EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic Retinopathy

Diabetic Retinopathy (DR) stands as a significant factor in the prevalence of blindness on a global scale, therefore, the timely identification and precise diagnosis of this condition are imperative for enabling proper management and care. Over the past few years, diverse methods have been employed...

Full description

Saved in:
Bibliographic Details
Main Authors: Maneesha Vadduri, P. Kuppusamy
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11072398/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849430834280398848
author Maneesha Vadduri
P. Kuppusamy
author_facet Maneesha Vadduri
P. Kuppusamy
author_sort Maneesha Vadduri
collection DOAJ
description Diabetic Retinopathy (DR) stands as a significant factor in the prevalence of blindness on a global scale, therefore, the timely identification and precise diagnosis of this condition are imperative for enabling proper management and care. Over the past few years, diverse methods have been employed for automated segmentation and classification of internal retinal components, but achieving high accuracy and efficiency remains a challenging task for clinical scenarios. This research introduces an innovative method for identifying and classifying internal retinal components that combine three advanced models: a dynamic segmentation model, an efficient Elephant Herding Optimization (EHO) Model, and an autoencoder model with Q-Learning Generative Adversarial Network (Q-LGAN) process. The dynamic segmentation model accurately identifies internal retinal components by incorporating advanced methods for image processing such as vessel enhancement, multi-scale analysis, and adaptive thresholding processes. The EHO Model extracts and selects multidimensional features from the segmented images, and the autoencoder model with Q-LGAN optimizes the classification and severity identification process. The proposed approach was evaluated on the EyePACS dataset and achieved efficient results in the context of accuracy, sensitivity, and specificity levels. Compared to existing methods, our approach demonstrated significant improvements in segmentation Accuracy and feature extraction efficiency, as well as improved classification and DR severity identification levels.
format Article
id doaj-art-9efc6ea94a444cda91b43177a577a7b8
institution Kabale University
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-9efc6ea94a444cda91b43177a577a7b82025-08-20T03:27:51ZengIEEEIEEE Access2169-35362025-01-011311895911897410.1109/ACCESS.2025.358670211072398EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic RetinopathyManeesha Vadduri0https://orcid.org/0000-0002-1543-7825P. Kuppusamy1https://orcid.org/0000-0001-5369-8121School of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, IndiaSchool of Computer Science and Engineering, VIT-AP University, Amaravati, Andhra Pradesh, IndiaDiabetic Retinopathy (DR) stands as a significant factor in the prevalence of blindness on a global scale, therefore, the timely identification and precise diagnosis of this condition are imperative for enabling proper management and care. Over the past few years, diverse methods have been employed for automated segmentation and classification of internal retinal components, but achieving high accuracy and efficiency remains a challenging task for clinical scenarios. This research introduces an innovative method for identifying and classifying internal retinal components that combine three advanced models: a dynamic segmentation model, an efficient Elephant Herding Optimization (EHO) Model, and an autoencoder model with Q-Learning Generative Adversarial Network (Q-LGAN) process. The dynamic segmentation model accurately identifies internal retinal components by incorporating advanced methods for image processing such as vessel enhancement, multi-scale analysis, and adaptive thresholding processes. The EHO Model extracts and selects multidimensional features from the segmented images, and the autoencoder model with Q-LGAN optimizes the classification and severity identification process. The proposed approach was evaluated on the EyePACS dataset and achieved efficient results in the context of accuracy, sensitivity, and specificity levels. Compared to existing methods, our approach demonstrated significant improvements in segmentation Accuracy and feature extraction efficiency, as well as improved classification and DR severity identification levels.https://ieeexplore.ieee.org/document/11072398/Retinal image processingblood vesselsdiabetic retinopathyq-learning GANfeature extractionfeature selection
spellingShingle Maneesha Vadduri
P. Kuppusamy
EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic Retinopathy
IEEE Access
Retinal image processing
blood vessels
diabetic retinopathy
q-learning GAN
feature extraction
feature selection
title EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic Retinopathy
title_full EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic Retinopathy
title_fullStr EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic Retinopathy
title_full_unstemmed EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic Retinopathy
title_short EHO-Q-LGAN: An EHO-Based Q-Learning GAN for the Timely Diagnosis of Diabetic Retinopathy
title_sort eho q lgan an eho based q learning gan for the timely diagnosis of diabetic retinopathy
topic Retinal image processing
blood vessels
diabetic retinopathy
q-learning GAN
feature extraction
feature selection
url https://ieeexplore.ieee.org/document/11072398/
work_keys_str_mv AT maneeshavadduri ehoqlgananehobasedqlearningganforthetimelydiagnosisofdiabeticretinopathy
AT pkuppusamy ehoqlgananehobasedqlearningganforthetimelydiagnosisofdiabeticretinopathy