基于带标记核磁共振图像的左心室运动分析-计算机应用技术专业论文.docxVIP

华中科技大学硕士学位论文 华 中 科 技 大 学 硕 士 学 位 论 文 II II Abstract The Tagged Magnatic Resonance (TMR) sequence provides an effective way for the analysis of complex cardiac motion. Current clinical diagnoses mainly rely on naked eye observation or manual calibration, which means heavy workload and imprecise evaluation criteria. Therefore, it’s necessary to do qualitative and quantitative analysis on cardiac motion by means of computer image processing method, and then provides reference data for doctors to diagnose. First of all, dividing the left ventricular internal and external wall using threshold method and bilinear Snake model method, so that finding out the interested myocardial region, and then getting the cardiac image sequence for next steps. Then, enhancing cardiac markers by Gabor filtering method according to the periodic characteristics of mark of TMR images. Above based, designing a movement grid energy model in order to extract and trace the tag line intersections of cardiac motion image. The initial position of the movement grid nodes (The intersection of the tag line) is decided by using gray projection method on the first frame, and the process of minimizing the Energy of Grid Mesh (EGM) determines the subsequent state of grid. EGM includes the structure energy of maintaining topology and mutual positional relationship, the characteristic energy of gradation grayscale trait of tagged intersection and the limited energy of motion relationship between adjacent frames. Choosing greedy algorithm to solve minimum value of the total energy so as to obtain the final position of tagged intersection. Finally, getting the dense displacement field of left ventricular myocardial area by computing the motion displacement of tagged intersection between the left ventricular internal and external walls. In the experiment, using image sequences with complete cardiac period from 15 cases as data, which are provided by Tongji Medical College of Huazhong University of III III Scie