A Comprehensive Review of CNN-Based Framework for Multi-Sign Detection of Diabetic Retinopathy in Fundus Images Using Public Datasets

Notice

This is an unedited manuscript accepted for publication and provided as an Article in Press for early access at the author’s request. The article will undergo copyediting, typesetting, and galley proof review before final publication. Please be aware that errors may be identified during production that could affect the content. All legal disclaimers of the journal apply.

Year : 2025 | Volume : 14 | 03 | Page :
    By

    Umesh Kumar,

  1. Assistant Professor,, Computer Engineering Department, JC Bose University of Science and Technology, YMCA, Faridabad,, Haryana, India

Abstract

document.addEventListener(‘DOMContentLoaded’,function(){frmFrontForm.scrollToID(‘frm_container_abs_203714’);});Edit Abstract & Keyword

Diabetic retinopathy (DR) is one of the main causes of vision impairment. Blindness prevention and
effective treatment depend on early detection. A thorough deep learning-based framework for the
automatic segmentation and simultaneous detection of exudates, hemorrhages, and microaneurysms –
three important DR indicators – from retinal fundus images is presented in this work. These three
pathological signs’ corresponding annotated image patches, along with background (no-sign) areas,
were used to train a ten-layer convolutional neural network (CNN). A post-processing algorithm is used
to improve localization and remove noise from the probability maps that the network produces for each
class. Receiver operating characteristic (ROC) curve analysis was utilized to determine the optimal
classification thresholds. The proposed model was evaluated on two publicly available datasets, one
for training and another for testing. It uses both patch-based and full-image analysis. To ensure
reliability, the model’s performance was tested across ten independent runs, consistently demonstrating
high accuracy and reliable detection of all DR indicators. Furthermore, the approach was compared
against several advanced models and techniques, including the tandem pulse coupled neural network
(TPCNN), deep learning-based support vector machine (DLBSVM), synergic deep learning (SDL), and
lesion-aware CNN (LACNN), using datasets, such as Messidor, DRIVE, CHASE_DB1, and OCT image
databases. The proposed system exhibited competitive or superior results in terms of accuracy,
specificity, and sensitivity. It confirms its potential as an effective and dependable tool for
ophthalmologists in clinical DR screening.

Keywords: Diabetic retinopathy, fundus images, CNN, DR

How to cite this article:
Umesh Kumar. A Comprehensive Review of CNN-Based Framework for Multi-Sign Detection of Diabetic Retinopathy in Fundus Images Using Public Datasets. Research and Reviews : Journal of Computational Biology. 2025; 14(03):-.
How to cite this URL:
Umesh Kumar. A Comprehensive Review of CNN-Based Framework for Multi-Sign Detection of Diabetic Retinopathy in Fundus Images Using Public Datasets. Research and Reviews : Journal of Computational Biology. 2025; 14(03):-. Available from: https://journals.stmjournals.com/rrjocb/article=2025/view=0


document.addEventListener(‘DOMContentLoaded’,function(){frmFrontForm.scrollToID(‘frm_container_ref_203714’);});Edit

References

Jebaseeli TJ, Durai CAD, Peter JD. Retinal blood vessel segmentation from diabetic retinopathy
images using tandem PCNN model and deep learning based SVM. Optik. 2020;199:163328.
2. Shankar K, Wahab Sait AR, Gupta D, Lakshmanaprabu SK, Khanna A, Pandey HM. Automated
detection and classification of fundus diabetic retinopathy images using synergic deep learning
model. Pattern Recognit Lett. 2020;133:210–6.
3. Srivastava R, Duan L, Wong DWK, Liu J, Wong TY. Detecting retinal microaneurysms and
hemorrhages with robustness to the presence of blood vessels. Comput Methods Programs Biomed.
2017;138:83–91.
4. Li X, Shen L, Shen M, Tan F, Qiu CS. Deep learning based early stage diabetic retinopathy
detection using optical coherence tomography. Neurocomputing. 2019;369:134–44.
5. Pires R, Avila S, Wainer J, Valle E, Abramoff MD, Rocha A. A data-driven approach to referable
diabetic retinopathy detection. Artif Intell Med. 2019;96:93–106.
6. Adem K. Exudate detection for diabetic retinopathy with circular Hough transformation and
convolutional neural networks. Expert Syst Appl. 2018;114:289–95.
7. Alyoubi WL, Shalash WM, Abulkhair MF. Diabetic retinopathy detection through deep learning
techniques: A review. Inform Med Unlocked. 2020;20:100377.
8. Lee SH, Perez ED, Brown MS. A survey of publicly available retinal image datasets for diabetic
retinopathy analysis. Biomed Signal Process Control. 2021;66:102409.
9. Sánchez C, Hornero R, López MI, Poza J. A novel automatic image processing algorithm for
detection of hard exudates based on retinal image analysis. Med Eng Phys. 2003;25(8):735–42.
10. Giancardo R, Meriaudeau F, Karnowski TP, Li Y, Tobin SS, Chaum E. Exudate-based diabetic
macular edema detection in fundus images using publicly available datasets. Med Image Anal.
2012;16(1):216–26.
11. Fraz MM, Jahangir W, Zahid S, Hamayun MM, Barman SA. Multiscale segmentation of exudates
in retinal images using contextual cues and ensemble classification. Biomed Signal Process Control.
2017;35:50–62.
12. Tang L, Ni B, Luo W, Wang J, Yan S. Detecting diabetic retinopathy with deep learning by dividing
fundus images into splats. Comput Med Imaging Graph. 2017;55:106–13.
13. Walter T, Massin P, Erginay A, Ordonez R, Jeulin C, Klein JC. Automatic detection of
microaneurysms in color fundus images. Med Image Anal. 2007;11(6):555–66.
14. van Grinsven F, van Ginneken G, ter Haar Romeny BM, Sánchez CI. Fast convolutional neural
network training using selective data sampling: Application to hemorrhage detection in color
fundus images. IEEE Trans Med Imaging. 2016;35(5):1273–84.
15. Shan J, Li L. A deep learning method for microaneurysm detection in fundus images. In: IEEE
International Conference on Image Processing (ICIP); 2016. p. 4320–4.
16. Agurto C, Murray V, Barriga E, Murillo S, Davis H, Pattichis PS, et al. Multiscale AM-FM methods
for diabetic retinopathy lesion detection. IEEE Trans Med Imaging. 2010;29(2):502–12.
17. Tan JH, Acharya UR, Subbhuraam VB, Lim CM, Ng KC, Krishnan SV. Automated diagnosis of
diabetic retinopathy using higher order spectra features and deep learning. Comput Methods
Programs Biomed. 2015;122(2):113–20.

18. Tan H, Tan YH, Chang CY. Patch-based detection and classification for diabetic retinopathy
lesions. In: IEEE EMBS Conf Biomed Eng Sci (IECBES); 2016. p. 512–7.

Ahead of Print Subscription Original Research
Volume 14
03
Received 18/05/2025
Accepted 14/07/2025
Published 25/08/2025
Publication Time 99 Days
64

[first_name] [last_name]

My IP

PlumX Metrics