Study On Motion Of Left Ventricle Based On Tagged MR Images

Using tagged Magnetic Resonance (MR) images, we can research deformation and motion of the heart muscle. The motion information of the Left Ventricle (LV) can reflect the pumping function of the heart, which can be used to diagnose many heart diseases. Therefore, analyzing motion of LV has become a hotspot in the relative researching filed. In this paper, we want to build a frame to analyze the left ventricle's motion based on the general tagged MR images, which is mainly composed of four parts: tag detecting and tracking, left ventricle segmentation, material markers extracting, and motion reconstruction and analysis. With this frame, we can get the 3D shape of the LV, the front displacement filed, the strain tensor of the heart muscle and global function parameters.This thesis restricts itself on the topic of tagged-MRI-based motion analysis. Some novel models and algorithms have been proposed based on the existed achievements of previous researchers, which are described as following:(1) It proposes a method using active contour model based on kinetic function to track the tag and uses three kinds of elastic potential energy, new internal energy and new image energy. The paper also changes the assumption that the initial velocity is zero and takes the optical flow as the velocity, which improves the veracity of the arithmetic. The method is tested on several sequences of cardiac systole MRI. The result indicates the method is effective.(2) It brings out a new tag tracking method based on Bayesian statistical approach. The method uses the grid model as the basic structure. Firstly, the method looks the grid nodes' positions as a Markov Random Field (MRF) model, and uses the EM algorithm to classify the nodes into two sorts by whether the node is inside the ventricles. Then, the method designs different prior distributions and likelihood function for different sorts based the function of which perform in the tracking progress. The method utilizes the Iterated Conditional Modes (ICM) algorithm to perform a Maximum A Posterior (MAP) estimate. The method is tested on several sequences of cardiac systole MRI. The result shows the method can exactly classify the grid nodes and track the SPAMM tag lines. Because of taking the gird model's Markov character into consideration, the grid can keep its topological shape during tracking progress.(3) A coupled polar active contour model is developed to simultaneity segment the 2D inside and outside edges of LV in Short Axis (SA) images. Using the polar model, the...