@inproceedings{e56d47c015c2449cb98171c3d024ab3f,
title = "Effect of domain-specific self-supervised pretraining on predictive uncertainty for colorectal polyp characterization",
abstract = "Colorectal polyps (CRPs) are potential precursors of colorectal cancer (CRC), one of the most common types of cancer worldwide. Computer-Aided Diagnosis (CADx) systems can play a crucial role as a second opinion for endoscopists in characterizing CRPs and contribute to the diagnostic performance of colonoscopies. Despite their potential, deep neural network-based systems often tend to overestimate the confidence about their decisions and provide predictive probabilities that are poorly related to their classification accuracy. Quantifying uncertainty of such supportive systems is crucial for optimal clinical workflow integration and physician's acceptance. Thus, a trustworthy CADx system is expected to provide accurate and well-calibrated classification confidence. Transfer learning from either natural image datasets, such as ImageNet, or other datasets with similar modalities, has been widely used for improving the accuracy of deep learning-based systems in medical image classification. In this paper, we study the impact of domain-specific pretraining on the calibration and the overall performance of a CADx system for CRP characterization. We evaluate our hypothesis on a fully deterministic and a hybrid Bayesian version of each approach using a generic ResNet50 architecture. Experimental results demonstrate the effectiveness of domain-specific pretraining in achieving a higher overall characterization AUC. Additionally, the in-domain and out-of-domain pretrained models portray similar calibration error rates, however, their corresponding hybrid Bayesian models offer higher robustness with improved calibration performance. A hybrid Bayesian version of a domain-specific pretraining approach has shown to significantly improve the accuracy and reliability of CADx systems used for CRP characterization and similar positive effects may be expected for other medical imaging applications.",
keywords = "Colorectal polyp characterization, Computer-aided diagnosis systems, Domain-specific pretraining",
author = "N. Dehghani and T. Scheeve and Boers, {T. G.W.} and {van der Zander}, {Q. E.W.} and A. Thijssen and R. Schreuder and Masclee, {A. A.M.} and Schoon, {E. J.} and {van der Sommen}, F. and {de With}, {P. H.N.}",
note = "Funding Information: We acknowledge the Dutch Cancer Society for funding the COMET-OPTICAL project (Project No. 12639). We gratefully acknowledge Prof. Bergman and his research group at the Amsdterdam UMC for their contribution of the GastroNet dataset. This work has not been submitted for publication anywhere else. Publisher Copyright: {\textcopyright} 2023 SPIE.; Medical Imaging 2023: Computer-Aided Diagnosis ; Conference date: 19-02-2023 Through 23-02-2023",
year = "2023",
month = jan,
day = "1",
doi = "10.1117/12.2653848",
language = "English",
isbn = "9781510660359",
volume = "12465",
series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",
publisher = "SPIE",
number = "124651G",
editor = "Iftekharuddin, {Khan M.} and Weijie Chen",
booktitle = "Medical Imaging 2023: Computer-Aided Diagnosis",
address = "United States",
edition = "1",
}