Dual Convolutional Neural Network with Explainability Attention Mechanism for Diabetic Retinopathy Classification

Diabetic retinopathy (DR) is a worldwide chronic eye disease which is a complication due to long-term diabetes. DR can gradually reduce a patient's vision and even lead to blindness. Therefore, in order to effectively prevent potential retinopathy, early diagnosis and timely intervention are particularly important. In recent years, with the rapid development of deep learning technology, convolutional neural network (CNN) has become an important analysis tool in the medical field. In this study, we propose a novel dual convolutional neural network model that combines the Convolutional Block Attention Module (CBAM) to enhance the feature extraction capability. Specifically, we fused the network “PD-NET” based on null convolution and pyramid pooling and the network “ResDSCM-NET” based on deeply separable convolution and multi-scale convolution to construct the final ensemble model. In order to verify the effectiveness of the proposed model, we conducted experiments using the DR dataset. The experimental results show that the model achieves 96.38% accuracy in detecting DR, which is significantly better than the existing methods. The results indicate that the proposed model has high accuracy and potential for early diagnosis of DR and is expected to play an important role in clinical applications.