Endoscopy-Driven Pretraining for Classification of Dysplasia in Barrett’s Esophagus with Endoscopic Narrow-Band Imaging Zoom Videos

• Published in: Applied sciences Vol. 10; no. 10; p. 3407
• Main Authors: van der Putten, Joost, Struyvenberg, Maarten, de Groof, Jeroen, Curvers, Wouter, Schoon, Erik, Baldaque-Silva, Francisco, Bergman, Jacques, van der Sommen, Fons, de With, Peter H.N
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 01-05-2020
• Subjects: Abnormalities; Accuracy; Algorithms; Barrett esophagus; Barrett’s esophagus; Classification; Datasets; Deep learning; Diagnosis; Dysplasia; endoscopic zoom imagery; Endoscopy; Esophagus; Gastroenterology; Imagery; Inspection; machine learning; Patients; Physiological aspects; Studies; Surveillance; Netherlands; Japan
• ISSN: 2076-3417; 2076-3417
• DOI: 10.3390/app10103407

Endoscopic diagnosis of early neoplasia in Barrett’s Esophagus is generally a two-step process of primary detection in overview, followed by detailed inspection of any visible abnormalities using Narrow Band Imaging (NBI). However, endoscopists struggle with evaluating NBI-zoom imagery of subtle abnormalities. In this work, we propose the first results of a deep learning system for the characterization of NBI-zoom imagery of Barrett’s Esophagus with an accuracy, sensitivity, and specificity of 83.6%, 83.1%, and 84.0%, respectively. We also show that endoscopy-driven pretraining outperforms two models, one without pretraining as well as a model with ImageNet initialization. The final model outperforms absence of pretraining by approximately 10% and the performance is 2% higher in terms of accuracy compared to ImageNet pretraining. Furthermore, the practical deployment of our model is not hampered by ImageNet licensing, thereby paving the way for clinical application.