Abstract
This study investigates the application of deep learning algorithms, specifically convolutional neural networks (CNNs), in the early detection and classification of diabetic retinopathy from retinal fundus images. Our model achieved an overall accuracy of 94.3% and an AUC-ROC of 0.97.
This study investigates the application of deep learning algorithms, specifically convolutional neural networks (CNNs), in the early detection and classification of diabetic retinopathy from retinal fundus images. We developed and validated a multi-class classification model trained on a dataset of 15,000 annotated retinal images across five severity levels. Our model achieved an overall accuracy of 94.3% and an AUC-ROC of 0.97, outperforming existing clinical screening benchmarks.