Coronary Artery Centerline Extraction Based On CT Image

Coronary atherosclerotic heart disease(referred to as coronary heart disease)is a health issue worthy of high attention.The incidence of complications is still on the rise,and has become a major disease that harms human health.Therefore,extracting the center line of coronary arteries based on X-ray computed tomography(CT)is extremely important.CT coronary angiography can not only understand the direction and shape of blood vessels,determine whether coronary arteries have stenosis,and assess whether coronary arteries have functional changes but also be used as the basis for various visualization technologies such as 3D visualization and multi-phase image registration.However,in the CT imaging process,due to the beating,contraction of the heart and the irregularity of the coronary direction,it is still very difficult to accurately extract the center line of the coronary artery.Therefore,this study hopes to improve the accuracy of the coronary centerline extraction and the degree of automation in the extraction process.The main work in this paper are shown below:First,a method of segmenting coronary artery based on CT images is proposed.Firstly,anisotropic diffusion filter is used to filter the original image to eliminate shot noise in the image;secondly,the Frangi vascular similarity function based on the Hessian matrix is used to enhance the vascular area in the image and suppress non-vascular region to make the image easier to segment;thirdly,the region growing method is used to segment the blood vessels in the CT sliced image.Second,proposing an algorithm of a method for coronary artery centerline extraction based on the model of inscribed ball of tubular tissue(MIBTT).First,the algorithm uses twice distance transform: the first time uses the distance transform to complete the boundary search of the original graphics and the second time uses the distance transform to calculate values of distance transform for all voxels,and deletes unnecessary voxels based on the values of distance transform,in order to complete the initial contraction of the vascular region and reduce the computational cost of the next step.Then use non-witness voxels to construct the largest inscribed ball model to find the skeleton voxels that can reflect the shape of the original graphics.Finally,on these preliminary extracted skeleton voxels,a principle similar to the dichotomy method is used to optimize the skeleton line to obtain the final coronary centerline.By evaluating the experimental results,the algorithm can accurately extract the centerline of the coronary arteries.Third,proposing an algorithm of a method for coronary artery centerline extraction based on twelve-direction topology thinning(TDTT).This method uses the twelve-direction topological thinning method in three-dimensional space to preserve the topological structure of the coronary arteries to the greatest extent,and then uses the Dijkstra algorithm to trim the thin branches generated during the thinning process,and finally obtains a smooth coronary centerline.Next,the results of the extraction of the coronary center line without the small branch removal and the small branch removal were evaluated,which proved that the coronary center line had higher accuracy after the small branch removal.At the same time,the algorithm is used to extract the centerlines of the complete coronary artery tree.The results prove that the algorithm also has a good effect on the centerlines extraction of a complete coronary artery tree.