Skip to content
Conference Open access

Ocularai: an Explainable and Confidence-Aware Deep Learning Framework for Multi-Class Retinal Disease Detection Using Fundus Imaging

Sep 2026 · 2026 International Conference on Sustainability, Innovation & Technology (ICSIT) · pp. 1-6 · 0 citations · 36 references

Abstract

Some of the major causes of permanent vision loss worldwide remain glaucoma, diabetic retinopathy and age-related macular degeneration. However, in developing countries such as India where the incidence of such diseases is the highest, access to specialized expert ophthalmological care is limited. Early detection is essential to avoid permanent visual impairment however, a lack of training and infrastructure to support these efforts are major challenges facing population health. The objective of this study was to introduce an automated deep learning framework, namely OcularAI, that classifies retinopathy in multi-class manner using the already existing fundus image databases. We use a customized and well-trained according to the transfer learning organic growth of an EfficientNet-B0 architecture trained on a carefully selected subset of 6,350 retinal images from seven diagnostic classes. For improving clinical interpretability, Grad-CAM was introduced to visualize diagnostically pertinent areas in retinal images. Furthermore, the classification is supported by a confidence-aware prediction mechanism and uncertain predictions are marked as low confident in case it is used in real-world scenarios for increased reliability. Experimental results show good generalization and are only indicated by the best validation accuracy of 89.21% with not much overfitting. It is being used as an internet-based program, which enables screening and decision support in real-time. The proposed approach emphasizes the importance of EAI, which also results in a bridge to automated diagnosis and clinical trust in ophthalmology.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.