基于形态学与局部特征结构的肺血管树重建

Reconstruction of Pulmonary Vessel Tree Based on Morphology and Local Feature Structure

孤立性肺结节(SPN)是肺癌的重要征象,基于肺部CT容积数据的血管树重建是SPN计算机辅助诊断的前提。提出一种形态学滤波和局部特征结构相结合的肺部血管树重建方法,通过形态学滤波对待分组织进行预分割,对特定组织体素计算Hessian矩阵的特征值,并按其特征结构进行分割进而重建。实验结果表明,该方法能快速、有效地实现血管树重建。

Vessel tree reconstruction for thoracic CT data is a prerequisite in Solitary Pulmonary Nodule（SPN） detection. This paper proposes a novel method to reconstruct the pulmonary vessel tree, integrating morphological filters and the local feature structure. It uses morphological operators to extract the general structure of the vessel tree, and removes the voxcls not belonging to the tissue according to the Hessian matrix eigenvalues of the special tissue voxels. Experimental results show the method is capable of quickly and effectively realizing the reconstruction of pulmonary vessel tree.