Publication

Abstract : Background Diabetic retinopathy is a major cause of visual impairment worldwide and a growing public health concern due to the rising prevalence of diabetes. Timely and accurate detection is critical to preventing irreversible vision loss. Objective This study proposes DcareNet, a novel deep learning algorithm designed to improve the robustness, accuracy, and efficiency of diabetic retinopathy classification. Methods DcareNet was developed to automatically detect key retinal biomarkers such as microaneurysms, exudates, and abnormal blood vessels. The model integrates advanced feature extraction techniques and data optimization strategies to overcome challenges related to class imbalance, poor image quality, and limited datasets. Performance was evaluated using standard diabetic retinopathy datasets under controlled experimental conditions. Results DcareNet achieved a classification accuracy of 94.4% with an error rate of 5.58%, outperforming traditional approaches. These results highlight the mode’s reliability and potential for real-world clinical application, particularly in environments with limited ophthalmological resources. Conclusion DcareNet provides a scalable, cost-effective solution for diabetic retinopathy screening. It supports early intervention and bridges the gap between diagnostic demand and medical expertise in both developed and resource-constrained settings. The proposed approach aligns with the United Nations Sustainable Development Goals, specifically Goal 3 (Good Health and Well-being) and Goal 10 (Reduced Inequality), by promoting accessible, high-quality healthcare.