| With the aggravation of environmental pollution,the incidence rate of lung cancer has been increasing in recent years.The initial manifestation of lung cancer is pulmonary nodules,and the incidence and mortality of the glass bulb lung nodules are much higher than those of other lung nodules.At present,CT has become the most effective medical detection method for early lung cancer detection.However,due to the progress of science and technology,the thickness of medical image slices is becoming smaller and smaller,which makes the final amount of slice data more and more large.A large number of CT images contradict the accuracy of artificial diagnosis.Therefore,the research on segmentation method of ground glass nodules has become the research of domestic and foreign experts Hot spots.In addition,the computer 3D reconstruction technology can display the overall information of the nodule,which is convenient for doctors to plan the later treatment plan.Therefore,it is of great significance to study the segmentation and reconstruction algorithm of ground glass pulmonary nodules.In this paper,a fully automatic segmentation and reconstruction method of pulmonary nodule is proposed.Firstly,the lung parenchyma is segmented to remove the interference of lung wall and other structures on the segmentation results of pulmonary nodule in the later stage.Then,the combination of super-pixel and density clustering is used to segment the pulmonary nodule.Finally,the segmented sequence image is reconstructed in three dimensions,so as to provide the research of pulmonary nodule More dimensional information.The main research contents are as follows:(1)Segmentation algorithm of lung.parenchyma.In order to ensure the accurate segmentation of pulmonary nodules,the first step is to segment the interference information including the lung wall to get the complete lung parenchyma.This paper presents an automatic segmentation method based on region growth and fuzzy clustering.This method combines the region growing method with the fuzzy c-means clustering method,and integrates the histogram information of the image to get the complete lung parenchyma image.(2)Segmentation algorithm of pulmonary nodules.In this paper,super-pixel segmentation and density clustering are combined.Firstly,the lung parenchyma image is segmented by super-pixel,and then the lung nodule is extracted by density clustering.Considering the particularity of the ground glass nodule,this paper integrates the edge features of the lung nodule into the super-pixel algorithm,changes the selection rules of the initial clustering center,and locates the lung nodule according to the brightness features,density features and roundness like degree of the lung nodule.Combined with the idea of the secondary density clustering,a lung nodule segmentation algorithm based on the edge sensitive SLIC and the secondary density clustering is proposed.Experimental results show that the algorithm proposed in this paper has good segmentation effect.(3)Three dimensional reconstruction model of pulmonary nodules.The premise of accurate 3D reconstruction is the accurate segmentation of 2D sequence,but the existing segmentation algorithm can not ensure that every layer of 2D sequence can locate the area where the pulmonary nodules are located.In this paper,an adaptive correction algorithm for two-dimensional interval sequences is proposed.The algorithm first calculates the mean value of the center point of the image sequence,then calculates the distance between the image and the mean value of the center and other images in the sequence respectively.By comparing the distance between the two,the image with the wrong location can be obtained,and the regional position of the pulmonary nodule can be obtained by using the information of the upper and lower layers,and then it can be located and segmented again.Finally,the moving cube algorithm in face rendering is used to target Reconstruction,multi-dimensional display of information,provides the basis for the follow-up treatment. |
PDF Full Text Request