An Enhanced DenseNet-169 Framework for Automated Multi-Class Diabetic Retinopathy Classification Using Transfer Learning and SMOTE
International Journal of Advanced Computer Science and Applications · 2026 · 17 (10)
DOI: 10.14569/ijacsa.2026.0171036 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Advanced Computer Science and Applications, 2026. Verfasst von Sinduja A und 5 weiteren Autor:innen.
Abstract (Originalsprache)
Diabetic Retinopathy (DR) is the most common microvascular consequence of diabetes mellitus and the most common cause of permanent blindness in the working-age population worldwide. Early diagnosis and quick intervention are critical for preventing severe visual impairment, but traditional screening methods rely on skilled ophthalmologists to manually examine retinal fundus images. In this article, a robust deep learning system is proposed for the automated multi-class labelling of diabetic retinopathy based on retinal fundus images. The proposed methodology is based on the DenseNet-169 convolutional neural network architecture as a key feature extractor owing to its dense connectivity mechanism, which permits efficient feature propagation, helps feature reuse, and alleviates the vanishing gradient problem that is common in deep neural networks. To increase convergence and lessen the reliance on large medical datasets, transfer learning is applied to fine-tune DenseNet-169, which is trained with ImageNet pretrained weights. Moreover, the Synthetic Minority Over-Sampling Technique (SMOTE) is employed to handle the critical class inequity problem in retinal image datasets by establishing representative samples for minority disease classes. The proposed architecture is tested on publicly available retinal fundus image datasets with five clinically recognised phases of diabetic retinopathy. Experimental results establish that the proposed amalgamation of transfer learning with DenseNet-169 and SMOTE significantly improves classification performance and sensitivity to minority disease classifications.
Zitieren
APA
A, S., Thangavel, A. H., V, D., Selvam, H., Narmadha, A. P., & Abinaya, K. (2026). An Enhanced DenseNet-169 Framework for Automated Multi-Class Diabetic Retinopathy Classification Using Transfer Learning and SMOTE. International Journal of Advanced Computer Science and Applications, 17(10). https://doi.org/10.14569/ijacsa.2026.0171036
Vancouver
A S, Thangavel AH, V D, Selvam H, Narmadha AP, Abinaya K. An Enhanced DenseNet-169 Framework for Automated Multi-Class Diabetic Retinopathy Classification Using Transfer Learning and SMOTE. International Journal of Advanced Computer Science and Applications. 2026;17(10). doi:10.14569/ijacsa.2026.0171036
BibTeX
@article{A2026,
author = {A, Sinduja and Thangavel, Agila Harshini and V, Deepa and Selvam, Hemavathi and Narmadha, A P and Abinaya, K},
title = {An Enhanced DenseNet-169 Framework for Automated Multi-Class Diabetic Retinopathy Classification Using Transfer Learning and SMOTE},
journal = {International Journal of Advanced Computer Science and Applications},
year = {2026},
volume = {17},
number = {10},
doi = {10.14569/ijacsa.2026.0171036}
}
EUVIMED