Detection of glaucoma using three-stage training with EfficientNet

This paper sets forth a methodology that is based on three-stage-training of a state-of-the-art network architecture previously trained on Imagenet, and iteratively finetuned in three steps; freezing first all layers, then re-training a specific number of them and finally training all the architectu...

Full description

Bibliographic Details
Authors: de Zarzà i Cubero, I., de Curtò y DíAz, J., Calafate, Carlos
Format: article
Status:Published version
Publication Date:2022
Country:España
Institution:Universitat Oberta de Catalunya (UOC)
Repository:O2, repositorio institucional de la UOC
OAI Identifier:oai:openaccess.uoc.edu:10609/149211
Online Access:http://hdl.handle.net/10609/149211
https://doi.org/10.1016/j.iswa.2022.200140
Access Level:Open access
Keyword:glaucoma
fundus images
efficientNet
id ES_ef276c8660bcd01ea2e184ca37ea0669
oai_identifier_str oai:openaccess.uoc.edu:10609/149211
network_acronym_str ES
network_name_str España
repository_id_str
spelling Detection of glaucoma using three-stage training with EfficientNetde Zarzà i Cubero, I.de Curtò y DíAz, J.Calafate, Carlosglaucomafundus imagesefficientNetThis paper sets forth a methodology that is based on three-stage-training of a state-of-the-art network architecture previously trained on Imagenet, and iteratively finetuned in three steps; freezing first all layers, then re-training a specific number of them and finally training all the architecture from scratch, to achieve a system with high accuracy and reliability. To determine the performance of our technique a dataset consisting of 17.070 color cropped samples of fundus images, and that includes two classes, normal and abnormal, is used. Extensive evaluations using baselines models (VGG16, InceptionV3 and Resnet50) are carried out, in addition to thorough experimentation with the proposed pipeline using variants of EfficientNet and EfficientNetV2. The training procedure is described accurately, putting emphasis on the number of parameters trained, the confusion matrices (with analysis of false positives and false negatives), accuracy, and F1-score obtained at each stage of the proposed methodology. The results achieved show that the intelligent system presented for the task at hand is reliable, presents high precision, its predictions are consistent and the number of parameters needed to train are low compared to other alternatives.Elsevier202320232022info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionapplication/pdfapplication/pdfhttp://hdl.handle.net/10609/149211https://doi.org/10.1016/j.iswa.2022.200140reponame:O2, repositorio institucional de la UOCinstname:Universitat Oberta de Catalunya (UOC)InglésIntelligent Systems with Applications, 2022, 16https://www.sciencedirect.com/science/article/pii/S2667305322000771?via%3Dihubinfo:eu-repo/grantAgreement/PID2021-122580NB-I00//CC BYhttps://creativecommons.org/licenses/by/4.0/info:eu-repo/semantics/openAccessoai:openaccess.uoc.edu:10609/1492112026-05-28T12:42:01Z
dc.title.none.fl_str_mv Detection of glaucoma using three-stage training with EfficientNet
title Detection of glaucoma using three-stage training with EfficientNet
spellingShingle Detection of glaucoma using three-stage training with EfficientNet
de Zarzà i Cubero, I.
glaucoma
fundus images
efficientNet
title_short Detection of glaucoma using three-stage training with EfficientNet
title_full Detection of glaucoma using three-stage training with EfficientNet
title_fullStr Detection of glaucoma using three-stage training with EfficientNet
title_full_unstemmed Detection of glaucoma using three-stage training with EfficientNet
title_sort Detection of glaucoma using three-stage training with EfficientNet
dc.creator.none.fl_str_mv de Zarzà i Cubero, I.
de Curtò y DíAz, J.
Calafate, Carlos
author de Zarzà i Cubero, I.
author_facet de Zarzà i Cubero, I.
de Curtò y DíAz, J.
Calafate, Carlos
author_role author
author2 de Curtò y DíAz, J.
Calafate, Carlos
author2_role author
author
dc.subject.none.fl_str_mv glaucoma
fundus images
efficientNet
topic glaucoma
fundus images
efficientNet
description This paper sets forth a methodology that is based on three-stage-training of a state-of-the-art network architecture previously trained on Imagenet, and iteratively finetuned in three steps; freezing first all layers, then re-training a specific number of them and finally training all the architecture from scratch, to achieve a system with high accuracy and reliability. To determine the performance of our technique a dataset consisting of 17.070 color cropped samples of fundus images, and that includes two classes, normal and abnormal, is used. Extensive evaluations using baselines models (VGG16, InceptionV3 and Resnet50) are carried out, in addition to thorough experimentation with the proposed pipeline using variants of EfficientNet and EfficientNetV2. The training procedure is described accurately, putting emphasis on the number of parameters trained, the confusion matrices (with analysis of false positives and false negatives), accuracy, and F1-score obtained at each stage of the proposed methodology. The results achieved show that the intelligent system presented for the task at hand is reliable, presents high precision, its predictions are consistent and the number of parameters needed to train are low compared to other alternatives.
publishDate 2022
dc.date.none.fl_str_mv 2022
2023
2023
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/10609/149211
https://doi.org/10.1016/j.iswa.2022.200140
url http://hdl.handle.net/10609/149211
https://doi.org/10.1016/j.iswa.2022.200140
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv Intelligent Systems with Applications, 2022, 16
https://www.sciencedirect.com/science/article/pii/S2667305322000771?via%3Dihub
info:eu-repo/grantAgreement/PID2021-122580NB-I00//
dc.rights.none.fl_str_mv CC BY
https://creativecommons.org/licenses/by/4.0/
info:eu-repo/semantics/openAccess
rights_invalid_str_mv CC BY
https://creativecommons.org/licenses/by/4.0/
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
application/pdf
dc.publisher.none.fl_str_mv Elsevier
publisher.none.fl_str_mv Elsevier
dc.source.none.fl_str_mv reponame:O2, repositorio institucional de la UOC
instname:Universitat Oberta de Catalunya (UOC)
instname_str Universitat Oberta de Catalunya (UOC)
reponame_str O2, repositorio institucional de la UOC
collection O2, repositorio institucional de la UOC
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869423836866281472
score 15,301603

Similar Items