EUVIMEDEUVIMED entdecken
Crossref

Unsupervised Single-Class Anomaly Detection in Rat Liver Using a Model of Normal Histomorphology

Thomas Forest iD, Raghav Amaravadi iD, Richard Baumgartner iD, Kyathanahalli Janardhan iD, Tiffany Jenkins, Andy Liaw iD, Geetank Raipuria, Falgun Shah, Nitin Singhal iD, Rajath Soans, Takayuki Tsuchiya iD, Roujia Wang iD

Toxicologic Pathology · 18.08.2026

DOI: 10.1177/01926233261475696 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Toxicologic Pathology, online erschienen am 18.08.2026. Verfasst von Thomas Forest und 11 weiteren Autor:innen. Beteiligte Einrichtung: Merck & Co., Inc., Rahway, New Jersey, USA. Die Arbeit nennt 36 Literaturangaben.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

An unsupervised, single-class anomaly detection approach based on a vision transformer architecture was developed to aid histopathology evaluation of rat liver from nonclinical toxicology studies. The approach was designed to detect any histopathology finding without requiring specific histopathology examples for training. The model was developed by training a vision transformer with approximately 135,000 whole slide images (WSIs) and then extracting high-dimensional feature representations from 750 unremarkable vehicle control WSIs, creating a rat liver histomorphology reference dataset representing typical biological variability. Each test WSI was compared against the reference dataset at the tile level, and the resulting anomaly detections were aggregated to derive an overall image-level anomaly score. Anomaly scores were higher in affected groups treated with established toxicants than in control groups, and based on comparing anomaly scores at the individual animal level with pathologist diagnoses, model sensitivity was 95.9% and specificity was 70.8%. Utility as a pathologist aiding tool was further studied by comparing tile-level anomaly annotations with diagnosis. For one treated group, histomorphology feature extraction was used to explore discordance between anomaly detection and diagnosis. Specificity was impacted by pre-analytical variables as well as detection of anomalies not considered noteworthy by pathologists.

Schlagwörter

Stichwörter der Autor:innen

anomaly detectorartificial intelligencefoundation modellivernonclinical toxicologyratsingle class

Autor:innen und Institutionen

  • Merck & Co., Inc., Rahway, New Jersey, USA.
  • Aira Matrix Private Limited, Thane, India.

ORCID-Profile: Thomas Forest, Raghav Amaravadi, Richard Baumgartner, Kyathanahalli Janardhan, Andy Liaw, Nitin Singhal, Takayuki Tsuchiya, Roujia Wang

Zitieren

APA

Forest, T., Amaravadi, R., Baumgartner, R., Janardhan, K., Jenkins, T., Liaw, A., Raipuria, G., Shah, F., Singhal, N., Soans, R., Tsuchiya, T., & Wang, R. (2026). Unsupervised Single-Class Anomaly Detection in Rat Liver Using a Model of Normal Histomorphology. Toxicologic Pathology. https://doi.org/10.1177/01926233261475696

Vancouver

Forest T, Amaravadi R, Baumgartner R, Janardhan K, Jenkins T, Liaw A, et al. Unsupervised Single-Class Anomaly Detection in Rat Liver Using a Model of Normal Histomorphology. Toxicologic Pathology. 2026. doi:10.1177/01926233261475696

BibTeX
@article{Forest2026,
  author = {Forest, Thomas and Amaravadi, Raghav and Baumgartner, Richard and Janardhan, Kyathanahalli and Jenkins, Tiffany and Liaw, Andy and Raipuria, Geetank and Shah, Falgun and Singhal, Nitin and Soans, Rajath and Tsuchiya, Takayuki and Wang, Roujia},
  title = {Unsupervised Single-Class Anomaly Detection in Rat Liver Using a Model of Normal Histomorphology},
  journal = {Toxicologic Pathology},
  year = {2026},
  doi = {10.1177/01926233261475696}
}
Kostenlos starten →