Research Of Three-dimensional Reconstruction Of Interested Vessel Segment In Coronary Artery Tree

Cardiovascular disease is mainly the heart beating disorder caused by insufficient myocardial blood supply,which has become the first major factor threatening human health.It has the characteristics of high morbidity and high mortality.One of the traditional method of cardiovascular disease diagnosis is digital subtraction angiography(DSA).DSA has good imaging quality and imaging speed,it is widely used in clinic as one of the main basis for the diagnosis and treatment of coronary heart disease,known as the "gold standard" for the diagnosis of coronary heart disease.However,X-ray imaging is to superimpose the 3D structure of blood vessels on the 2D image through X-ray projection,and the vascular structures may overlap each other in the angiography path,which eventually leads to the inaccuracy of doctors’ observation and diagnosis.It is also the main defect of X-ray imaging.Therefore,3D reconstruction of coronary artery is helpful to improve the accuracy of quantitative analysis of clinical medical parameters.With the continuous development of medical imaging technology and computer image processing technology,as well as the research and exploration of relevant researchers,the computer analysis and diagnosis technology for coronary artery image has made a great leap,and has increasingly played a huge role in clinical application.This paper deeply studies the core of image registration and 3D reconstruction,explores the 3D reconstruction algorithm suitable for DSA vascular images.Firstly,the algorithm optimizes the system parameters recorded in the angiography equipment.Secondly,according to the characteristics of coronary artery images,a new method of vessel point pair matching is proposed.Based on this method,the global optimal point pair matching results can be achieved.Then,the 3D reconstruction is carried out by using the principle of binocular stereo vision in computer vision.Finally,rendering the vessel surface by reconstructed coronary artery centerline and vessel radius.The main research work of this paper is as follows:1.The parameters of angiography system were optimized.Aiming at the system parameter error caused by the isocentric deviation of the single arm radiography system and the movement of the patient’s heart.Combining the characteristics of trust region method and traditional Levenberg-Marquard(LM)algorithm,this paper proposes a parameter adjusting LM(PALM)algorithm to optimize the geometric transformation matrix between two imaging coordinate systems with different angles.2.Point pairs matching optimization.In this paper,a new matching method of blood vessel corresponding points based on curve descriptors and projective geometric constraints is proposed.Firstly,the curvature and brightness order curve descriptor(CBOCD)is used to represent the self occlusion of vessels during angiography.Secondly,the matching cost matrix defined by epipolar matching error is constructed,and a dynamic programming algorithm with self-adaptive step size is designed to realize the global optimal matching point pairs of blood vessels by integrating the smoothness of blood vessels and the self occlusion described by CBOCD.3.Point based 3D reconstruction.Based on the global optimal matching point pairs,the 3D reconstruction of blood vessel is carried out according to the camera parameters and external parameters by using the principle of binocular stereo vision in computer vision.4.Vascular surface reconstruction and rendering.Using the recorded vessel radius of coronary artery image after segmentation and thinning,the cross-section sequence corresponding to the reconstructed vessel centerline was constructed,and the equal angle sampling was performed on the constructed cross-section curve to obtain the surface point cloud of coronary artery.Finally,the surface of coronary artery was reconstructed by using the implicit surface reconstruction technology of 3D point cloud.This paper finally construct a 3D reconstruction algorithm for DSA vascular images.In the aspect of mean square error of back projection and vascular smoothness,the reconstruction results are significantly improved.The effectiveness and applicability of the proposed method meet the needs of clinical application.