Research on Arteriovenous Vessel Segmentation in Fundus Images Based on Topology

Aiming at the shortcomings of traditional image processing methods and existing deep learning models in fundus image segmentation, this paper proposes a deep learning model based on topology, which combines the structural advantages of U-Net and introduces the Dice loss function based on cartilage skeletonisation to optimise the segmentation effect of the vascular structure. Experimental results show that the model significantly improves the segmentation accuracy and classification accuracy of arterial and venous blood vessels while maintaining high computational efficiency, especially in handling fine blood vessels and distinguishing arterial and venous veins with superior performance. The research in this paper not only helps to improve the performance of fundus image segmentation techniques, but also provides important support for the use of deep learning to compute retinal blood vessels to assist clinical diagnosis.