You're viewing our updated article page. If you need more time to adjust, you can return to the old layout.

ORIGINAL RESEARCH article

Front. Med.

Sec. Ophthalmology

Enhancing Fundus Image Analysis for Diabetic Retinopathy Using CheXNet with CBAM and Grad-CAM Visualization

  • 1. Yarmouk University, Irbid, Jordan

  • 2. Jordan University of Science and Technology, Irbid, Jordan

  • 3. Imam Muhammad Ibn Saud Islamic University, Riyadh, Saudi Arabia

Article metrics

View details

272

Views

The final, formatted version of the article will be published soon.

Abstract

Diabetic retinopathy (DR) is a leading cause of vision impairment among people with diabetes; therefore, early detection and the right diagnosis are necessary to save sight. Unfortunately, making the models for the complex fundus images' understanding and diagnosis is still a major challenge. In the paper, a new method is proposed for the diabetic retinopathy diagnosis from fundus images employing the hybrid model named CheXNet_CBAM. The model is an upgrade of the DenseNet121 architecture. For better feature representation, a convolutional block attention module (CBAM) is added to the CheXNet model. This module integrates both channel and spatial attention, allowing the network to put more attention on the retinal locations that are clinically significant. The Grad-CAM technique is also employed to produce the heat maps, which show the areas of the fundus images that are most important for classification. The proposed model is evaluated against deep learning models (CheXNet, DenseNet121, MobileNetV2, VGG19, and ResNet50). The suggested CheXNet_CBAM model surpasses the other models in diabetic retinopathy diagnosis from fundus images with a remarkably high accuracy of 96.12% over the APTOS-2019 dataset and 96.33% for the DDR dataset. CheXNet_CBAM not only offers a more reliable diagnostic foundation for doctors, but also its excellent diabetic retinopathy classification skills can significantly help patients with early detection and treatment of the ailment. This presents a remarkable clinical impact in the diabetic retinopathy diagnosis area.

Summary

Keywords

deep learning, Diabetic Retinopathy, Fundus imaging, Grad-CAM, image classification

Received

25 October 2025

Accepted

06 February 2026

Copyright

© 2026 Al-Dolat, Alhatamleh, Alqudah, Alhazimi, Amin, Daamseh, Madain, Malkawi, Al- Omari, Almarek and aljefri. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Amro Alhazimi

Disclaimer

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Outline

Share article

Article metrics