摘要
目的提出一种从CT图像中自动分割出肺实质的方法。方法第一步,利用灰度阈值把肺部区域从背景中提取出来;第二步,对提取出的肺部区域进一步处理,除去大气管、平滑肺部边缘;第三步,利用数学形态学方法分开左右两肺。结果利用该方法对三个病人的肺部CT图像序列进行处理,证明该方法能针对不同厚度的CT图像,自动选取合适的阈值进行分割,并能去除独立的气管/支气管,最后完整提取出肺实质。结论本文提出的方法为进一步利用计算机辅助诊断方法对肺结节进行识别和标记提供了基础。
Objective To suggest an auto-segmentation method for lung parenchyma of CT image. Methods An automated lung segmentation method was approached for some 3D X-ray lung CT images, which consisted of three steps. Firstly, the lung region was extracted from the CT images by gray-level threshold. Secondly, the trachea or bronchi was removed from lung parenchyma and the lung borders were smoothed by morphological operations. Finally, the left and right lungs were separated by mathematical morphology method. Results The method had been tested by processing 3D CT data sets, which had different slice thickness. The results showed that this method can automatically select thresholds and remove isolated trachea/bronchi. Conclusion The automated lung segmentation method will be helpful for nodule detection in computer-aided diagnosis.
出处
《中国医学影像技术》
CSCD
北大核心
2006年第9期1428-1431,共4页
Chinese Journal of Medical Imaging Technology
基金
上海市重点学科建设项目资助(P0502)。
关键词
自动分割
肺实质
体层摄影术
X线计算机
Automated segmentation Lung parenchyma
Tomography, X-ray computed
