Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification

High annotation costs are a substantial bottleneck in applying deep learning architectures to clinically relevant use cases, substantiating the need for algorithms to learn from unlabeled data. In this work, we propose employing self-supervised methods. To that end, we trained with three self-superv...

Descripción completa

Detalles Bibliográficos
Autores principales: Taleb, Aiham, Rohrer, Csaba, Bergner, Benjamin, De Leon, Guilherme, Rodrigues, Jonas Almeida, Schwendicke, Falk, Lippert, Christoph, Krois, Joachim
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2022
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9140204/
https://www.ncbi.nlm.nih.gov/pubmed/35626392
http://dx.doi.org/10.3390/diagnostics12051237
_version_ 1784715041539358720
author Taleb, Aiham
Rohrer, Csaba
Bergner, Benjamin
De Leon, Guilherme
Rodrigues, Jonas Almeida
Schwendicke, Falk
Lippert, Christoph
Krois, Joachim
author_facet Taleb, Aiham
Rohrer, Csaba
Bergner, Benjamin
De Leon, Guilherme
Rodrigues, Jonas Almeida
Schwendicke, Falk
Lippert, Christoph
Krois, Joachim
author_sort Taleb, Aiham
collection PubMed
description High annotation costs are a substantial bottleneck in applying deep learning architectures to clinically relevant use cases, substantiating the need for algorithms to learn from unlabeled data. In this work, we propose employing self-supervised methods. To that end, we trained with three self-supervised algorithms on a large corpus of unlabeled dental images, which contained 38K bitewing radiographs (BWRs). We then applied the learned neural network representations on tooth-level dental caries classification, for which we utilized labels extracted from electronic health records (EHRs). Finally, a holdout test-set was established, which consisted of 343 BWRs and was annotated by three dental professionals and approved by a senior dentist. This test-set was used to evaluate the fine-tuned caries classification models. Our experimental results demonstrate the obtained gains by pretraining models using self-supervised algorithms. These include improved caries classification performance (6 p.p. increase in sensitivity) and, most importantly, improved label-efficiency. In other words, the resulting models can be fine-tuned using few labels (annotations). Our results show that using as few as 18 annotations can produce ≥45% sensitivity, which is comparable to human-level diagnostic performance. This study shows that self-supervision can provide gains in medical image analysis, particularly when obtaining labels is costly and expensive.
format Online
Article
Text
id pubmed-9140204
institution National Center for Biotechnology Information
language English
publishDate 2022
publisher MDPI
record_format MEDLINE/PubMed
spelling pubmed-91402042022-05-28 Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification Taleb, Aiham Rohrer, Csaba Bergner, Benjamin De Leon, Guilherme Rodrigues, Jonas Almeida Schwendicke, Falk Lippert, Christoph Krois, Joachim Diagnostics (Basel) Article High annotation costs are a substantial bottleneck in applying deep learning architectures to clinically relevant use cases, substantiating the need for algorithms to learn from unlabeled data. In this work, we propose employing self-supervised methods. To that end, we trained with three self-supervised algorithms on a large corpus of unlabeled dental images, which contained 38K bitewing radiographs (BWRs). We then applied the learned neural network representations on tooth-level dental caries classification, for which we utilized labels extracted from electronic health records (EHRs). Finally, a holdout test-set was established, which consisted of 343 BWRs and was annotated by three dental professionals and approved by a senior dentist. This test-set was used to evaluate the fine-tuned caries classification models. Our experimental results demonstrate the obtained gains by pretraining models using self-supervised algorithms. These include improved caries classification performance (6 p.p. increase in sensitivity) and, most importantly, improved label-efficiency. In other words, the resulting models can be fine-tuned using few labels (annotations). Our results show that using as few as 18 annotations can produce ≥45% sensitivity, which is comparable to human-level diagnostic performance. This study shows that self-supervision can provide gains in medical image analysis, particularly when obtaining labels is costly and expensive. MDPI 2022-05-16 /pmc/articles/PMC9140204/ /pubmed/35626392 http://dx.doi.org/10.3390/diagnostics12051237 Text en © 2022 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
spellingShingle Article
Taleb, Aiham
Rohrer, Csaba
Bergner, Benjamin
De Leon, Guilherme
Rodrigues, Jonas Almeida
Schwendicke, Falk
Lippert, Christoph
Krois, Joachim
Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification
title Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification
title_full Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification
title_fullStr Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification
title_full_unstemmed Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification
title_short Self-Supervised Learning Methods for Label-Efficient Dental Caries Classification
title_sort self-supervised learning methods for label-efficient dental caries classification
topic Article
url https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9140204/
https://www.ncbi.nlm.nih.gov/pubmed/35626392
http://dx.doi.org/10.3390/diagnostics12051237
work_keys_str_mv AT talebaiham selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification
AT rohrercsaba selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification
AT bergnerbenjamin selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification
AT deleonguilherme selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification
AT rodriguesjonasalmeida selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification
AT schwendickefalk selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification
AT lippertchristoph selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification
AT kroisjoachim selfsupervisedlearningmethodsforlabelefficientdentalcariesclassification