基于CT图像的肺实质细分割综合方法.pdfVIP

22 5 () V o.l 22 No. 5 2010 10 Journal of Chongqing University of Posts and Tele omm uni ations( Natural S ien e Edition) O t. 2010 DO I: 10. 3979/ j. issn. 1673825X. 2010. 05. 028 CT 卞晓月, 武 妍 ( , 201804) : 肺实质的分割是肺结节检测和诊断的基础, 是肺部疾病计算机辅助诊断的关键 骤之一针对肺部计算 机断层扫描 ( com puter tom ograp y, CT) 图像, 采用最佳阈值法进行初 分割, 去除背景, 用结合上下文分析的区域生 长法去掉气管支气管, 对左右肺连结的情况进行像素分析, 分开左右肺, 对提取出来的肺区用滚球法进行修复,得 到肺实质图像去除气管和分割左右肺的算法是针对肺部 CT 图像的特征提出的, 具有简单高效的特点实验表 明, 该综合方法的准确性和可靠性较高, 有较好的应用前景 : 肺实质分割; 最佳阈值; 滚球法; 上下文分析 : TP391 : A : 1673825X( 2010) A m ethod of areful lung segm entation based on CT im ages BIAN X iaoyue, WU Y an ( Sc ool of Electronics and Inform ation, TongjiUniversity, S ang ai 201804, P. R. C ina) Abstra t: T e segmentation of lung parenc ym a is t e foundation of lung nodu le detection and diagnosis. It is also one of t e key stages in computer aided d iagnosis system s to detect lung diseases. T erefore, t is articlem ade roug segm entation to get rid of t e background w it optmi al t res old algorit m, and used t e region grow t algorit m and contextual analysis to reject t e trac ea and m ain bronc .i T en, p ixel analys is w ere used to m ake t e left and rig t lung segm entation. At last, t e lung fieldsw ere gained and repairedw it fallingball algorit m. T e algorit m used to reject t e trac ea and seg m ent t e left and rig t lung was especially des igned according to t e feature of t e lung CT mi ages, w ic w ere smi ple and efficien t. T e expermi ent s ow s t at t is m et od as good accuracy, robustness and future app lications. K ey words: lung parenc yma segm en tation; optmi al t res o