The Cardiac Movement Tracking Research Of Three-dimensional Echocardiography And MR Image

Statistics show that heart diseases have become the first killer of human healthnowadays，threatening and endangering people’s health. The role of the left ventricle inthe heart is the most important which pump the blood into the whole body and organsthrough the cyclical shrinkage, diastole and twist. Therefore, we can use the abnormalmovements and abnormalities of left ventricle to judge whether a body has a disease ornot. We can give evaluation objectively by tracking the movement of the left ventricular.Researches show that the unusual form of cardiac motion will generate large changes inmany diseases. Therefore, there is positive significance to heart function by tracking themotion of the left ventricle.Real-time three-dimensional ultrasound technology and cardiac MR technology areall noninvasive heart observation methods. The tracking method of cardiac motion,which is based on image processing, is proposed in this paper, including analysis ofimage filtering, image registration, contour tracking of left ventricle and time sequenceimage segmentation. This paper use left ventricle time sequence images forming bythree-dimensional ultrasound technology to analysis the movement of left ventricle, andthree-dimensional reconstruction of left ventricle time sequence images in cardiac cycleare realized.This main work and results mainly includes the following aspects:①Left ventricular ultrasound image preprocessing is raised. Several commonnonlinear filtering methods of left ventricular ultrasound images, such as medianfiltering, Lee filtering and morphological filtering, are proposed. By comparing theresults, we choose morphological filtering as image preprocessing method in this paper.②Left ventricular MR image registration is researched. We present a novel imageregistration algorithm combining Radon transformation and power spectrum, which hasa very good result in the heart MR image registration. Edge detection in this algorithmnot only reduce noises in the image, but also reduce the amount of input information.Radon transformation’s dimension reduction, rapid extraction useful information andthe power spectrum’s rotation invariant features can make Left ventricular MR imageregistration come true commendably.③This paper put forward two kinds of left ventricular ultrasound imagesegmentation methods. The first kind of segmentation method is time independent pulse coupled neural network, which can successfully extract edges of left ventricularultrasound image and maintain the integrity of the profile at the same time. The secondkind of segmentation method can realize the automatic segmentation of the leftventricular time sequence ultrasound image, providing excellent material for furtherimage registration. We make use of projection method to locate coarsely the leftventricular, and then use C-V level set method to make accurate segmentation to get theaccurate location of left ventricular.④Realizing the motion tracking and3D reconstruction of left ventricular. Thepaper makes the edge artificial marker tracking to the left ventricular ultrasonicsequence image after segmentation, and measure motion parameters. On OpenGLplatform, we realize3D-surface reconstruction of the left ventricular by using theshortest diagonals method, and realize3D-surface-reconstruction of left ventricularbased on contour connection.The cardiac movement tracking algorithm referred in this page that based on imageprocessing involved the arithmetic of image registration, image segmentation isfeasible and valuable in clinical application. It’s hoped that this algorithm can beconsummated in the future research, and be applied in clinical analysis as quickly aspossible.