快速边界行进算法:一种CT图像肺实质自动分割策略  

Fast border marching algorithm:automatic lung segmentation on chest CT images

作　　者：张林[1] 何中市[1] 张杰慧[1] 

机构地区：[1]重庆大学计算机学院,重庆400044

出　　处：《计算机应用研究》2011年第9期3556-3558,共3页Application Research of Computers

基　　金：国家科技重大专项资助项目(2008ZX07315-001);重庆市重大科技专项资助项目(CSTC;2008AB5038);中央高校基本科研业务费资助项目(CDJXS11180019)

摘　　要：提出了快速边界行进算法分割CT图片中的肺实质,解决常用分割方法不能正确分割出与胸膜和纵膈相连肺结节的问题。本算法首先通过高斯平滑、阈值分割等方法预处理CT图片,获取肺实质区域;然后采用Douglas-Peucker算法提取边缘的关键点以压缩肺实质边缘;最后,采用自适应边界行进算法修补肺实质边缘。实验结果证明了本算法能有效地分割出与胸膜和纵膈相连的肺结节,并减小了时间开销。This paper presented an automatic lung segmentation algorithm called fast border matching algorithm to extract lung parenchyma,revising the segmentation defects caused by juxtapleural nodules.Fast border matching algorithm firstly used some typical methods to extract lung parenchyma,such as Gaussian smoothing,gray-level thresholding,FloodFill non-lung region,etc.Then,introduced Douglas-Peucker algorithm to extract key points from the border of the lung parenchyma.Lastly,it employed adaptive border marching algorithm to smooth the lung border and reliably included juxtapleural nodules.Experimental results show fast border marching algorithm can extract juxtapleural and mediastinal nodules,and reduce the time overhead.

关 键 词：肺实质分割 肺结节 快速边界行进算法 D-P算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]