Study On Segmentation Of Pulmonary Artery And Risk Assessment Of Acute Pulmonary Embolism Based On CTPA Images

Acute pulmonary embolism,with the characteristics of low diagnostic rate and high mortality,is an emergency and critical cardiovascular disease.If the disease can be diagnosed and treated in time,the mortality can be greatly cut down.In order to reduce the false positive rate,false negative rate,and the labor and time costs,this article conducts a study on the computer-aided diagnosis,including segmentation of pulmonary artery and risk assessment of pulmonary embolism.The study defines the pulmonary artery obstruction index of pulmonary embolism according to location,volume,and other important factors of emboli,which helps doctors improve the accuracy of diagnosis and helps patients save valuable diagnosis and treatment time.The study mainly comprises segmentation of pulmonary artery and its branches,and definition of risk assessment index of the emboli.The first part proposes an improved three-dimensional region growth method based on anatomical characteristics and three-dimensional connectivity of pulmonary artery.The algorithm combines the mathematical projection theorem to introduce an isolation plane,which prevents segmentation leakage and meets requirements for segmentation accuracy.The second part defines the risk assessment index,focusing on the position and volume of the emboli,and proposes emboli locating and risk grading method based on three-dimensional skeleton.The parallel topology refinement algorithm is used to extract the three-dimensional skeleton of the pulmonary artery.The hierarchical agglomerative clustering algorithm is used to simplify the bifurcation,and then realizes the grading of the pulmonary artery.And the severity index of the emboli position is obtained.In this paper,three-dimensional segmentation and reconstruction of the pulmonary artery and its branches are carried out with multiple sets of pulmonary CTPA images provided by the China-Japan Friendship Hospital.The improved algorithm can effectively segment the pulmonary artery to the lung subsegments,ensuring good three-dimensional connectivity and accuracy.At the same time,a large number of experiments were conducted to verify the accuracy of the emboli position.For entire pulmonary artery,the accuracy of position recognition is 90.00%.For central pulmonary artery,the accuracy is 98.07%,and for peripheral pulmonary artery,the accuracy is 81.25%.This paper establishes an overall framework of pulmonary embolism computer-aided diagnosis system and has a research on segmentation and risk assessment in theory and experiments.It is proved that the proposed method is feasible and lays a solid foundation for clinical application in the pulmonary embolism computer-aided diagnosis system.