One-Vote Veto: Semi-Supervised Learning for Low-Shot Glaucoma Diagnosis
IEEE Transactions on Medical Imaging · 2023-12 · 42 (12) · 3764-3778
DOI: 10.1109/tmi.2023.3307689 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Medical Imaging, 2023-12. Verfasst von Rui Fan und 6 weiteren Autor:innen. Hauptthemen laut MEDLINE: Glaucoma. Die Arbeit wurde bisher 30-mal zitiert (Crossref) und nennt 69 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NEI NIH HHS, Fundamental Research Funds for the Central Universities, Awards from the National Eye Institute.
Abstract (Originalsprache)
Convolutional neural networks (CNNs) are a promising technique for automated glaucoma diagnosis from images of the fundus, and these images are routinely acquired as part of an ophthalmic exam. Nevertheless, CNNs typically require a large amount of well-labeled data for training, which may not be available in many biomedical image classification applications, especially when diseases are rare and where labeling by experts is costly. This article makes two contributions to address this issue: 1) It extends the conventional Siamese network and introduces a training method for low-shot learning when labeled data are limited and imbalanced, and 2) it introduces a novel semi-supervised learning strategy that uses additional unlabeled training data to achieve greater accuracy. Our proposed multi-task Siamese network (MTSN) can employ any backbone CNN, and we demonstrate with four backbone CNNs that its accuracy with limited training data approaches the accuracy of backbone CNNs trained with a dataset that is 50 times larger. We also introduce One-Vote Veto (OVV) self-training, a semi-supervised learning strategy that is designed specifically for MTSNs. By taking both self-predictions and contrastive predictions of the unlabeled training data into account, OVV self-training provides additional pseudo labels for fine-tuning a pre-trained MTSN. Using a large (imbalanced) dataset with 66,715 fundus photographs acquired over 15 years, extensive experimental results demonstrate the effectiveness of low-shot learning with MTSN and semi-supervised learning with OVV self-training. Three additional, smaller clinical datasets of fundus images acquired under different conditions (cameras, instruments, locations, populations) are used to demonstrate the generalizability of the proposed methods.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Förderung
- NEI NIH HHS
- Fundamental Research Funds for the Central Universities
- Awards from the National Eye Institute
- National Center on Minority Health and Health Disparities, National Institutes of Health
- Horncrest Foundation
- Department of Ophthalmology and Visual Sciences, Washington University in St. Louis
- NIH Vision Core
- Merck Research Laboratories, Pfizer Inc., White House Station, NJ, USA
Zitieren
APA
Fan, R., Bowd, C., Brye, N., Christopher, M., Weinreb, R. N., Kriegman, D. J., & Zangwill, L. M. (2023). One-Vote Veto: Semi-Supervised Learning for Low-Shot Glaucoma Diagnosis. IEEE Transactions on Medical Imaging, 42(12), 3764-3778. https://doi.org/10.1109/tmi.2023.3307689
Vancouver
Fan R, Bowd C, Brye N, Christopher M, Weinreb RN, Kriegman DJ, et al. One-Vote Veto: Semi-Supervised Learning for Low-Shot Glaucoma Diagnosis. IEEE Transactions on Medical Imaging. 2023;42(12):3764-3778. doi:10.1109/tmi.2023.3307689
BibTeX
@article{Fan2023,
author = {Fan, Rui and Bowd, Christopher and Brye, Nicole and Christopher, Mark and Weinreb, Robert N and Kriegman, David J and Zangwill, Linda M},
title = {One-Vote Veto: Semi-Supervised Learning for Low-Shot Glaucoma Diagnosis},
journal = {IEEE Transactions on Medical Imaging},
year = {2023},
volume = {42},
number = {12},
pages = {3764--3778},
doi = {10.1109/tmi.2023.3307689}
}
EUVIMED