دورية أكاديمية
أعرض في EDS

Improved Training Efficiency for Retinopathy of Prematurity Deep Learning Models Using Comparison versus Class Labels.

التفاصيل البيبلوغرافية
العنوان: Improved Training Efficiency for Retinopathy of Prematurity Deep Learning Models Using Comparison versus Class Labels.
المؤلفون: Hanif A; Department of Ophthalmology, Oregon Health & Science University, Portland, Oregon., Yıldız İ; Department of Electrical and Computer Engineering, Northeastern University, Boston, Massachusetts., Tian P; Department of Electrical and Computer Engineering, Northeastern University, Boston, Massachusetts., Kalkanlı B; Department of Electrical and Computer Engineering, Northeastern University, Boston, Massachusetts., Erdoğmuş D; Department of Electrical and Computer Engineering, Northeastern University, Boston, Massachusetts., Ioannidis S; Department of Electrical and Computer Engineering, Northeastern University, Boston, Massachusetts., Dy J; Department of Electrical and Computer Engineering, Northeastern University, Boston, Massachusetts., Kalpathy-Cramer J; Department of Radiology, Athinoula A. Martinos Center for Biomedical Imaging Clinical Computational Neuroimaging Group, Charlestown, Massachusetts., Ostmo S; Department of Ophthalmology, Oregon Health & Science University, Portland, Oregon., Jonas K; Department of Ophthalmology, University of Illinois at Chicago College of Medicine, Chicago, Illinois., Chan RVP; Department of Ophthalmology, University of Illinois at Chicago College of Medicine, Chicago, Illinois., Chiang MF; National Eye Institute, National Institutes of Health, Bethesda, Maryland., Campbell JP; Department of Ophthalmology, Oregon Health & Science University, Portland, Oregon.
المصدر: Ophthalmology science [Ophthalmol Sci] 2022 Feb 02; Vol. 2 (2), pp. 100122. Date of Electronic Publication: 2022 Feb 02 (Print Publication: 2022).
نوع المنشور: Journal Article
اللغة: English
بيانات الدورية: Publisher: Elsevier, B.V Country of Publication: Netherlands NLM ID: 9918230896206676 Publication Model: eCollection Cited Medium: Internet ISSN: 2666-9145 (Electronic) Linking ISSN: 26669145 NLM ISO Abbreviation: Ophthalmol Sci Subsets: PubMed not MEDLINE
أسماء مطبوعة: Original Publication: [Amsterdam] : Elsevier, B.V., [2021]-
مستخلص: Purpose: To compare the efficacy and efficiency of training neural networks for medical image classification using comparison labels indicating relative disease severity versus diagnostic class labels from a retinopathy of prematurity (ROP) image dataset.
Design: Evaluation of diagnostic test or technology.
Participants: Deep learning neural networks trained on expert-labeled wide-angle retinal images obtained from patients undergoing diagnostic ROP examinations obtained as part of the Imaging and Informatics in ROP (i-ROP) cohort study.
Methods: Neural networks were trained with either class or comparison labels indicating plus disease severity in ROP retinal fundus images from 2 datasets. After training and validation, all networks underwent evaluation using a separate test dataset in 1 of 2 binary classification tasks: normal versus abnormal or plus versus nonplus.
Main Outcome Measures: Area under the receiver operating characteristic curve (AUC) values were measured to assess network performance.
Results: Given the same number of labels, neural networks learned more efficiently by comparison, generating significantly higher AUCs in both classification tasks across both datasets. Similarly, given the same number of images, comparison learning developed networks with significantly higher AUCs across both classification tasks in 1 of 2 datasets. The difference in efficiency and accuracy between models trained on either label type decreased as the size of the training set increased.
Conclusions: Comparison labels individually are more informative and more abundant per sample than class labels. These findings indicate a potential means of overcoming the common obstacle of data variability and scarcity when training neural networks for medical image classification tasks.
(© 2022 by the American Academy of Ophthalmology.)
References: Radiology. 2019 Feb;290(2):537-544. (PMID: 30422093)
AMIA Annu Symp Proc. 2014 Nov 14;2014:1902-10. (PMID: 25954463)
Ophthalmology. 2016 Nov;123(11):2345-2351. (PMID: 27566853)
Ophthalmology. 2021 Oct;128(10):e51-e68. (PMID: 34247850)
Arch Ophthalmol. 2007 Jul;125(7):875-80. (PMID: 17620564)
Dermatol Ther (Heidelb). 2020 Jun;10(3):365-386. (PMID: 32253623)
Arch Ophthalmol. 2011 May;129(5):591-6. (PMID: 21555612)
Neural Netw. 2019 Oct;118:65-80. (PMID: 31254769)
J Am Coll Radiol. 2020 Dec;17(12):1653-1662. (PMID: 32592660)
Nat Methods. 2019 Jan;16(1):67-70. (PMID: 30559429)
Arch Ophthalmol. 1984 Aug;102(8):1130-4. (PMID: 6547831)
JAMA Ophthalmol. 2018 Jul 1;136(7):803-810. (PMID: 29801159)
Transl Vis Sci Technol. 2020 Feb 27;9(2):14. (PMID: 32704420)
Ophthalmology. 2016 Nov;123(11):2338-2344. (PMID: 27591053)
Psychol Rev. 2005 Oct;112(4):881-911. (PMID: 16262472)
Neural Netw. 2015 Jan;61:85-117. (PMID: 25462637)
Br J Ophthalmol. 2019 Feb;103(2):167-175. (PMID: 30361278)
فهرسة مساهمة: Keywords: ANOVA, analysis of variance; AUC, area under the receiver operating characteristic curve; Artificial intelligence; Deep learning; ICROP, International Classification of Retinopathy of Prematurity; Labels; Neural networks; ROP, retinopathy of prematurity; Retinopathy of prematurity; i-ROP, Imaging and Informatics in ROP
تواريخ الأحداث: Date Created: 20221017 Latest Revision: 20221019
رمز التحديث: 20231215
مُعرف محوري في PubMed: PMC9560533
DOI: 10.1016/j.xops.2022.100122
PMID: 36249702
قاعدة البيانات: MEDLINE
الوصف
تدمد:2666-9145
DOI:10.1016/j.xops.2022.100122