FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attention

Diabetic retinopathy (DR) is a major cause of vision loss. Accurate grading of DR is critical to ensure timely and appropriate intervention. DR progression is primarily characterized by the presence of biomarkers including microaneurysms, hemorrhages, and exudates. These markers are small, scattered...

Full description

Saved in:
Bibliographic Details
Main Authors: Chang Yu, Qian Ma, Jing Li, Qiuyang Zhang, Jin Yao, Biao Yan, Zhenhua Wang
Format: Article
Language:English
Published: AIMS Press 2025-02-01
Series:Electronic Research Archive
Subjects:
Online Access:https://www.aimspress.com/article/doi/10.3934/era.2025033
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850264453637996544
author Chang Yu
Qian Ma
Jing Li
Qiuyang Zhang
Jin Yao
Biao Yan
Zhenhua Wang
author_facet Chang Yu
Qian Ma
Jing Li
Qiuyang Zhang
Jin Yao
Biao Yan
Zhenhua Wang
author_sort Chang Yu
collection DOAJ
description Diabetic retinopathy (DR) is a major cause of vision loss. Accurate grading of DR is critical to ensure timely and appropriate intervention. DR progression is primarily characterized by the presence of biomarkers including microaneurysms, hemorrhages, and exudates. These markers are small, scattered, and challenging to detect. To improve DR grading accuracy, we propose FF-ResNet-DR, a deep learning model that leverages frequency domain attention. Traditional attention mechanisms excel at capturing spatial-domain features but neglect valuable frequency domain information. Our model incorporates frequency channel attention modules (FCAM) and frequency spatial attention modules (FSAM). FCAM refines feature representation by fusing frequency and channel information. FSAM enhances the model's sensitivity to fine-grained texture details. Extensive experiments on multiple public datasets demonstrate the superior performance of FF-ResNet-DR compared to state-of-the-art models. It achieves an AUC of 98.1% on the Messidor binary classification task and a joint accuracy of 64.1% on the IDRiD grading task. These results highlight the potential of FF-ResNet-DR as a valuable tool for the clinical diagnosis and management of DR.
format Article
id doaj-art-bcdea22e9d72417e89dc16a30eb580b8
institution OA Journals
issn 2688-1594
language English
publishDate 2025-02-01
publisher AIMS Press
record_format Article
series Electronic Research Archive
spelling doaj-art-bcdea22e9d72417e89dc16a30eb580b82025-08-20T01:54:41ZengAIMS PressElectronic Research Archive2688-15942025-02-0133272574310.3934/era.2025033FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attentionChang Yu0Qian Ma1Jing Li2Qiuyang Zhang3Jin Yao4Biao Yan5Zhenhua Wang6College of Information Technology, Shanghai Ocean University, Shanghai 201306, ChinaGeneral Hospital of Ningxia Medical University, Ningxia 750001, ChinaEye Institute and Department of Ophthalmology, Eye and ENT Hospital, Fudan University, Shanghai 201114, ChinaDepartment of Ophthalmology and Optometry, The Affiliated Eye Hospital, Nanjing Medical University, Nanjing 210029, ChinaDepartment of Ophthalmology and Optometry, The Affiliated Eye Hospital, Nanjing Medical University, Nanjing 210029, ChinaDepartment of Ophthalmology, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200080, ChinaCollege of Information Technology, Shanghai Ocean University, Shanghai 201306, ChinaDiabetic retinopathy (DR) is a major cause of vision loss. Accurate grading of DR is critical to ensure timely and appropriate intervention. DR progression is primarily characterized by the presence of biomarkers including microaneurysms, hemorrhages, and exudates. These markers are small, scattered, and challenging to detect. To improve DR grading accuracy, we propose FF-ResNet-DR, a deep learning model that leverages frequency domain attention. Traditional attention mechanisms excel at capturing spatial-domain features but neglect valuable frequency domain information. Our model incorporates frequency channel attention modules (FCAM) and frequency spatial attention modules (FSAM). FCAM refines feature representation by fusing frequency and channel information. FSAM enhances the model's sensitivity to fine-grained texture details. Extensive experiments on multiple public datasets demonstrate the superior performance of FF-ResNet-DR compared to state-of-the-art models. It achieves an AUC of 98.1% on the Messidor binary classification task and a joint accuracy of 64.1% on the IDRiD grading task. These results highlight the potential of FF-ResNet-DR as a valuable tool for the clinical diagnosis and management of DR.https://www.aimspress.com/article/doi/10.3934/era.2025033diabetic retinopathy (dr)grading modelclassification modelcolor fundus imageattention mechanism
spellingShingle Chang Yu
Qian Ma
Jing Li
Qiuyang Zhang
Jin Yao
Biao Yan
Zhenhua Wang
FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attention
Electronic Research Archive
diabetic retinopathy (dr)
grading model
classification model
color fundus image
attention mechanism
title FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attention
title_full FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attention
title_fullStr FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attention
title_full_unstemmed FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attention
title_short FF-ResNet-DR model: a deep learning model for diabetic retinopathy grading by frequency domain attention
title_sort ff resnet dr model a deep learning model for diabetic retinopathy grading by frequency domain attention
topic diabetic retinopathy (dr)
grading model
classification model
color fundus image
attention mechanism
url https://www.aimspress.com/article/doi/10.3934/era.2025033
work_keys_str_mv AT changyu ffresnetdrmodeladeeplearningmodelfordiabeticretinopathygradingbyfrequencydomainattention
AT qianma ffresnetdrmodeladeeplearningmodelfordiabeticretinopathygradingbyfrequencydomainattention
AT jingli ffresnetdrmodeladeeplearningmodelfordiabeticretinopathygradingbyfrequencydomainattention
AT qiuyangzhang ffresnetdrmodeladeeplearningmodelfordiabeticretinopathygradingbyfrequencydomainattention
AT jinyao ffresnetdrmodeladeeplearningmodelfordiabeticretinopathygradingbyfrequencydomainattention
AT biaoyan ffresnetdrmodeladeeplearningmodelfordiabeticretinopathygradingbyfrequencydomainattention
AT zhenhuawang ffresnetdrmodeladeeplearningmodelfordiabeticretinopathygradingbyfrequencydomainattention