Multiclass Classification of Retinal Diseases from OCT Images Using ResNet50-Based Transfer Learning

Septia Arifta; Tanaya Anindita Irawan; Ade Rizky Darmawan; Aviolla Terza Damaliana; Shindi Shella May Wara5

doi:10.59934/jaiea.v5i3.2445

Authors

Septia Arifta Universitas Pembangunan Nasional Veteran Jawa Timur
Tanaya Anindita Irawan
Ade Rizky Darmawan
Aviolla Terza Damaliana
Shindi Shella May Wara5 Universitas Pembangunan Nasional Veteran Jawa Timur

DOI:

https://doi.org/10.59934/jaiea.v5i3.2445

Keywords:

deep learning, multiclass classification, ResNet50, retinal disease, transfer learning

Abstract

Retinal diseases are among the leading causes of visual impairment and blindness when not detected at an early stage. Advances in artificial intelligence, particularly deep learning, provide opportunities to support automated retinal disease diagnosis using Optical Coherence Tomography (OCT) images. This study aims to develop a multiclass retinal disease classification model using ResNet50-based transfer learning on the Retinal OCT C8 dataset, which consists of eight retinal condition categories: Age-related Macular Degeneration (AMD), Choroidal Neovascularization (CNV), Central Serous Retinopathy (CSR), Diabetic Macular Edema (DME), Diabetic Retinopathy (DR), Drusen, Macular Hole (MH), and Normal. The research stages included data preprocessing, image augmentation, transfer learning-based model training, and fine-tuning of the final network layers. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics. Experimental results demonstrated that the proposed ResNet50 model achieved an accuracy of 93%, precision of 93%, recall of 93%, and F1-score of 93% on the testing dataset. These findings indicate that ResNet50 is effective in identifying multiple retinal diseases from OCT images. The proposed approach has potential applications in computer-aided diagnostic systems to assist clinicians in performing faster and more accurate retinal disease screening and early detection.

Downloads

Download data is not yet available.

References

S. V. Viraktamath, K. B. Kadakuntla, and S. Begumpur, “Detection of Diabetic Maculopathy,” in Proceedings of the International Conference on Artificial Intelligence and Smart Computing (ICAISC), IEEE, 2023. doi: 10.1109/ICAISC58445.2023.10199599.

N. Eladawi et al., “Optical Coherence Tomography: A Review,” in Artificial Intelligence in Retina, Academic Press, 2020, pp. 191–221. doi: 10.1016/B978-0-12-817440-1.00007-3.

A. Awasthi, S. B. Rao, and K. Acharjya, “Efficient Image Feature Extraction Using Convolutional Neural Networks,” in Proceedings of the International Conference on Networking and Advanced Technologies (ICoNAT), IEEE, 2024, pp. 1–5. doi: 10.1109/ICONAT61936.2024.10775088.

P. Dey and P. Dey, “Transfer Learning: Leveraging Pre-trained Knowledge for Efficient Models in Deep Learning,” in Advances in Artificial Intelligence and Machine Learning, Springer, 2025, pp. 251–261. doi: 10.1007/978-981-95-3272-8_17.

A. Puspitasari, D. S. Salsabila, and D. Roliawati, “Application of ResNet50-CNN for Classification Optimization on Fashion Datasets,” Indones. J. Data Sci., vol. 3, no. 1, pp. 1–12, 2025, doi: 10.30989/ijds.v3i1.1533.

L. Xingyu, “ResNet: Enabling Deep Convolutional Neural Networks Through Residual Learning,” ArXiv Prepr., 2025, doi: 10.48550/arXiv.2510.24036.

H. Çetiner and S. Metlek, “Analysis of Different Pooling Functions on a Convolutional Neural Network-Based Model,” Int. J. 3D Print. Technol. Digit. Ind., vol. 8, no. 2, pp. 266–276, 2024, doi: 10.46519/ij3dptdi.1484354.

I. L. Kharisma, S. Susanti, R. Rustiani, R. A. Pratama, and K. Kamdan, “Implementation of the VGG19 Model with Transfer Learning for Retinal Disease Diagnosis: A Study on Normal Eyes, Diabetic Retinopathy, Cataract, and Glaucoma Datasets,” Eng. Proc., vol. 107, no. 1, p. 111, 2025, doi: 10.3390/engproc2025107111.

Y. Xia and J. Sun, “Basic Performance Metrics for Machine Learning Models,” in Machine Learning Fundamentals and Applications, CRC Press, 2026, pp. 485–496. doi: 10.1201/9781003610281-15.

A. Shaykhulova, “Model Performance on Domain Test Set,” 2026, doi: 10.60797/irj.2026.163.75.3.

S. Simhadri, M. Ponnam, R. Rajitha, and R. Balamurugan, “Enhanced Multi-Class Model Evaluation: Analyzing BERT, GPT-2, and LLaMA with Precision, Recall, and F1-Score Metrics,” in Proceedings of the International Conference on Intelligent Machine Intelligence and Applications (ICIMIA), IEEE, 2025, pp. 984–989. doi: 10.1109/ICIMIA67127.2025.11200914.