2D Information Processing And 3D Reconstruction Of Coronary Artery From X-Ray Angiograms

X-ray angiography is one of the significant imaging techniques to diagnose coronary artery disease. However, the traditional treatment methods based on two-dimensional image have serious limitations. The automatic computation of the 3D coronary arterial tree allows the clinicians to visualize the arteries and can also support the geometric measurements for better prediction of stenosis. The main work of this thesis is carried out about the extraction of the coronary artery, vessel skeleton, diameter and 3-D reconstruction and optimization. The works can be included in four parts as follows:Vessel segmentation is difficulty in angiogram image processing. Our method makes full use of the information of the vessel movement, shape, scale and intensity. For the case of slow movement of the background tissue and the more dramatic activity of the target artery, a sharpening frequency domain filter is employed to improve the image contrast. Then, the improved morphological method is used to remove nearly stationary noise and highlight the artery tree. At last, after relative-threshold comparison and multi-scale images combination, we obtain the result. By the method in this paper, we can automatically extract almost entire coronary artery tree from the low S/N x-ray angiographies, especially for the small and distal vessel part.In the aspect of 2D information processing, morphologic thinning method is employed on the segmented results to pick up the vessel skeleton and 8 chain code is used to represent the skeleton. Then we separately use'vertical line'and'disk rolling'method to extract the vessel diameter along skeleton. Experiments validate our method yield good performance on binary image.In the 3D reconstruction of coronary artery, we firstly discuss the geometry model of angiographic system and the 3D coordinates transformation to educe the 3D point reconstruction method. Then B-Spline interpolation method is used to increase the original sampling rate to make skeleton smoother and have better visibility.Finally, we optimize the exterior geometric parameter of the angiography system. The adaptive simulated annealing method is used to search the optimization solutions. The optimization method has good efficiency and the optimization results are improved compared to the original parameters.