针对目前临床肝脏病灶切除、移植等手术中存在的依赖医生经验、安全性有待提高的临床实际问题，该项目提出了一种新的基于组合稀疏轮廓（CSC）先验模型和最小监督分类思想的病变肝区、肝脏实质、门静脉和肝静脉的分割算法思想并根据分割结果自动对肝脏进行Couinaud八段划分的解决方案。通过相关算法的理论研究，为计算机辅助肝脏切割手术尤其是活体肝脏移植手术的治疗计划提供理论支持。新算法针对病变肝脏边界模糊的难题首次提出组合稀疏轮廓（CSC）先验模型；同时从医学图像中提取统计及空间信息来进行感兴趣区的最小监督分类，在精度上拥有较大的优势，有望达到临床应用的标准并实现自主创新。开发出的算法将经过大量的临床医学实验（主要针对三维CT腹部数据）验证其精度、可靠性和实用性，在医学图像处理、虚拟器官建模、模式识别等领域完成一系列高水平的研究成果,发表高水平学术论文，并实现有完全自主知识产权的创新性临床应用。

The traditional liver surgery such as tumor excision and transplantation relies much on doctors' surgical experience, where the safty requirs significant improvement. In order to solve the problems, this project proposes a new segmentation algorithm, which is based on a novel composed sparse contour model and a minimum supervised classification, for lesion area, liver parenchyma, portal vein and hepatic vein. Using the segmentation result, it can also automatically divide the liver into eight Couinaud segments. The proposed algorithm will provide a fundamental theory, algorithm and surgical planning for image guided liver surgery and liver transplantation.The new algorithm inludes a novel composed sparse contour model, and uses the statistical and spatial information extracted from the medical image to achieve the minimum supervised classification of regions of interest. It can solve the difficult problem of fuzzy boundary of liver lesions and has a large advantage in accuracy, which is expected to reach the standard of the clinical application and realizes independent innovation. The efficiency and robustness of the proposed algorithm will be tested by using plenty of clinical datasets ( mainly 3D CT abdomen data). The implementation of this project will bring in a series of high level research outputs in medical image processing, virtual modeling, pattern recognition, publish a bunch of high level research papers, and also achieve clinical application which has completely independent intellectual property rights.