Vessels Extraction Of Coronary Angiogram Based On Empirical Mode Decomposition And Modulus Maxima

Coronary heart disease,which is caused by coronary artery stenosis, is one of the most common heart disease and the occurrence of it is closely related to the number of coronary artery and the degree of coronary artery stenosis. Analyzing coronary angiography images is the the standard method of diagnosis of the disease. Vascular contrast image is equivalent to the two-dimensional projection of the tubular blood vessels in the three-dimensional space.In the imaging process, the imaging mechanism of medical imaging equipment, the display device and other factors reduce the quality of coronary angiography images,for example,it makes the image fuzzy especially the fuzzy of vascular edges.To help doctors make an accurate diagnosis of the disease,we apply vascular tracking technology in coronary angiography images to extract characteristics of the blood vessels and suppress noise in the angiogram.In order to ensure the integrity and accuracy of vascular in feature extracting process, according to the image features of coronary angiography image, this paper proposes a image preprocessing method which based on linear interpolation. This method makes the coronary angiography image more smooth in area without pixel transition and better meet the requirements of the contrast image extraction algorithms. Cardiovascular is tubular structures in three-dimensional space,so after the irradiation of X-ray the cardiovascular has the following characteristics:slow change of image intensity in the direction of blood flow but rapid change of image intensity in the direction perpendicular to the direction of blood flow. With this feature, the contrast image is decomposed into several layers using empirical mode decomposition method. Thereafter, applying feature extraction methods in sub-images to get clear and obvious vascular segment. Using modulus maxima to achieve the goal of image cross-referencing, the blood vessels information in the layers is summarized in the resulting image to re-constitute a significant cardiovascular image. This image preprocessing, image decomposition, image feature extraction and image reconstructing are all based on Matlab.