Research On And Implementation Of The Algorithm Of CT-based Cardiac Coronary Artery Segmentation

The extraction of coronary artery and vessel centerline and the measurement of vessel stenosis are the focuses in the field of medical image processing and analyis. According to the scale space theory, a qauntiative analysis method of coronary artery Computer Aided Diagnosis (CAD) based on CT dataset is proposed.In scale space, the member function of tubular structure is constructed considering the responses in different scales on the base of differential geometry theory of tubular structure to extract the seed points in coronary artery automatically. The algorithm of extracting seed points automatically have implemented on Cardiac CAD software platform. For the high-quality CT images, the accuracy rate of seed point extraction is above 90%. By the way of a shape based region growing algorithm, the leakages where the image information ambiguous is restrained. The results show that the algorithm can improve the automation of Cardiac CAD system by one-click vessel extraction. The algorithm has been successfully applied in coronary arteries analysis of Neusoft Cardiac CAD.Skeleton is frequently applied in medical image analysis. Thinning is a process which produces the skeleton of an object while its topological structure is preserved; it is the important step of medical image analysis. A 3D Thinning algorithm extracting medial lines of objects was described, which remove six kinds of simple points layer by layer from the outmost layer by iteration until only a medial axis line remains. Then the resulted central path was smoothed, providing a root of every tree. At last the thinning result had been validated and evaluated by phantoms.