Diabetic Retinopathy Detection Using Deep Convolutional Neural Networks Architecture Resnet-18.

Abstract

Listen

Translate article

Diabetic Retinopathy (DR) is a major cause of blindness among individuals with diabetes, emphasizing the need for early detection to prevent severe vision loss. A deep learning method based on a ResNet-18 model is employed to automatically detect DR from retinal images. Data augmentation techniques, such as random rotations and horizontal flips, are utilized to improve the model’s ability to generalize to unseen data. Retinal images undergo pre-processing and normalization before being fed into the ResNet-18 network, which is fine-tuned for binary classification to identify the presence or absence of DR. The model is trained using the Adam optimizer and cross-entropy loss, with performance monitored over several training epochs. Accuracy and loss are measured on both training and testing datasets to evaluate the model's effectiveness. Results show that the model achieves strong accuracy in DR detection. Visualizations of the loss and accuracy trends offer insights into the learning process, demonstrating the potential of deep learning in automated DR screening for early diagnosis in clinical settings.