基于SMDC连接代价算子的肺血管分割算法研究  

Algorithm of Lung Vessel Segmentation Based on SMDC-Connection Cost

作　　者：黄煜峰[1] 王兴家[1] 赖凯[1] 童同[1] 李传富[2] 冯焕清[1] 

机构地区：[1]中国科学技术大学电子科学与技术系,合肥230027 [2]安徽中医学院第一附属医院影像中心,合肥230031

出　　处：《北京生物医学工程》2010年第3期235-240,共6页Beijing Biomedical Engineering

基　　金：国家自然科学基金(60771007)资助

摘　　要：针对CT肺图像中的血管分割,提出一种基于SMDC连接代价算子的分割方法。首先设计一种背景填充技术,对灰度变换范围有选择性地收缩或拉伸,有效地消除背景噪声对血管分割的影响,然后应用SMDC连接代价算子对肺部区域进行分割,最后用实验证明了该方法的有效性。相似度分析表明自动与手工分割结果吻合率高于95%,并能较好地保留细节。In order to solve the difficulty of lung blood vessel＇ s extraction, we propose a new segmentation method based on selective marking and depth constrained （SMDC）-connection cost. First, a new method of painting background based on intensity nonlinear mapping is brought forward for getting rid of the influences of background. Then, SMDC-eonnection is applied to segment the lung blood vessel. Finally, comparative experiments show the efficacies of this method. Then a case with normal slices is used to analyze the similarity between the results from computerized segmentation and handwork, and a superposition proportion above 95% is achieved and the details are well presented.

关 键 词：SMDC连接代价 肺血管分割 数学形态学 灰度映射 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]