Benchmarking CNN Architectures for Early Glaucoma Detection from Retinal Fundus Images Using Deep Learning Techniques
International Journal of Mathematics, Statistics, and Computing · 10.06.2026 · 4 (2) · 71-79
DOI: 10.46336/ijmsc.v4i2.321 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Mathematics, Statistics, and Computing, online erschienen am 10.06.2026. Verfasst von Riza Andrian Ibrahim und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Glaucoma is a progressive optic nerve disease and one of the leading causes of irreversible blindness worldwide. Early detection is essential to prevent vision loss; however, traditional diagnostic methods often require specialized expertise and are not always accessible. This study proposes a deep learning-based approach for automated glaucoma detection using retinal fundus images. Three CNN architectures—ResNet-50, VGG-19, and EfficientNet-B0—were implemented and evaluated to determine the most effective model. The research methodology includes dataset preparation, image preprocessing, model training, and performance evaluation using multiple classification metrics. The dataset consists of labeled retinal fundus images categorized into glaucoma and non-glaucoma classes. Preprocessing techniques such as resizing and data augmentation were applied to improve model generalization. Each model was trained under the same experimental conditions to ensure a fair comparison. The results show that EfficientNet-B0 achieves the best performance among the evaluated models, with the highest accuracy, precision, recall, and F1-score, as well as the lowest loss and fastest training time. Visualization through confusion matrix and ROC curve analysis further confirms its superior classification capability and robustness in distinguishing between classes. In contrast, VGG-19 demonstrates the lowest performance, while ResNet-50 provides moderate results. In conclusion, EfficientNet-B0 is identified as the most reliable and efficient model for automated glaucoma detection. This study highlights the potential of deep learning in improving early diagnosis and supporting accessible, scalable ophthalmology screening systems.
Zitieren
APA
Ibrahim, R. A., & Setyawan, D. P. (2026). Benchmarking CNN Architectures for Early Glaucoma Detection from Retinal Fundus Images Using Deep Learning Techniques. International Journal of Mathematics, Statistics, and Computing, 4(2), 71-79. https://doi.org/10.46336/ijmsc.v4i2.321
Vancouver
Ibrahim RA, Setyawan DP. Benchmarking CNN Architectures for Early Glaucoma Detection from Retinal Fundus Images Using Deep Learning Techniques. International Journal of Mathematics, Statistics, and Computing. 2026;4(2):71-79. doi:10.46336/ijmsc.v4i2.321
BibTeX
@article{Ibrahim2026,
author = {Ibrahim, Riza Andrian and Setyawan, Deva Putra},
title = {Benchmarking CNN Architectures for Early Glaucoma Detection from Retinal Fundus Images Using Deep Learning Techniques},
journal = {International Journal of Mathematics, Statistics, and Computing},
year = {2026},
volume = {4},
number = {2},
pages = {71--79},
doi = {10.46336/ijmsc.v4i2.321}
}
EUVIMED