Understanding inherent image features in CNN-based assessment of diabetic retinopathy

Abstract Diabetic retinopathy (DR) is a leading cause of blindness and affects millions of people throughout the world. Early detection and timely checkups are key to reduce the risk of blindness. Automated grading of DR is a cost-effective way to ensure early detection and timely checkups. Deep lea...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Roc Reguant, Søren Brunak, Sajib Saha
Formato:	article
Lenguaje:	EN
Publicado:	Nature Portfolio 2021
Materias:	Medicine R Science Q
Acceso en línea:	https://doaj.org/article/9fba7cc53af54f118f3f3f3889663b8b
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:9fba7cc53af54f118f3f3f3889663b8b
record_format	dspace
spelling	oai:doaj.org-article:9fba7cc53af54f118f3f3f3889663b8b2021-12-02T14:49:43ZUnderstanding inherent image features in CNN-based assessment of diabetic retinopathy10.1038/s41598-021-89225-02045-2322https://doaj.org/article/9fba7cc53af54f118f3f3f3889663b8b2021-05-01T00:00:00Zhttps://doi.org/10.1038/s41598-021-89225-0https://doaj.org/toc/2045-2322Abstract Diabetic retinopathy (DR) is a leading cause of blindness and affects millions of people throughout the world. Early detection and timely checkups are key to reduce the risk of blindness. Automated grading of DR is a cost-effective way to ensure early detection and timely checkups. Deep learning or more specifically convolutional neural network (CNN)—based methods produce state-of-the-art performance in DR detection. Whilst CNN based methods have been proposed, no comparisons have been done between the extracted image features and their clinical relevance. Here we first adopt a CNN visualization strategy to discover the inherent image features involved in the CNN’s decision-making process. Then, we critically analyze those features with respect to commonly known pathologies namely microaneurysms, hemorrhages and exudates, and other ocular components. We also critically analyze different CNNs by considering what image features they pick up during learning to predict and justify their clinical relevance. The experiments are executed on publicly available fundus datasets (EyePACS and DIARETDB1) achieving an accuracy of 89 ~ 95% with AUC, sensitivity and specificity of respectively 95 ~ 98%, 74 ~ 86%, and 93 ~ 97%, for disease level grading of DR. Whilst different CNNs produce consistent classification results, the rate of picked-up image features disagreement between models could be as high as 70%.Roc ReguantSøren BrunakSajib SahaNature PortfolioarticleMedicineRScienceQENScientific Reports, Vol 11, Iss 1, Pp 1-12 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Medicine R Science Q
spellingShingle	Medicine R Science Q Roc Reguant Søren Brunak Sajib Saha Understanding inherent image features in CNN-based assessment of diabetic retinopathy
description	Abstract Diabetic retinopathy (DR) is a leading cause of blindness and affects millions of people throughout the world. Early detection and timely checkups are key to reduce the risk of blindness. Automated grading of DR is a cost-effective way to ensure early detection and timely checkups. Deep learning or more specifically convolutional neural network (CNN)—based methods produce state-of-the-art performance in DR detection. Whilst CNN based methods have been proposed, no comparisons have been done between the extracted image features and their clinical relevance. Here we first adopt a CNN visualization strategy to discover the inherent image features involved in the CNN’s decision-making process. Then, we critically analyze those features with respect to commonly known pathologies namely microaneurysms, hemorrhages and exudates, and other ocular components. We also critically analyze different CNNs by considering what image features they pick up during learning to predict and justify their clinical relevance. The experiments are executed on publicly available fundus datasets (EyePACS and DIARETDB1) achieving an accuracy of 89 ~ 95% with AUC, sensitivity and specificity of respectively 95 ~ 98%, 74 ~ 86%, and 93 ~ 97%, for disease level grading of DR. Whilst different CNNs produce consistent classification results, the rate of picked-up image features disagreement between models could be as high as 70%.
format	article
author	Roc Reguant Søren Brunak Sajib Saha
author_facet	Roc Reguant Søren Brunak Sajib Saha
author_sort	Roc Reguant
title	Understanding inherent image features in CNN-based assessment of diabetic retinopathy
title_short	Understanding inherent image features in CNN-based assessment of diabetic retinopathy
title_full	Understanding inherent image features in CNN-based assessment of diabetic retinopathy
title_fullStr	Understanding inherent image features in CNN-based assessment of diabetic retinopathy
title_full_unstemmed	Understanding inherent image features in CNN-based assessment of diabetic retinopathy
title_sort	understanding inherent image features in cnn-based assessment of diabetic retinopathy
publisher	Nature Portfolio
publishDate	2021
url	https://doaj.org/article/9fba7cc53af54f118f3f3f3889663b8b
work_keys_str_mv	AT rocreguant understandinginherentimagefeaturesincnnbasedassessmentofdiabeticretinopathy AT sørenbrunak understandinginherentimagefeaturesincnnbasedassessmentofdiabeticretinopathy AT sajibsaha understandinginherentimagefeaturesincnnbasedassessmentofdiabeticretinopathy
_version_	1718389464237080576

Understanding inherent image features in CNN-based assessment of diabetic retinopathy

Ejemplares similares