CTNet: convolutional transformer network for diabetic retinopathy classification

This work proposes a lightweight diabetic retinopathy classification method—CTNet, using the combination of CNN and Transformers on fundus images to capture both local and global spatial features.