Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
@article{210919,
author = {Monisha Birlin T and Divya C},
title = {Computerized Diagnosis of Diabetic Retinopathy using Dual Attention Mechanisms},
journal = {International Journal of Innovative Research in Technology},
year = {2026},
volume = {13},
number = {6},
pages = {2870-2876},
issn = {2349-6002},
url = {https://ijirt.org/article?manuscript=210919},
abstract = {Diabetic retinopathy (DR) is a leading cause of preventable vision loss, and computerized grading of retinal fundus photographs can support large-scale screening. DR grading is difficult because the decisive lesions (microaneurysms, haemorrhages, and exudates) are small, scattered, and unevenly distributed across severity grades. Attention mechanisms address this by teaching a convolutional neural network where to look and which features to trust. This paper reviews dual attention mechanisms for DR diagnosis, that is, designs that combine two complementary attention branches, most often channel attention ("what") and spatial attention ("where"), or class-agnostic global attention and class-specific category attention. We explain the building blocks (squeeze-and-excitation, CBAM, and the dual attention network), group ten published DR models into six design families, and compare their reported results on EyePACS, APTOS 2019, Messidor, Messidor-2, IDRiD, and DDR. In the two ablation studies reviewed, combining two attention branches improved accuracy by 1.4 to 3.1 percentage points over the better single branch. However, results are reported under different datasets, splits, and metrics; few studies test on external data, and performance falls on unseen datasets. We close with recommendations for fair evaluation: quadratic weighted kappa, patient- and duplicate-aware splits, external validation and full reporting of ablations.},
keywords = {Attention mechanism, Channel attention, Deep learning, Diabetic retinopathy, Dual attention, Fundus images, Spatial attention},
month = {November},
}
Submit your research paper and those of your network (friends, colleagues, or peers) through your IPN account, and receive 800 INR for each paper that gets published.
Join NowNational Conference on Sustainable Engineering and Management - 2024 Last Date: 15th March 2024
Submit inquiry