Human–machine interaction in optical polyp diagnosis: decision-making after CADx polyp diagnosis
Journal of the Canadian Association of Gastroenterology · 01.08.2026 · 9 (4) · 204-209
DOI: 10.1093/jcag/gwag016 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of the Canadian Association of Gastroenterology, online erschienen am 29.04.2026. Verfasst von Delphine Dubois und 8 weiteren Autor:innen. Beteiligte Einrichtung: University of Montreal Hospital Research Center (CRCHUM), Montreal, Quebec, H2X0A9, Canada. Die Arbeit nennt 16 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Fonds de Recherche du Québec Santé.
Abstract (Originalsprache)
Background
Artificial intelligence (AI)-based computer-aided diagnosis (CADx) systems are increasingly used to classify colorectal polyps during colonoscopy, but their impact depends on how endoscopists interpret AI predictions. This study aimed to evaluate endoscopists’ decision-making after reviewing CADx outputs, using histopathology diagnosis as ground truth. Patients and methods We performed a secondary analysis of a prospective study (NCT06822816) conducted at our centre between 2022 and 2025. All patients and polyps ≤10 mm with documented CADx output and endoscopist diagnosis were included. Endoscopists either accepted CADx classifications (neoplastic vs hyperplastic) or provided alternatives. Our primary outcome was the accuracy of accepted CADx diagnoses; secondary outcomes assessed reclassification patterns after CADx output rejection.
Results
Our cohort included 754 polyps <10 mm from 367 patients. Accepted CADx diagnoses were correct for 89.1% of neoplastic and 68.7% of hyperplastic CADx predictions. Endoscopists accepted incorrect CADx classifications more frequently when these were neoplastic compared to hyperplastic (64.9% vs 26.5%). Upon rejecting correct CADx neoplastic predictions, endoscopists misclassified 23.4% as sessile serrated lesions (SSLs). Similarly, 29.5% of rejected correct CADx hyperplastic predictions were incorrectly reassigned as SSLs.
Conclusions
Endoscopists exhibit a diagnostic bias toward neoplasia, particularly misclassifying hyperplastic polyps as SSLs. While this may enhance safety, it may risk unnecessary surveillance colonoscopies and increase costs.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- University of Montreal Hospital Research Center (CRCHUM), Montreal, Quebec, H2X0A9, Canada.
- Department of Veterans Affairs Medical Center, White River Junction, VT and The Geisel School of Medicine at Dartmouth, Hanover, NH 05009, United States.
- Division of Gastroenterology/Hepatology, IN University School of Medicine, Indianapolis, IN 46202, United States.
- Gastroenterology, Endoscopy Unit, IRCCS Humanitas Research Hospital, Rozzano, 20089, Italy.
- Division of Gastroenterology, University of Montreal Hospital Center (CHUM), Montreal, Quebec, H2X0C1, Canada.
ORCID-Profile: Delphine Dubois, Roupen Djinbachian
Förderung
- Fonds de Recherche du Québec Santé
Zitieren
APA
Dubois, D., Jalal, Y., Pohl, H., Rex, D. K., Hassan, C., Djinbachian, R., Oleksiw, M., Michal, V., & von Renteln, D. (2026). Human–machine interaction in optical polyp diagnosis: decision-making after CADx polyp diagnosis. Journal of the Canadian Association of Gastroenterology, 9(4), 204-209. https://doi.org/10.1093/jcag/gwag016
Vancouver
Dubois D, Jalal Y, Pohl H, Rex DK, Hassan C, Djinbachian R, et al. Human–machine interaction in optical polyp diagnosis: decision-making after CADx polyp diagnosis. Journal of the Canadian Association of Gastroenterology. 2026;9(4):204-209. doi:10.1093/jcag/gwag016
BibTeX
@article{Dubois2026,
author = {Dubois, Delphine and Jalal, Yousr and Pohl, Heiko and Rex, Douglas K and Hassan, Cesare and Djinbachian, Roupen and Oleksiw, Megan and Michal, Victoire and von Renteln, Daniel},
title = {Human–machine interaction in optical polyp diagnosis: decision-making after CADx polyp diagnosis},
journal = {Journal of the Canadian Association of Gastroenterology},
year = {2026},
volume = {9},
number = {4},
pages = {204--209},
doi = {10.1093/jcag/gwag016}
}
EUVIMED