Retinal Vessel Segmentation in Medical Diagnosis using Multi-scale Attention Generative Adversarial Networks

With the advancement of medical technology, the demand for efficient and precise medical diagnosis is growing. Retinal vessel segmentation using artificial intelligence techniques is vital for medical diagnosis of degenerative retinal diseases. In this paper, we introduce a multi-scale generative adversarial network with class activation mapping, which can effectively improve the efficiency and accuracy of vessel segmentation using artificial intelligence. The task of vessel segmentation is better achieved due to our proposed architecture, which incorporates an attention mechanism and a multi-scale discrimination. It not only strengthens the ability to locate and segment fine retinal vessels, but also enables the model to have the ability to discriminate different receptive fields. To tackle the instability problem caused by unsupervised learning of generative adversarial networks, we introduce a supervised segmentation loss to improve model stability and convergence speed. And we propose a data augmentation method by reconstructing and combining fundus images to make the model obtain better generalization ability. We compare our method with previous models by several metrics and perform ablation study on each component of the model, demonstrating the superiority and effectiveness of the model.

Automated summary

In this paper, a multi-scale generative adversarial network with class activation mapping is proposed to enhance the efficiency and accuracy of vessel segmentation using artificial intelligence. The incorporation of attention mechanism and multi-scale discrimination improves the ability to locate and segment fine retinal vessels and discriminate different receptive fields. The instability problem caused by unsupervised learning is addressed by introducing a supervised segmentation loss, and a data augmentation method is proposed for better generalization ability. Experimental results and comparisons with previous models demonstrate the superiority and effectiveness of the proposed model.