Realization And Application Of An Automatic Segmentation Algorithm For Human Airway Tree

With the development of CT,PET,MRI and other modern medical imaging technology,medical images play an important role in the diagnosis of disease.Because of the X-rays through the air and tissue attenuation degree of different,so by CT scan,the images became important choice of features are extracted from the images,therefore by CT scan,the images in the observation of the lung airway tree to diagnose lung related diseases has a significant role.But in another respect due to the CT scan data information amount is large and complex,so for doctors,most of the data are not the doctor want,such complex data information will bring trouble diagnosis of doctors.For this problem a solution is we used some of the characteristics of the lung airway tree the doctors need to inside part of lung CT image segmentation and 3D display,the segmented results are presented to the doctor.So it is very important for the algorithm to segment the image of the trachea in the lung CT image accurately and quickly.Today,the 3D segmentation of the tracheal tree display this kind of computer aided technology is not in clinical medical application,so the research purpose of this thesis is to through the algorithm to automatically find a fast and accurate segmentation of CT image of lung gas pipe tree scheme.In this paper,based on the morphological structure of the tracheal tree proposed a new algorithm to extract human lung CT image of the bronchial tree,this algorithm including clamp,Gaussian smoothing filter,3D region growing,three-dimensional wave propagation,three-dimensional morphological detection,closed operation fill in the hole of the five components.In this paper,the automatic extraction algorithm has been applied to 28 different cases,and can successfully split the 10 levels above the bronchus,the whole of the algorithm of the trachea use the time of the 15s.This thesis analyzes and discusses the application of CT data types and algorithms based on the segmentation results.And on COPD and lung cancer patients with the lung image segmentation,by comparing the tracheal data and tracheal tree form to summarize and guess the impact of these diseases on the airway tree,provides the data basis for the future research.The segmentation method is fast and accurate,which can provide a reference value for the diagnosis of human lung and trachea.