Detection of Diabetic Retinopathy in Retinal Fundus Images Using CNN Classification Models

被引:26
作者
Asia, Al-Omaisi [1 ]
Zhu, Cheng-Zhang [1 ,2 ]
Althubiti, Sara A. [3 ]
Al-Alimi, Dalal [4 ]
Xiao, Ya-Long [1 ,2 ]
Ouyang, Ping-Bo [5 ]
Al-Qaness, Mohammed A. A. [6 ,7 ]
机构
[1] Cent South Univ, Sch Comp Sci & Engn, Changsha 410083, Peoples R China
[2] Cent South Univ, Coll Literature & Journalism, Changsha 410083, Peoples R China
[3] Majmaah Univ, Coll Comp & Informat Sci, Dept Comp Sci, Al Majmaah 11952, Saudi Arabia
[4] China Univ Geosci, Sch Comp Sci, Wuhan 430074, Peoples R China
[5] Cent South Univ, Xiangya Hosp 2, Dept Ophthalmol, Changsha 410083, Peoples R China
[6] Wuhan Univ, State Key Lab Informat Engn Surveying Mapping & R, Wuhan 430079, Peoples R China
[7] Sanaa Univ, Fac Engn, Sanaa 12544, Yemen
基金
国家重点研发计划;
关键词
classification; diabetic retinopathy; deep learning; CNN; ResNet; VggNet; CONVOLUTIONAL NEURAL-NETWORKS; BLOOD-VESSELS; VALIDATION;
D O I
10.3390/electronics11172740
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Diabetes is a widespread disease in the world and can lead to diabetic retinopathy, macular edema, and other obvious microvascular complications in the retina of the human eye. This study attempts to detect diabetic retinopathy (DR), which has been the main reason behind the blindness of people in the last decade. Timely or early treatment is necessary to prevent some DR complications and control blood glucose. DR is very difficult to detect in time-consuming manual diagnosis because of its diversity and complexity. This work utilizes a deep learning application, a convolutional neural network (CNN), in fundus photography to distinguish the stages of DR. The images dataset in this study is obtained from Xiangya No. 2 Hospital Ophthalmology (XHO), Changsha, China, which is very large, little and the labels are unbalanced. Thus, this study first solves the problem of the existing dataset by proposing a method that uses preprocessing, regularization, and augmentation steps to increase and prepare the image dataset of XHO for training and improve performance. Then, it takes the advantages of the power of CNN with different residual neural network (ResNet) structures, namely, ResNet-101, ResNet-50, and VggNet-16, to detect DR on XHO datasets. ResNet-101 achieved the maximum level of accuracy, 0.9888, with a training loss of 0.3499 and a testing loss of 0.9882. ResNet-101 is then assessed on 1787 photos from the HRF, STARE, DIARETDB0, and XHO databases, achieving an average accuracy of 0.97, which is greater than prior efforts. Results prove that the CNN model (ResNet-101) has better accuracy than ResNet-50 and VggNet-16 in DR image classification.
引用
收藏
页数:20
相关论文
共 62 条
[1]  
Aguiree F., 2013, IDF DIABETES ATLAS
[2]   Diabetic Retinopathy Fundus Image Classification and Lesions Localization System Using Deep Learning [J].
Alyoubi, Wejdan L. ;
Abulkhair, Maysoon F. ;
Shalash, Wafaa M. .
SENSORS, 2021, 21 (11)
[3]  
[Anonymous], 2014, DIARETDB0 Database Medical Image Understanding and Analysis
[4]  
[Anonymous], 2017, P WORKSH 32 AAAI C A
[5]  
[Anonymous], 2016, DIAB RENT DET CHALL
[6]  
[Anonymous], 2010, 2010 AAAI SPRING S S
[7]  
BAUDOIN CE, 1984, REV EPIDEMIOL SANTE, V32, P254
[8]   SURF: Speeded up robust features [J].
Bay, Herbert ;
Tuytelaars, Tinne ;
Van Gool, Luc .
COMPUTER VISION - ECCV 2006 , PT 1, PROCEEDINGS, 2006, 3951 :404-417
[9]  
Bindu C.H, 2019, INDIAN J SCI TECHNOL, V12, P3, DOI [10.17485/ijst/2018/v12i3/141597, DOI 10.17485/ijst/2018/v12i3/141597]
[10]   Causes of vision loss worldwide, 1990-2010: a systematic analysis [J].
Bourne, Rupert R. A. ;
Stevens, Gretchen A. ;
White, Richard A. ;
Smith, Jennifer L. ;
Flaxman, Seth R. ;
Price, Holly ;
Jonas, Jost B. ;
Keeffe, Jill ;
Leasher, Janet ;
Naidoo, Kovin ;
Pesudovs, Konrad ;
Resnikoff, Serge ;
Taylor, Hugh R. .
LANCET GLOBAL HEALTH, 2013, 1 (06) :E339-E349