TY - JOUR
T1 - Endoscopy-driven pretraining for classification of dysplasia in barrett's esophagus with endoscopic narrow-band imaging zoom videos
AU - van der Putten, Joost
AU - Struyvenberg, Maarten
AU - de Groof, Jeroen
AU - Curvers, Wouter
AU - Schoon, Erik
AU - Baldaque-Silva, Francisco
AU - Bergman, Jacques
AU - van der Sommen, Fons
AU - de With, Peter H.N.
PY - 2020/5/1
Y1 - 2020/5/1
N2 - 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.
AB - 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.
KW - Barrett's esophagus
KW - Classification
KW - Deep learning
KW - Endoscopic zoomimagery
KW - Machine learning
UR - http://www.scopus.com/inward/record.url?scp=85085697661&partnerID=8YFLogxK
U2 - 10.3390/APP10103407
DO - 10.3390/APP10103407
M3 - Article
AN - SCOPUS:85085697661
VL - 10
JO - Applied Sciences
JF - Applied Sciences
SN - 2076-3417
IS - 10
M1 - 3407
ER -