Research On Key Techniques Of 3D Cardiac MRI Aided Diagnosis System

Cardiovascular disease is one of the most recognized around the world as the killer, and myocardial infarction is a higher mortality in cardiovascular disease. At present, some experts pointed out that in recent 5 years, due to myocardial infarction and cause death toll accounted for half of the total number of cardiovascular deaths, and the mortality rate also showed a clear upward momentum. Therefore, early on this kind of disease were diagnostic and prognostic evaluation, will lower the mortality of this disease in a large extent.In recent years, the diagnosis of cardiovascular disease most dependent on CT, X ray, ultrasonic and cardiac magnetic resonance imaging(Cardiac Magnetic Resonance Imaging, CMRI) and so on these more sophisticated imaging tools. Among these techniques, the CMRI technology with its unique imaging principle, can not only provide high resolution images of different sections of the heart, and can this kind of soft tissue on the myocardium can also clearly shows the whole, thus become an important tool in diagnosis and evaluation function of heart disease. However, the hospital every day there will be a lot of CMRI, so in this invisible to increase the workload of doctors. Therefore, in order to make better use of these image assisted doctor the patient’s diagnosis, it is necessary to develop a set of perfect system to manage and retrieve the image.This paper systematically analysis of three-dimensional cardiac diagnosis system development and design, the segmentation of CMRI in left ventricular epicardial, study on the segmentation of myocardial scar, and put forward corresponding methods, the main works and contributions of this thesis are as follows:(1) For the problem of segmentation left ventricular CMRI, proposed a segmentation method based on saliency detection of left ventricular inner and outer membrane. Methods based on visual saliency detection and based on mathematical morphology open operation, left ventricular blood pool area extraction; then to extract the blood pool region contours as the initial contour of the lining of the heart, in the band shape constrained active contour model, evolution to get accurate left ventricular epicardial. Experimental results show that, the method more accurate, can make the initial contour converge rapidly to the endocardial border, divided by the outer membrane is relatively accurate.(2) For the problem of segmentation cardiac MR image of myocardial scar, an improved k-means clustering method is proposed in this paper. The specific method is on the basis of traditional K-means clustering, combined with the image gray value, three characteristics of mean and median value consists of a three-dimensional vector to represent the pixel information, finally use the 3D vector to substitute grey as sample points is used to segment the image and its use to contain noise CMRI segmentation. Experimental results show that the algorithm for noisy CMRI segmentation myocardial scar area is more accurate.(3) Finally, based on the realization of the two algorithms, development and design of a three dimensional cardiac magnetic resonance image aided diagnosis system. System development platform used in Visual Studio 2010, combined tool library function ITK, VTK read Dicom format images, the left ventricular inner and outer membrane of segmentation, segmentation and myocardial scar the three-dimensional reconstruction of the heart, the system has the function of 3D visualization and analysis of cardiac function, and the reference data is just can be an important basis for diagnosis of the doctor to the patient’s condition.