EUVIMEDEUVIMED entdecken
Crossref

Advancing Ophthalmic Diagnostics with AI: A Deep Learning Framework for Early Detection of Retinal Diseases from OCT Data

Sirin Mohammed Hejazi, Khaled Eskaf

Midocean Journal for Research and Studies · 12.06.2026 · 2 (3) · 342-365

DOI: 10.71335/vkamx642 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Midocean Journal for Research and Studies, online erschienen am 12.06.2026. Verfasst von Sirin Mohammed Hejazi und 1 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Objectives

This study aims to develop and evaluate a deep learning–based framework for the automated classification of retinal diseases using Optical Coherence Tomography (OCT) images to support early diagnosis and clinical decision-making. Methodology: A CRISP-DM–guided methodology was adopted. OCT images were preprocessed and augmented to address class imbalance and improve model generalization. Five convolutional neural network architectures—3-layer CNN, 7-layer CNN, VGG16, ResNet50, and DenseNet121—were trained and evaluated using standard performance metrics, including accuracy, precision, recall, F1-score, and AUC-ROC. Model interpretability was enhanced using Grad-CAM visual explanations. The study utilized a publicly available OCT dataset collected from international clinical sources and accessed through the Kaggle platform.

Results

The experimental results demonstrated that DenseNet121 achieved the best overall performance, with 96% accuracy and an AUC-ROC score of 0.97, while the custom 7-layer CNN also showed competitive performance. CNV was identified as the most accurately classified condition, whereas Drusen remained the most challenging category due to subtle structural retinal features. Furthermore, explain ability analysis confirmed that the model predictions were based on clinically meaningful retinal regions.

Conclusion

The proposed framework demonstrated strong potential as a clinical decision-support tool for retinal disease screening and early diagnosis. The findings highlight the effectiveness of deep learning models, particularly DenseNet121, in improving diagnostic accuracy from OCT images. Future research should focus on multi-center validation, integration of 3D OCT volumes, and uncertainty estimation techniques to enhance clinical applicability and reliability.

Zitieren

APA

Hejazi, S. M., & Eskaf, K. (2026). Advancing Ophthalmic Diagnostics with AI: A Deep Learning Framework for Early Detection of Retinal Diseases from OCT Data. Midocean Journal for Research and Studies, 2(3), 342-365. https://doi.org/10.71335/vkamx642

Vancouver

Hejazi SM, Eskaf K. Advancing Ophthalmic Diagnostics with AI: A Deep Learning Framework for Early Detection of Retinal Diseases from OCT Data. Midocean Journal for Research and Studies. 2026;2(3):342-365. doi:10.71335/vkamx642

BibTeX
@article{Hejazi2026,
  author = {Hejazi, Sirin Mohammed and Eskaf, Khaled},
  title = {Advancing Ophthalmic Diagnostics with AI: A Deep Learning Framework for Early Detection of Retinal Diseases from OCT Data},
  journal = {Midocean Journal for Research and Studies},
  year = {2026},
  volume = {2},
  number = {3},
  pages = {342--365},
  doi = {10.71335/vkamx642}
}
Kostenlos starten →