The Study On 3D Reconstruction Of Coronary Arterial Tree's Skeleton Based On Single-plane Coronary Angiograms

Despite of advances in cardiovascular imaging, coronary angiography remains the"gold standard" for the assessment of coronary artery disease. 3D (three dimensional)reconstruction of coronary arterial tree based on the two single plane angiograms has beenan active area of research over the past years. The automatic computation of the 3Dcoronary arterial tree allows the clinicians to visualize the arteries and can also supportthe geometric measurements for better prediction of stenosis. Clearly, 3D reconstructionof the arterial tree is significant for providing accurate quantification of the coronaryvasculature and stenosis assessment. Major contents of this dissertation including the following: Discussion of the imaging theory of coronary angiogram, introduction of assessment of stenosis and comparison of 2D QCA (quantitative coronary arterial) and 3D QCA. The projection model of coronary angiogram is presented according to the perspective principle. And the theoretical model of 3D reconstruction is built up based on the transformation between the two angiograms. The method for 3D reconstructing a spatial point is presented with the reconstruction error analysis. Study on automatic identification of the feature points, segmentation of 2D skeleton tree and description of topological structure using directed binary tree. Registration bifurcations and endpoints according to the topological constraints and epipolar constraints, and matching the vessel segments. Improving the accuracy of matching by optimizing the transformation. After registration of pixels and interpolation, 3D coronary arterial skeleton is reconstructed. Based on the object-oriented idea a reproducible graphic class is established called COpenGL, with which to show 3D skeleton. Results are provided with error analysis on the computer-simulated and human angiogram data to critically evaluate performance of this method.