Super-Resolution Image Generation for Diabetic Retinopathy Detection by SRGAN

Yi Zhao

2025

Abstract

As computer vision technology progresses, the Super-Resolution method is essential in medical image enhancement. In this article, Super-Resolution Generative Adversarial Network (SRGAN) is trained to produce high-resolution diabetic retinopathy images, aiming to assist numerous model training processes such as U-net and ResU-net. As a result of improving the original SRGAN framework, the resolution and quality of images reach a higher level, capturing more detailed information. Through this way, segmentation models can more accurately determine the location of lesions and tumor nodules, which enables early disease prediction and precise localization. Nowadays, many advanced segmentation studies work with high-resolution processing of medical images. The experiment results indicate that SRGAN has commendable efficacy in the APTOS-2019 dataset, achieving PSNR-43 SSIM-0.93, Precision-0.965, Recall-0.913, F1-Score-0.937, which demonstrates its superiority in detail restoration. SRGAN provides strong support for subsequent disease detection tasks, definitely facilitating more accurate diagnostic outcomes and ascending the reliability of medical image analysis.

Paper Citation

in Harvard Style

Zhao Y. (2025). Super-Resolution Image Generation for Diabetic Retinopathy Detection by SRGAN. In Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE; ISBN 978-989-758-765-8, SciTePress, pages 400-403. DOI: 10.5220/0013698200004670

in Bibtex Style

@conference{icdse25,
author={Yi Zhao},
title={Super-Resolution Image Generation for Diabetic Retinopathy Detection by SRGAN},
booktitle={Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE},
year={2025},
pages={400-403},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013698200004670},
isbn={978-989-758-765-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Data Science and Engineering - Volume 1: ICDSE
TI - Super-Resolution Image Generation for Diabetic Retinopathy Detection by SRGAN
SN - 978-989-758-765-8
AU - Zhao Y.
PY - 2025
SP - 400
EP - 403
DO - 10.5220/0013698200004670
PB - SciTePress