一种基于水平集的肺部血管快速分割方法

肺部血管结构微小,CT片有噪音、血管边界模糊等原因的影响,使分割肺部血管成为肺部分割的难点.针对此问题,提出了应用基于阈值的水平集三维分割肺部血管的方法.水平集分割方法是基于几何形变模型而不是参数形变模型,因此方法可根据肺部解剖结构确定肺部血管不同部位的阈值范围和肺部血管的分布范围,从而解决由于边界模糊产生肺部血管分割不准确的问题,并在此基础上,提出并描述基于水平集肺部血管树枝杈点的判定方案.

3D Rapid Segmentation of Pulmonary Vessel and Feature Extraction Through Threshold Level Set

The segmentation of pulmonary vessel is a challenging problem because of its tiny vessel,noisy CT image and blurred boundary.A method based on threshold level-set is therefore proposed for the 3-D segmentation of pulmonary vessel,which introduces the geometric contour-varying models instead of parametric contour-varying ones.As a result,this method is available to solve the inaccuracy of blurred boundary of the pulmonary vessel by determining the range of threshold of the position of pulmonary vessel and its distribution in accordance to anatomical structure.Thus,the segmentation of pulmonary vessel can be done more accurately.In addition,a judging scheme of branching node of pulmonary vessel tree is proposed on level-set basis.