee H and Natarajan, Anand and Ravi, Sudha and Oh, Kevin S and Cheng, Chee L},
title = {Deep learning model for automated detection of Helicobacter pylori and intestinal metaplasia on gastric biopsy digital whole slide images},
journal = {American Journal of Clinical Pathology},
year = {2025},
volume = {164},
number = {5},
pages = {830--839},
doi = {10.1093/ajcp/aqaf110}
}
EUVIMED