Computer Vision Based Coronary Angiography Image Modeling And Analysis

Heart vascular disease is one of the most dangerous diseases for human health; while the requirement of its pre-discovering and wound-less diagnosing poses a severe problem to medical researchers. The rapid development of modern medical image formation technique can provide doctors with global information of cardiac structure. This thesis carries out a study on the modeling and analysis of coronary images by employing computer vision methods; also we have investigated several key problems. However, considering the fact that the heart's function is closely related to its motion pattern, we also performed quantitative analysis to the motion information, so as to provide the doctors with a powerful facility of diagnosing and analysis heart vascular disease.The organization of the study of this thesis is as follows:1) Modeling and visualization of 3D heart vascular, two main parts are included: we first introduce the construction of correspondence in 2D space using 2D contours; 2D center lines and tree structures, and we then introduce the reconstruction of 3D vascular using stereo vision based method.2) Considering the fact that the heart's function is closely related to its motion pattern, quantitative analysis to the motion information is very important. Afterwards, we have fulfilled motion analysis based on the vasuclar models which we have already built. We have brought forward some new ideas in motion analysis, thus designed and completed String Matching Method and Thin Plate Spline Model based Method. Further, we have invented a mixed model which combined feature based model and intensity based model together, so as to compensate both disadvantages. To be more convincible, we have analysis the experimental results which include both simulate data and real data.3) After the correspondence information of structures has been established, we will furthermore employ these kinds of information and other structure based prophetic information, thus to ameliorate the image segmentation procedure, which tend to solve these problems that can not be solved by single frame image segmentation separately.4) Finally, we have integrated every modules into an united platform, thus designed and completed a software system, which is aimed at the real application in the future.