Classification of Diabetic Retinopathy Using Efficientnet-B7 with Hyperparameter Optimization

Diabetic Retinopathy (DR) is a microvascular complication of Diabetes Mellitus caused by retinal blood vessel damage, potentially leading to permanent blindness. Early detection is crucial to preventing disease progression. However, studies show that DR is often detected only in its advanced stages. This research classifies DR images using the EfficientNet-B7 Convolutional Neural Network (CNN) architecture with Hyperparameter Optimization (HPO) to achieve optimal results. Experiments were conducted with different data splits, dense layer configurations, and learning rates. The best training performance was achieved with a 90%-10% data split, 256 dense units, and a 0.01 learning rate, reaching 95.48% accuracy. The best testing performance was obtained with a 90%-10% data split, 32 dense units, and a 0.001 learning rate, achieving 95.81% accuracy. These results demonstrate that EfficientNet-B7, combined with optimized hyperparameters, enhances DR classification accuracy and provides a promising approach for early DR detection.