Identification of the Diabetic Retinopathy Using ResNet-18

In the realm of healthcare, trauma, aging, and diseases like diabetes can cause a variety of abnormalities in human eyes. Diabetic retinopathy stands out as a prevalent global cause of blindness. Early identification and diagnosis of diabetic retinopathy is essential for prompt treatment and blindness prevention. The analysis of medical images can contribute to the identification of diabetic retinopathy, assisting in the diagnostic process. This study uses a deep learning (DL) model, ResNet-18, to identify diabetic retinopathy. The dataset used in this study acquired from Kaggle is divided into training and testing. ResNet-18 accomplished a training accuracy of 99.91% and a testing accuracy of 96.65%. The findings demonstrate the effectiveness of utilizing DL models like ResNet-18 for the early identification of diabetic retinopathy, potentially revolutionizing the screening process in clinical settings. By streamlining the diagnostic process, this proposed model has the potential to significantly enhance early intervention and treatment strategies, ultimately mitigating the risk of blindness in diabetic patients. Incorporating such technology into the healthcare system could improve patient outcomes and resource allocation.