Optimization For 3D Reconstruction Of Coronary Artery Tree By Two-stage Levenberg-Marquardt Algorithm
Zeyu Fu, Zhuang Fu, Zening Gong, Xin Feng, Haoran Gu, Rongli Xie, Jun Zhang, Jian Fei · 2021 27th International Conference on Mechatronics and Machine Vision in Practice (M2VIP) · 2021
Three dimensional(3D) information acquisition and optimization of coronary artery tree is a key technology for the evaluation and diagnosis of cardiovascular disease. Based on the X-ray coronary angiography images in two directions, 3D reconstruction method establishes a projection model using the theory of perspective projection. In this study, on the basis of two-stage Levenberg-Marquardt(LM) algorithm, a novel method is employed for refinement of the 3D structure of vessel skeletons, considering deviation of angiography angle and distance. The first stage LM algorithm(LM1) optimizes the epipolar line matching error and the second stage LM algorithm(LM2) optimizes the back projection error. The experimental results show that this algorithm manages to obtain reconstruction model more accurately within a shorter time, verifying the efficacy of the proposed method.