基于先验信息水平集方法的肝脏CT序列图像自动分割  

Automatic segmentation of livers from CT series based on level set method with prior knowledge

作　　者：赵于前[1,2,3] 闫桂霞[1] 徐效文[1] 邹润民[2] 

机构地区：[1]中南大学地球科学与信息物理学院,湖南长沙410083 [2]中南大学信息科学与工程学院,湖南长沙410083 [3]中南大学"移动医疗"教育部-中国移动联合实验室,湖南长沙410083

出　　处：《中南大学学报（自然科学版）》2015年第4期1310-1317,共8页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(61172184;61379107;61174210);新世纪优秀人才支持计划项目(NCET130603);高等学校博士学科点专项科研基金资助项目(20130162110016);湖南省科技基本建设项目(20131199)~~

摘　　要：应用二次区域生长法获取肝脏的初步分割结果,将其作为先验知识,构造新的边缘指示函数和水平集能量函数,有效地解决了因肝脏形状多变以及弱边界问题带来的分割难题。在序列的分割过程中,将上一层分割结果进行距离变换,并提取大于既定阈值的部分生成下一层区域生长的种子点,避免了分割过程中的误差积累,并使序列的分割具有延续性和自动性。算法最后应用自适应边缘行进算法修补边缘,使得分割结果更加完整。实验结果证明了该方法的有效性和可行性。The livers were preliminarily segmented by two-step region growth, and then the segmented results were taken as prior knowledge to construct a new edge indicator and level set energy function. The segment effectively solves the problems of variable liver shapes and weak boundaries. During the liver CT series segmentation, the distance transform that its value is greater than a fixed threshold is extracted from the current slice segmentation to generate the region growth seed points for the next slice segmentation, which avoids the error accumulation and ensures the continuity and automation of liver CT series segmenting process. Finally, the adaptive border marching method was used to correct the segmentation defects. The experimental results demonstrate the effectiveness and feasibility of the proposed method.

关 键 词：水平集 肝脏分割 先验信息 CT序列 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]