基于CT图像的肺气管树3D分割方法的研究

Three-dimensional Pulmonary Airway Tree Segmentation from CT Images

目的:对肺部气管树的分割在临床上具有重要应用价值。针对目前肺气管树分割存在的问题,本文提出了一种结合区域生长和形态学方法的气管树3D分割的方法。方法:首先,采用基于3D联通区域与形态学的方法分割出CT序列图像中的肺实质;其次,利用3D区域生长法初步提取气管树;然后,利用形态学分割方法选取细小气管候选区域,并与上一步分割结果合成三维肺气管区域;最后,再次利用区域生长法去除伪气管区域,提取出最终的气管树。结果:实验结果表明,三维区域生长方法能够很好地获得气管/主支气管、段气管及主要的气管分支,而形态学方法能够有效地检测出细小气管区域。所以利用本文方法可以简单、有效地提取出肺气管树,并防止区域生长过程中的遗漏现象。结论:本文方法可为肺部气管的定量分析奠定基础,具有十分重要的临床诊断意义。

Objective： Pulmonary airway tree segmentation is very valuable in clinic application.Aiming at solving segmentation problems presently,a three-dimensional segmentation method for extracting pulmonary airway tree from CT images is proposed in this paper.Methods： Firstly,the lung parenchyma was obtained from CT images using a three-dimensional（3D） segmentation method.Then,the main pulmonary airway was segmented using 3D region growing method.Next,the morphology method was used to segment the small airway regions.The final airway tree was obtained through region growing once again.Results： Experimental results showed that 3D region growing can obtain main tracheal while branch the morphology method can find out tiny tracheal area very well.So the proposed method can be effectively used in the segmentation of pulmonary airway tree from CT images while avoiding omission during region growing.Conclusions： It builds a good foundation for the follow-up quantitative analysis of pulmonary airway and was applicable in the process of clinical diagnosis.