Disease Classification in Retinal Images Using Vision Transformer and Capsule Network

Disease classification in retinal image analysis is of great importance in terms of accelerating the diagnostic process and increasing its accuracy in ophthalmology. Early diagnosis of retinal diseases plays a critical role in preventing the progression of diseases and managing treatment processes more effectively. Deep learning classification models offer an innovative and effective solution for the diagnosis of retinal diseases. In this study, a hybrid model combining Vision Transformer and Capsule Network is proposed. Vision Transformer provides in-depth extraction of visual features. Capsule Network increases the classification success by preserving the spatial relationships of these features. In experiments using RFMiD dataset, the performance of the proposed model is compared with VGG-16, VGG-19, Xception and ViT models. The experimental results show that the model improves the classification accuracy. The findings of the study show that new methods can be effective in computer-aided diagnosis of retinal diseases.