Methods Research On Segmentation And Volume Calculation Of Pulmonary Artery In CT Images

Pulmonary embolism poses a great threat to the patients' safety and health,and the computer-aided diagnosis system for this kind of disease can greatly improve the diagnostic efficiency of doctors and is of great significance for the successful treatment of the disease.Computed tomography(CT)images are the "gold standard" for the diagnosis of pulmonary embolism;how to segment the pulmonary artery targets from the CT image sequence,and how to obtain the volume of the three-dimensional model for the pulmonary artery sequence,the effective solutions of these two problems are the key to the realization of computer-aided diagnosis.In this paper,the CT data of the chest are taken as the research objects,and the methods of segmentation and volume calculation of pulmonary artery in CT images are carried out.The main contents of this paper include the following three aspects:(1)In term of CT image preprocessing,the CT data conversion method based on adaptive window level and window width and the separation method of pulmonary artery and superior vena cava based on Hessian matrix are studied.The gray values of the CT image are converted from the CT values,but the images obtained by the traditional conversion methods may have the problems that the target is not obvious and the contrast is low..In response to these problems,in this paper,the CT value probability distribution function and probability change rate function are used to automatically select the window level and window width,respectively,and these two values are used to construct a new mapping relationship to complete the conversion of CT data to CT images.In addition,the pulmonary artery target in CT images is often connected to the superior vena cava,affecting the segmentation of the target.To solve this problem,the relationship between Hessian matrix eigenvalues,eigenvectors and image structure is used to construct a.mask with curved structure,and the mask is mapped to the original image to achieve;the separation of the target and the superior vena cava,The experimental results show that the CT images obtained by the conversion method proposed in this paper have clear targets and high contrast,and the separation method of pulmonary artery and superior vena cava based on Hessian matrix can successfully achieve the separation of target and background.(2)In term of segmentation of pulmonary artery in CT image,in view of the problem that the traditional active contour model cannot adapt to the changes in the number and shape of targets in the image sequence,from the perspective of a single image,the GVF-CV model and the region growth method based on adaptive seed point are proposed;and from the perspective of the overall image sequence,a segmentation method of pulmonary artery tree based on edge tracking is proposed.Among them,the GVF-CV model limits the range of targets' edge detection by introducing "Region of Interest"to reduce the false detection rate,and on this basis,the level set evolution is performed to adapt to the detection and segmentation of multiple targets;the region growth method based on adaptive seed point determines the screening conditions of seed point through the relationship between the GVF field and the contour,and using the proposed radial function-based detection method of pulmonary artery morphology to verify the feasibility of the method,and based on the selected seed point,to fill in the missing parts of the target;the method of segmenting the pulmonary artery tree based on edge tracking takes the above methods as the kernel,and determines the selection method of the initial contour set of each layer in the image sequence and the criteria for eliminating noise points.Experimental verification using data from multiple sets of clinical actual scans shows that the segmentation accuracy of the methods proposed in this paper are better than traditional methods and they are suitable for segmentation of pulmonary arteries.(3)In the term of calculation of pulmonary artery volume,for the shortcoming that the traditional pixel statistics method and voxel accumulation method cannot accurately obtain the volume,improved voxel accumulation method is proposed in this paper.The algorithm subdivides the types of voxels located on the surface of the reconstructed three-dimensional target,and determines the judgment conditions of various types of voxels;the effective volume values of various voxels are derived,and the volume index table is established;based on these,the overall flow of volume calculation for pulmonary artery is proposed.In this paper,the sequences of segmented pulmonary artery target are used as the experimental objects to analyze the effectiveness of the proposed algorithm,and the experimental results show that the algorithm has less error than traditional methods and is more suitable for the calculation of pulmonary artery volume.Finally,this paper proves the important significance of the proposed algorithm in assisting doctors through the evaluation of the degree of pulmonary embolism.