Comprehensive Review on the Application of U-Net Architecture for Hepatic Neoplasm Segmentation in Computed Tomography Imaging

肝脏肿瘤(Hepatic Neoplasm)作为全球公共卫生领域的重大威胁，其发病率与致死率持续攀升的态势亟需精准诊疗技术的突破。在多种计算机断层扫描(computed tomography, CT)图像进行肝脏及肝脏肿瘤分割的深度学习(Deep Learning, DL)方案对比中，U-Net及其变种表现较优秀。为此，本文对近年来U-Net在肝脏及肝脏肿瘤CT图像分割中的应用以及具体优化方法进行了归纳，对比分析不同模型间的优劣并提出可能的发展方向，以期为进一步研究提供参考。Hepatic neoplasm pose a significant global public health threat, with their persistently rising incidence and mortality rates necessitating breakthroughs in precise diagnostic and therapeutic technologies. Among deep learning (DL) solutions for liver and hepatic neoplasm segmentation in computed tomography (CT) images, U-Net and its variants demonstrate relatively superior performance. To this end, this paper provides a summary of the applications of U-Net in the segmentation of liver and hepatic neoplasm CT images in recent years, as well as specific optimization methods. It compares and analyzes the strengths and weaknesses of different models and proposes potential directions for future development, with the aim of providing a reference for further research.