Research On The Segmentation And 3D Visualization Of Coronary Artery CTA Images Based On U-Net

Coronary atherosclerotic heart disease(referred to as coronary heart disease)refers to a heart disease caused by myocardial ischemia,hypoxia or necrosis due to narrowing or obstruction of the blood vessel cavity,which seriously endangers the life and health of people around the world.In order to accurately and quickly diagnose the degree and location of coronary artery stenosis,and to develop an effective treatment plan,the accurate extraction method of coronary blood vessels based on computer tomography angiography(CTA)has attracted the attention of medical scientific research institutions around the world.This paper uses U-Net network to perform rough segmentation of coronary arteries in CTA images of human thorax,and then uses a level set algorithm with multiple conditional constraints to perform contour curve evolution to achieve accurate segmentation and extraction of coronary arteries.Finally,a 3D visualization system of coronary arteries was designed and implemented.The main work of this article is as follows:(1)In order to improve the contrast of the coronary arteries in the CTA image,the image enhancement processing is performed by a multi-scale enhancement filter based on the Hessian matrix.While filtering the background interference information in the area around the coronary arteries,it effectively highlights the imaging effect of the coronary arteries.(2)In order to solve the problem that the position and shape of the coronary arteries in the CTA image are large,which easily leads to segmentation errors.In this paper,fast training and accurate feature recognition methods of U-Net network are used to locate and coarsely segment the coronary artery region of interest in the CTA image of human chest.Then based on the unique characteristics of the extracted coronary arteries,a variety of conditionally constrained level set algorithms were designed to perform contour evolution,and finally accurate segmentation and extraction of coronary arteries were achieved.(3)A 3D visualization system of coronary arteries is designed by the moving cube algorithm.The system will perform 3D modeling of the coronary artery sequence images extracted by the algorithm in this paper,and can display various forms of interactive operations according to user requirements.By inputting a sample of coronary CTA image sequence for experiments,the segmentation algorithm designed in this paper can increase the average overlap rate up to 12.41% compared with other typical algorithms,which proves the effectiveness of the method in this paper.At the same time,a three-dimensional visualization system is designed to display coronary arteries in three-dimensional space,making it easier and more intuitive for doctors to observe lesion information and three-dimensional stereoscopic models of regions of interest.Provide clearer and key supporting information for doctors to develop treatment strategies to improve the cure rate of coronary heart disease.