| dc.description.abstract | Coronary angiography is an imaging technique that uses X-rays to photograph the cardiovascular system to assist in cardiac catheterization. During the operation, the doctor will take a large number of X-ray images, which are usually in two directions ( AP/LA direction). Through the positional relationship of the two images and empirical judgments, it is determined whether the catheter position is correct at each time. Sometimes in order to obtain an image that clearly shows the structure of the blood vessel, it is necessary to take multiple shots to obtain an appropriate image.
To provide physicians with more intuitive blood vessel flow information and reduce X-ray shooting, in this study, we hope to establish an algorithm for reconstructing a three-dimensional heart vessel model from two two-dimensional X-ray images. The algorithm includes the pre-processing of the image, the method to obtain the corresponding points of the two images, and finally calculate the three-dimensional space coordinates. The purpose of the image pre-processing part is to segment the blood vessel image from the background, including noise removal, uneven light calibration, and enhancement filtering; Finding the centerlines of the vessel by image skeleton then can help to find the bifurcation points of the centerlines. Then calculate the corresponding point based on the epipolar geometry and finally measure the position of the three-dimensional model in space. In the experiment of verifying the reconstruction method, 2D C-arm was used to shoot the self-made cardiovascular prosthesis, two X-ray images were obtained for reconstruction, and the verification tool was held to verify the correctness of the reconstructed spatial position. The reconstruction position error is about 1.84mm, and the standard deviation is 0.74mm. | en_US |