Developing a Deep Learning Methodology to Anticipate the Onset of Diabetic Retinopathy at an Early Stage
摘要
Diabetic Eye Disease (DED) is a common condition in diabetic individuals, leading to vision loss. Detecting DED early is crucial, but manual assessment of retinal images is time-consuming and prone to errors. This research introduces advanced deep learning models, including a hybrid VGG16-XGBoost, DenseNet 50, and DenseNet 121, for efficient DED detection. Using the APTOS 2019 dataset, the models achieved high accuracy: hybrid network (81.50%), DenseNet 50 (97.20%), and DenseNet 121 (98.60%). The DenseNet 121 model outperformed existing techniques, emphasizing its effectiveness in automated DED diagnosis, potentially improving precision and efficiency for both practitioners and patients.