Deep Learning-Based Classification of Eye Diseases Using Convolutional Neural Network for OCT Images

Abstract

In the contemporary era, the medical industry has faced considerable challenges on diagnosing healthcare conditions due to limitations in technology and diagnostic equipment. With the advancement of computer science, cutting-edge solutions such as connected devices (IoT), cloud technologies, and AI, and Deep Learning has significantly enhanced detection of medical disorders, especially in ophthalmology. However, many diagnostic tasks are still performed manually by ophthalmologists, making the process resource-intensive and likely to result in inaccuracies due to overlapping symptoms among various retinal diseases. Although several automated systems exist, their performance often falls short of state-of-the-art accuracy. An automated, AI-driven diagnostic system is presented for identifying retinal disorders using Optical Coherence Tomography (OCT) images. This technique utilizes various techniques like attention-based mechanisms, transfer learning, and convolutional neural networks to enhance the accuracy of classification. The model utilizes various techniques to classify images correctly as Choroidal Neovascularization (CNV), Diabetic Macular Edema (DME), Age-Related Macular Degeneration (AMD), etc. From the performance assessment, it can be seen that the proposed approach achieves accuracy rates of 99.08% during training, 97.52% during validation, and 97.39% during the test phase. Using a publicly available image dataset of OCT images, the proposed system can classify retinal images into four classes: Normal, Diabetic Macular Edema (DME), Choroidal Neovascularization (CNV), and Age-Related Macular Degeneration (AMD). The good accuracy of the proposed system indicates the potential of the proposed model to assist doctors in the diagnosis of retinal diseases such as Diabetic Retinopathy.

Introduction

Visual impairment is rapidly increasing worldwide due to various ocular diseases such as Diabetic Retinopathy, Age-Related Macular Degeneration, and conditions like CNV and DME, which are major causes of blindness. Factors such as aging, diabetes, excessive screen use, genetics, and environmental exposure contribute significantly. Global data highlights the growing burden, with millions affected and many cases remaining undiagnosed despite the fact that early detection could prevent most vision loss.

Modern diagnostic imaging techniques, especially Optical Coherence Tomography, play a crucial role in detecting retinal diseases by providing detailed, non-invasive images. The integration of AI, Machine Learning, and Deep Learning has further improved diagnostic accuracy, efficiency, and cost-effectiveness, supporting clinicians in early detection and treatment.

The research focuses on using Convolutional Neural Networks (CNNs) to classify retinal diseases from OCT images. Prior studies demonstrate that deep learning models—particularly CNNs with transfer learning—achieve high accuracy (often above 90%) in detecting diseases like DR, though challenges such as limited datasets, class imbalance, and high computational cost remain.

The proposed methodology uses a large dataset of over 84,000 OCT images categorized into CNV, DME, Drusen, and normal classes. By leveraging AI-driven analysis, the system aims to enable accurate, automated, and early diagnosis of ocular diseases, ultimately improving clinical decision-making and reducing the burden on healthcare systems.

Conclusion

The study used an AI-based algorithm to detect the three major eye problems that are prevalent worldwide: Diabetic Macular Edema (DME), Choroidal Neovascularization (CNV), and Drusen, which are major causes of visual impairment worldwide. To reduce the rising incidence of these eye problems, screening and testing are the way forward. The proposed VGG16_Attention model combines the attention mechanism with deep convolutional neural networks (CNNs) and transfer learning (TL) techniques. The proposed approach is able to classify OCT scans with high accuracy without compromising on competitive performance, without the need for deep learning hardware or large datasets. The experimental results demonstrate that the proposed model effectively and accurately classifies OCT images of ocular diseases, indicating its potential for automated eye condition diagnosis. Furthermore, incorporating OCT scans from multiple device manufacturers in the training and testing datasets could help develop a more generalizable and universally applicable diagnostic model. While the proposed system provides a cost-effective and efficient approach for detecting various eye conditions, it is not intended to serve as a fully autonomous or infallible diagnostic tool. Given the critical importance of patient safety, the system should be used as a decision-support aid, with final medical judgments made by qualified ophthalmologists or healthcare professionals.

References

[1] A. Moraru, D. Costin, R. Moraru, and D. Branisteanu, “Artificial intelligence and deep learning in ophthalmology—present and future (review),” Experimental and Therapeutic Medicine, 2020. doi: 10.3892/etm.2020.9118. [2] D. S. W. Ting et al., “Artificial intelligence and deep learning in ophthalmology,” British Journal of Ophthalmology, vol. 103, no. 2, pp. 167–175, 2019. doi: 10.1136/bjophthalmol-2018-313173. [3] A. R. Ran et al., “Deep learning in glaucoma with optical coherence tomography: a review,” Eye, vol. 35, no. 1, pp. 188–201, 2021. doi: 10.1038/s41433-020-01191-5. [4] A. J. Paul, “Advances in classifying the stages of diabetic retinopathy using convolutional neural networks in low memory edge devices,” in Proc. MASCON, 2021, pp. 1–8. doi: 10.1109/mascon51689.2021.9563584. [5] D. S. Kermany et al., “Identifying medical diagnoses and treatable diseases by image-based deep learning,” Cell, vol. 172, no. 5, pp. 1122–1131.e9, 2018. doi: 10.1016/j.cell.2018.02.010. [6] U. Ishtiaq et al., “Diabetic retinopathy detection through artificial intelligent techniques: a review and open issues,” Multimedia Tools and Applications, vol. 79, no. 21–22, pp. 15209–15252, 2020. doi: 10.1007/s11042-018-7044-8. [7] V. Lakshminarayanan, H. Kheradfallah, A. Sarkar, and J. J. Balaji, “Automated detection and diagnosis of diabetic retinopathy: a comprehensive survey,” Journal of Imaging, 2021. doi: 10.3390/jimaging7090165. [8] B. K. Swenor, M. J. Lee, V. Varadaraj, H. E. Whitson, and P. Y. Ramulu, “Aging with vision loss: a framework for assessing the impact of visual impairment on older adults,” The Gerontologist, vol. 60, no. 6, pp. 989–995, 2020. doi: 10.1093/geront/gnz117. [9] Z. L. Teo et al., “Global prevalence of diabetic retinopathy and projection of burden through 2045: systematic review and meta-analysis,” Ophthalmology, vol. 128, no. 11, pp. 1580–1591, 2021. doi: 10.1016/j.ophtha.2021.04.027. [10] World Health Organization, World Report on Vision, 2019. [Online]. Available: https://www.who.int/publications/i/item/9789241516570 [11] B. Correspondent, “80–90% of blindness cases in India are preventable: experts,” Biovoice News, 2018. [Online]. Available: https://www.biovoicenews.com/80-90-of-blindness-cases-in-india-are-preventable-experts/ [12] R. Nuzzi, G. Boscia, P. Marolo, and F. Ricardi, “The impact of artificial intelligence and deep learning in eye diseases: a review,” Frontiers in Medicine, vol. 8, 2021. doi: 10.3389/fmed.2021.710329. [13] K. Alsaih et al., “Machine learning techniques for diabetic macular edema (DME) classification on SD-OCT images,” Biomedical Engineering Online, vol. 16, no. 1, pp. 1–12, 2017. doi: 10.1186/s12938-017-0352-9. [14] N. M. Al-Moosawi and R. S. Khudeyer, “ResNet-34/DR: a residual convolutional neural network for the diagnosis of diabetic retinopathy,” Informatica, vol. 45, no. 7, pp. 115–124, 2021. doi: 10.31449/inf. v45i7.3774. [15] H. Tariq et al., “Performance analysis of deep-neural-network-based automatic diagnosis of diabetic retinopathy,” Sensors, vol. 22, no. 1, pp. 1–15, 2022. doi: 10.3390/s22010205. [16] P. M. Burlina et al., “Automated grading of age-related macular degeneration from color fundus images using deep convolutional neural networks,” JAMA Ophthalmology, vol. 135, no. 11, pp. 1170–1176, 2017. doi: 10.1001/jamaophthalmol.2017.3782. [17] V. Gulshan et al., “Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs,” JAMA, vol. 316, no. 22, pp. 2402–2410, 2016. doi: 10.1001/jama.2016.17216. [18] U. Schmidt-Erfurth et al., “Artificial intelligence in retina,” Progress in Retinal and Eye Research, 2018. doi: 10.1016/j.preteyeres.2018.07.004. [19] M. D. Abramoff, M. K. Garvin, and M. Sonka, “Retinal imaging and image analysis,” IEEE Reviews in Biomedical Engineering, vol. 3, pp. 169–208, 2010. doi: 10.1109/RBME.2010.2084567. [20] ?. ??lu, M. ??lu, S. Giovanzana, and R. D. Shah, “The history and use of optical coherence tomography in ophthalmology,” Human and Veterinary Medicine, vol. 3, no. 1, pp. 29–32, 2011. [21] A. Ran and C. Y. Cheung, “Deep learning-based optical coherence tomography and optical coherence tomography angiography image analysis: an updated summary,” Asia-Pacific Journal of Ophthalmology, vol. 10, no. 3, pp. 253–260, 2021. doi: 10.1097/APO.0000000000000405. [22] S. S. M. Sheet et al., “Retinal disease identification using upgraded CLAHE filter and transfer convolution neural network,” ICT Express, 2021. doi: 10.1016/j.icte.2021.05.002. [23] A. Kulkarni, D. Chong, and F. A. Batarseh, “Foundations of data imbalance and solutions for a data democracy,” in Data Democracy, Amsterdam: Elsevier, 2020, pp. 83–106. [24] J. Han, “The design of diabetic retinopathy classifier based on parameter optimization SVM,” in Proc. ICIIBMS, 2018. doi: 10.1109/ICIIBMS.2018.8549947. [25] S. Wan, Y. Liang, and Y. Zhang, “Deep convolutional neural networks for diabetic retinopathy detection by image classification,” Computers & Electrical Engineering, vol. 72, pp. 274–282, 2018. doi: 10.1016/j.compeleceng.2018.07.042. [26] M. Jena, S. P. Mishra, and D. Mishra, “Detection of diabetic retinopathy images using a fully convolutional neural network,” in Proc. ICDSBA, 2018, pp. 523–527. doi: 10.1109/ICDSBA.2018.00103. [27] M. U. Rehman, S. H. Khan, Z. Abbas, and S. M. Danish Rizvi, “Classification of diabetic retinopathy images based on customized CNN architecture,” in Proc. AICAI, 2019, pp. 244–248. doi: 10.1109/AICAI.2019.8701231. [28] F. J. Martinez-Murcia et al., “Deep residual transfer learning for automatic diagnosis and grading of diabetic retinopathy,” Neurocomputing, vol. 452, pp. 424–434, 2021. doi: 10.1016/j.neucom.2020.04.148. [29] R. Chopra, S. K. Wagner, and P. A. Keane, “Optical coherence tomography in the 2020s—outside the eye clinic,” Eye, vol. 35, no. 1, pp. 236–243, 2021. doi: 10.1038/s41433-020-01263-6. [30] A. K. Gangwar and V. Ravi, Diabetic Retinopathy Detection Using Transfer Learning and Deep Learning, vol. 1176, Singapore: Springer, 2021. [31] ADCIS, “Messidor Dataset,” [Online]. Available: https://www.adcis.net/en/third-party/messidor/ [32] Asia Pacific Tele-Ophthalmology Society, “APTOS 2019 Blindness Detection,” 2020. [Online]. Available: https://www.kaggle.com/c/aptos2019-blindness-detection [33] F. A. Medeiros, A. A. Jammal, and A. C. Thompson, “An OCT-trained deep learning algorithm for objective quantification of glaucomatous damage in fundus photographs,” Ophthalmology, vol. 126, no. 4, pp. 513–521, 2019. doi: 10.1016/j.ophtha.2018.12.033. [34] D. Le et al., “Transfer learning for automated OCTA detection of diabetic retinopathy,” Translational Vision Science & Technology, vol. 9, no. 2, pp. 1–9, 2020. doi: 10.1167/tvst.9.2.35. [35] P. Chowdhury et al., “Transfer learning approach for diabetic retinopathy detection using efficient network with two-phase training,” in Proc. I2CT, 2021, pp. 1–6. doi: 10.1109/I2CT51068.2021.9418111. [36] California Healthcare Foundation, “Diabetic Retinopathy Detection,” 2015. [Online]. Available: https://www.kaggle.com/c/diabetic-retinopathy-detection [37] G. An et al., “Hierarchical deep learning models using transfer learning for disease detection and classification based on small number of medical images,” Scientific Reports, vol. 11, pp. 1–9, 2021. doi: 10.1038/s41598-021-83503-7. [38] K. R. A. Kumar, P. M. Megha, and K. Meenakshy, “Diabetic retinopathy detection & classification techniques: a review,” International Journal of Scientific & Technology Research, vol. 9, no. 3, pp. 1621–1628, 2020. [39] A. Samanta et al., “Automated detection of diabetic retinopathy using convolutional neural networks on a small dataset,” Pattern Recognition Letters, vol. 135, pp. 293–298, 2020. doi: 10.1016/j.patrec.2020.04.026. [40] S. Albahli et al., “Recognition and detection of diabetic retinopathy using DenseNet-65 based Faster-RCNN,” Computer Modeling in Engineering & Sciences, vol. 67, no. 2, pp. 1333–1351, 2021. doi: 10.32604/cmc.2021.014691. [41] S. I. Pao et al., “Detection of diabetic retinopathy using bichannel convolutional neural network,” Journal of Ophthalmology, 2020. doi: 10.1155/2020/9139713. [42] R. S. Salvi et al., “Predictive analysis of diabetic retinopathy with transfer learning,” in Proc. ICNET, 2021. doi: 10.1109/ICNTE51185.2021.9487789. [43] R. S. Rajkumar et al., “Transfer learning approach for diabetic retinopathy detection using residual network,” in Proc. ICICT, 2021, pp. 1189–1193. doi: 10.1109/ICICT50816.2021.9358468. [44] Y. S. Boral and S. S. Thorat, “Classification of diabetic retinopathy based on hybrid neural network,” in Proc. ICCMC, 2021, pp. 1354–1358. doi: 10.1109/ICCMC51019.2021.9418224. [45] A. Bhowmik, S. Kumar, and N. Bhat, Eye Disease Prediction from OCT Images with Transfer Learning, vol. 1000, Cham: Springer, 2019. [46] M. Shelar et al., “Detection of diabetic retinopathy and its classification from fundus images,” in Proc. ICCCI, 2021, pp. 3–8. doi: 10.1109/ICCCI50826.2021.9402347. [47] W. Lu et al., “Deep learning-based automated classification of multi-categorical abnormalities from optical coherence tomography images,” Translational Vision Science & Technology, 2018. doi: 10.1167/tvst.7.6.41. [48] K. M. Hasan et al., “Cataract disease detection by using transfer learning-based intelligent methods,” Computational and Mathematical Methods in Medicine, 2021. doi: 10.1155/2021/7666365. [49] Larxel, “Ocular disease recognition,” 2020. [Online]. Available: https://www.kaggle.com/andrewmvd/ocular-disease-recognition-odir5k [50] D. Kermany, K. Zhang, and M. Goldbaum, “Large dataset of labeled OCT and chest X-ray images,” 2018. [Online]. Available: https://data.mendeley.com/datasets/rscbjbr9sj/3 [51] T. J. Perumanoor, “What is VGG16? Introduction to VGG16, 2021. [Online]. Available: https://medium.com/@mygreatlearning/what-is-vgg16-introduction-to-vgg16-f2d63849f615 [52] P. Varshney, “VGGNet-16 architecture: a complete guide,” 2019. [Online]. Available: https://www.kaggle.com/blurredmachine/vggnet-16-architecture-a-complete-guide [53] S. Narkhede, “Understanding confusion matrix,” 2018. [Online]. Available: https://towardsdatascience.com/understanding-confusion-matrix-a9ad42dcfd62 [54] P. Singh, N. Singh, K. K. Singh, and A. Singh, “Diagnosing disease using machine learning,” in Machine Learning and the Internet of Medical Things in Healthcare, Amsterdam: Elsevier, 2021, pp. 89–111.

Copyright

Copyright © 2026 Piyush Verma, Mahak , Garima Yadav, Hari Om, Harendra Singh. This 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.