Application of an Improved U-Net-Based Edge Optimization Model in the Adjuvant Treatment of Bladder Cancer

In recent years, image segmentation has been widely used in medical image analysis. However, due to the problem that traditional segmentation methods have too many complex image features, it is difficult to extract effective information from medical images with complex semantic information, so the segmentation performance is poor. To address the above problems, this paper proposes an edge optimization network based on an improved U-Net architecture. The network uses ResNet152 as the encoder, and uses multi-level feature fusion in the decoder to enhance the information fusion ability of the network; and adds an edge enhancement input module to refine the target boundary in order to better achieve boundary segmentation and improve the overall segmentation accuracy. Experiments show that the proposed network can accurately segment the lesion area in cystoscopic images, which is useful for early diagnosis of cancer.