Automatic 3D Reconstruction Of Coronary Angiography Method Based On Model Evolution

Coronary heart disease is one of the major diseases that endanger human health at present. Coronary angiography was known as one of the main basises for clinical diagnosis and treatment of coronary heart disease. With the development of technology, the rotational coronary angiography springs up in recent years. It has many advantages and also makes the 3D reconstruction of coronary called a hotspot. The main problem of coronary reconstruction is due to the motion of the heart and the influence of respiratory movement. The motion will leads to the non-homomorphisms of coronary in different projection angles. This paper studies the reconstruction of the coronary arteries in this case.This paper presents an automatic 3D reconstruction of coronary angiography method based on model evolution. Firstly, respiratory movement is corrected for the projections which belong to one cardiac phase, so it can avoid the influence on reconstruction results from the respiratory movement. Secondly, direct reconstruction result is poor cased by the irregular background in the projections, so image segmentation is done for the projections. Combined vascular particularity, we used a vessel segmentation algorithm based on the shortest path technology. The method can effectively split the vessel from the background. Thirdly, use the results of segmentation to get an initial 3D coronary tree by back projection reconstruction algorithm. Then, based on the initial coronary volume data, this paper presents an efficient algorithm for parameterizetion of the volume data. The method first extracte the skeleton from the volume, then determine vascular node topology based on the continuity and other features, at last, estimate vessel radius.We will evolve the parametric model so that it can match the original data as possible. We use an iterative gradient descent optimization method which includes a smoothness constraint item to optimize the initial 3D parametric model, the constraint item enhanced anti-jamming capability in optimization process and make the vessel more smooth.Experimental results show that the reconstructin algorithm presents in this paper can effectively reconstruct the coronary arteries and achieve high reconstruction accuracy, and the stenosis of coronary artery can also be effectively detected, and also effectively reduce the manual intervention.