UNet-Based Polyp Segmentation: A Survey

Colorectal polyps are common intestinal diseases, in which image segmentation of colorectal polyps is crucial for diagnosis and treatment. With the development of deep learning, semantic segmentation technology has been widely used in colorectal polyp image analysis. Among them, the U-Net model is a widely used deep learning model, which is simple and effective and shows good performance in medical image segmentation tasks. This article mainly introduces the semantic segmentation technology of colorectal polyp images based on U-Net, including the basic structure, characteristics and advantages of the U-Net model, and its application in semantic segmentation of colorectal polyp images. At the same time, this article also reviews colorectal polyp image segmentation technology based on deep learning in recent years, focuses on the related research on the U-Net model and its improvements, and compares and analyzes various methods. In the methods and experiments section, this article introduces the structure, data set and experimental settings of the U-Net model in detail, as well as analyzes and compares the experimental results of U-Net-based colorectal polyp image semantic segmentation technology. Finally, this article summarizes the main contributions of the research and discusses prospects and challenges for future applications.