Based On Speckle Tracking Heart Dimensional Ultrasound Image Motion Analysis

The heart motion analysis is an important field of medical image processing, at present the most studies are about magnetic resonance imaging(MRI); the studies of supersonic image are also few. Echocardiographic has become a useful method in motion analyzing field, along with the development of Doppler tissue imaging (DTI).Because of the rapid development of speckle tracking imaging (STI), many foreign researchers and scholars take STI as a new research direction.This thesis research a series of speckles information in the minor axis images, in view of the left ventricular (LV)'s two-dimensional supersonic image motion sequence, and analyse the speckles information, using to motion analyzing. The main work of this thesis is:1. Segmenting the images, using the improved fast Snake model with greedy optimization, which bases on that the supersonic image noise is big, and that the speckles of the cardiac muscle and the speckles of the noise are the identical type information of the image. By evolving cardiac muscle's inside and outside outlines with the different starting curve, we can confirm the area of cardiac muscle's moving, which is the limiting condition of the next work.2. Introducing the image registration algorithm of the image, which is used frequently, this thesis bring a new motion analysis method based on characteristic speckles, according to the own characteristic of supersonic image. This method has three parts. First extract a lot of characteristic spots using the Harris operator, because the Harris operator is very sensitive to the vertex information, and the cardiac muscle spots are generally the reflecting luminescent spots of the organization on supersonic instrument; second matching these spots by correlation coefficient, and then matching again by approximate distance proportion restraint, the region size of cardiac muscle, the state of heart motion; calculating the partial and each motion parameter by using the mathematic means analyzing the pair spots, getting the parameter of each region that those spots belong to. Through tracking the motion of the cardiac muscle from the diastole to the systole, arriving at the diastole again, we get the motion curve of each region, and then finish the motion analysing.Using these algorithms, this thesis make series of experiments, has proven that this method is feasible, and the motion parameter is credible.