一种非实质性肺结节分割的新方法

A new method for segmenting nonsolid nodules on CT images

本文提出一种分割非实质性肺结节的新方法。首先通过阈值操作去除高亮的实质部分;其次根据模糊数学建立非实质性肺结节的模糊隶属度模型,并根据该模型计算各体素对非实质性肺结节的隶属度,以此为特征向量进行体素分类;然后基于Hessian矩阵的特征值进行血管增强,通过阈值去除血管;最后通过三维连通域标记获取非实质性肺结节。实验结果表明,应用本文方法能够比较准确地将非实质性肺结节分割出来,分割效果优于目前文献报道的其他方法。本文方法可作为分割非实质性肺结节的工具。

To solve the current segmenting problems in nonsolid pulmonary nodules,a new three-dimensional segmentation method based on a fuzzy membership model was developed.Firstly,a single intensity threshold was used to separate the solid voxels.Secondly,a fuzzy membership model of nonsolid nodule was built and combining with linear discriminant a voxel classifier is produced.Then blood vessels were enhanced using hessian matrix eigenvalue and deleted by a simple threshold.Finally,the nonsolid nodule was gotten through 3D connected component labeling.This method was experimented on clinical CT data,and achieved good results,showing that the proposed method can be used as a good tool to segment nonsolid nodules in CAD systems.