ARTIFICIAL INTELLIGENCE AND PANENDOSCOPY – MULTI-DEVICE AUTOMATIC DETECTION OF PLEOMORPHIC LESIONS DURING DEVICE-ASSISTED ENTEROSCOPY
Journal of Coloproctology · 01.12.2024 · 44 (S 01 S 01) · S1-S138
DOI: 10.1055/s-0045-1808889 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Coloproctology, erschienen am 01.12.2024. Verfasst von Francisco Mendes und 7 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Introduction
Device-assisted enteroscopy is the only diagnostic and therapeutic procedure that allows evaluation of the entire gastrointestinal tract. During the procedure, different types of lesions are detected, including vascular lesions, protruding lesions, ulcers, and erosions. However, its diagnostic accuracy is suboptimal. Convolutional neural networks, which are artificial intelligence models developed and optimized for image analysis, have shown their usefulness in various areas of medicine, although there is limited data on their utility in device-assisted enteroscopy.
Objective
The present study aimed to develop a multi-device convolutional neural network for the detection and differentiation of pleomorphic lesions during device-assisted enteroscopy.
Methods
A retrospective evaluation was performed on 338 procedures conducted at two reference centers, including 152 single-balloon enteroscopies (Fujifilm®), 172 double-balloon enteroscopies (Olympus®), and 14 spiral enteroscopies (Olympus®). The convolutional neural network was trained with 40,655 images (esophageal, gastric, enteric, and colonic), with 90% (36,599 images) used for the training dataset in a cross-validation design with 5 subgroups, and the remaining 10% (4,066 images) used for validation. The network's classification was compared with the expert consensus classification and evaluated based on sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and the area under the precision-recall curve.
Results
The convolutional neural network detected pleomorphic lesions with a sensitivity of 88.9%, specificity of 98.9%, positive predictive value of 95.8%, and negative predictive value of 97.1%, with an accuracy of 96.8% and an area under the precision-recall curve of 0.97.
Conclusion
Our group developed the first convolutional neural network worldwide for multi-device panendoscopic detection of pleomorphic lesions during device-assisted enteroscopy. This work demonstrates the importance of developing deep learning models with high accuracy to improve diagnostic accuracy in panendoscopic evaluation of pleomorphic lesions. Publication History Article published online: 25 April 2025 © 2025. The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution 4.0 International License, permitting copying and reproduction so long as the original work is given appropriate credit (https://creativecommons.org/licenses/by/4.0/) Thieme Revinter Publicações Ltda. Rua do Matoso 170, Rio de Janeiro, RJ, CEP 20270-135, Brazil
Zitieren
APA
Mendes, F., Mascarenhas, M., Ribeiro, T., Afonso, J., Cardoso, P., Martins, M., Saraiva, M. M., & Macedo, G. (2024). ARTIFICIAL INTELLIGENCE AND PANENDOSCOPY – MULTI-DEVICE AUTOMATIC DETECTION OF PLEOMORPHIC LESIONS DURING DEVICE-ASSISTED ENTEROSCOPY. Journal of Coloproctology, 44(S 01 S 01), S1-S138. https://doi.org/10.1055/s-0045-1808889
Vancouver
Mendes F, Mascarenhas M, Ribeiro T, Afonso J, Cardoso P, Martins M, et al. ARTIFICIAL INTELLIGENCE AND PANENDOSCOPY – MULTI-DEVICE AUTOMATIC DETECTION OF PLEOMORPHIC LESIONS DURING DEVICE-ASSISTED ENTEROSCOPY. Journal of Coloproctology. 2024;44(S 01 S 01):S1-S138. doi:10.1055/s-0045-1808889
BibTeX
@article{Mendes2024,
author = {Mendes, Francisco and Mascarenhas, Miguel and Ribeiro, Tiago and Afonso, João and Cardoso, Pedro and Martins, Miguel and Saraiva, Miguel Mascarenhas and Macedo, Guilherme},
title = {ARTIFICIAL INTELLIGENCE AND PANENDOSCOPY – MULTI-DEVICE AUTOMATIC DETECTION OF PLEOMORPHIC LESIONS DURING DEVICE-ASSISTED ENTEROSCOPY},
journal = {Journal of Coloproctology},
year = {2024},
volume = {44},
number = {S 01 S 01},
pages = {S1--S138},
doi = {10.1055/s-0045-1808889}
}
EUVIMED