Automatic Skeletonization,Labeling And Structural Analysis Of Lung Airway Tree

Human lung airway tree,consisting of the trachea,the main bronchi and the lobe,segment bronchi(about 23 generations),is a critical part of the respiratory system,and its changes in the structure and function are the major reasons and manifestations of the respiratory diseases.Multi-detector computed tomography(MDCT)is an important and noninvasive method for the airway disease evaluation.Using MDCT,it only takes a breath holding to get the whole lung's structural images that are continuous and(or)overlapped,with the approximate isotropic voxels,and with high resolution and thin layer(0.75-1.5 mm).However,the huge amount of data acquired by MDCT has become a huge challenge for the scientific diagnosis and research,and it is necessary to perform the automatic skeletonization and structural analysis of the airway tree using the advanced image processing technologies.The purpose of this paper is to propose a method for the analysis of lung airway tree structure,which can realize the automatic skeletonization,labeling and quantitative analysis of the lung airway tree.This paper uses two homemade models and 26 CT image data collected from ShengJing Hospital of China Medical University to carry out the algorithm verification.Firstly,the topology thinning algorithm is verified by the established models,and the improved method of packet connectivity test is proposed.Then,the improved thinning algorithm is applied to the airway tree models segmented from the CT image.Specifically,after dealing the image with morphological processing,the topological thinning method is applied to skeletonize the lung airway,then the simple points detected by the Euler characteristic and connectivity are deleted,and the final skeleton of the lung airway tree is generated.On the basis of the lung airway tree skeleton,the branching points and the leaf nodes of the skeleton are extracted,and the bronchi with different generations are connected with different color,and the tree structure of the lung airway is obtained.Finally,the length and the bifurcation of the bronchi are measured based on the tree structure.The improved topology thinning algorithm can get completely correct results for the two models constructed.In addition,this paper analyzed 26 sets of data,successfully extracted all data of lung airway tree skeleton,and succeeded in generating the tree structure of 22 groups of skeleton,14 of which do not have wrong bronchi.In this paper,the highest generation of the lung airway tree structure can reach 15,and the maximum of the numbers of the obtained leaf nodes is 52.At last,it is found that the average value of the length of the left main bronchus is 76.43 mm,the average length of the right main bronchus is 37.06 mm,the average value of the bifurcation angle between the left and right main bronchus is 109.76 degrees.The length and the bifurcation angle are consistent with the reported results.The results show that the proposed method of pulmonary airway tree analysis can realize the automatic extraction,labeling and quantification of skeleton to some pulmonary airway trees.The preprocessing of pulmonary airway tree is essential for the extraction of skeleton,and the lung airway tree with smooth surface and without holes is the basis of the correct extraction of the skeleton.This method has some potential value for the anatomical measurement of airway tree,the better understanding of the topological structure of the airway,and the quantitative diagnosis of airway diseases.