This Is AuburnElectronic Theses and Dissertations

Four-Chamber Endocardial Surface Reconstruction from MR Images and Application to Geometry and Mechanics Analysis on Healthy Subjects and Mitral Regurgitation Patients




Zhang, Xiaoxia

Type of Degree

PhD Dissertation


Electrical and Computer Engineering


Heart failure is a common cause of hospitalization and the leading cause of death in the world [1]. It is usually secondary to a primary cardiovascular disease (CVD) and it is estimated that by 2030, about 23.6 million people will die from a type of CVD. CVD is a problem that crosses both gender and ethnicity, and worsens with age. Therefore, effective diagnosis and early treatment of CVDs are significantly important. Medical imaging techniques, providing useful information in diagnosis, visualization and monitoring, have been dramatically developed and improved. Magnetic resonance imaging (MRI) is one such technique, becoming a leading imaging modality for advanced clinical research, drug studies and patient management due to its high image resolution, minimal invasion and reproducibility compared with other conventional imaging techniques. Functional and geometric analysis of the left ventricle (LV) of the human heart have been well developed using MRI, however, analysis of the LV has its limitations in the presence of some cardiovascular disease [2-4]. Research in shape development and functional analysis of right ventricle (RV), left atrium (LA) and right atrium (RA) are limited by their complicated geometries. LA functions along with volume-time curves provide powerful and incremental information on LA performance and are considered as predictors of cardiovascular morbidity and mortality [5-8]. However, in current clinical practice, changes in the LA geometry and function have been defined using long-axis views, which use a shape assumption, and the result depends on the orientation of the long axis views. Similar problems occur in the evaluation of RV and RA geometry and function. RV functional parameters are major markers in a variety of cardiovascular diseases, including pulmonary hypertension, ventricular ischemia or infarction, pulmonary or tricuspid valvular heart diseases [9]. RA volume and size could be early markers of RV dysfunction. Therefore, the analysis of all four chambers and the interaction among them can provide extra information that might suggest subtle abnormalities of cardiac function in patients with normal LV functions. Mitral valve regurgitation (MR) is one of the causes of heart failure, in which the mitral valve does not operate properly [20]. Correct identification of MR surgical candidates and optimizing the timing of surgery are key in patient management. Currently, the most common surgery guidelines are based on LV function and diameter, which is LV ejection fraction < 60% and/or LV end-systolic dimension >40 mm [22]. However, patients are reported not receiving timely surgery followed the surgery guidelines with worse outcomes and increased morbidity and mortality [96] [99] [100]. As we described, all four chambers have clinical potentials independently and mutually. Therefore, we are motivated to explore more parameters in volume function, geometry and mechanics of LA, RA and RV in addition to LV for optimized surgery timing. In this dissertation, we first developed a novel 4D (temporal and spatial) surface fitting algorithm to reconstruct subject-specific heart surface of all four chambers from cardiac MRI images. The reconstructed surface had vertex-to-vertex correspondence, which provided more accurate measurements and comprehensive analysis of the heart. This algorithm was validated on two groups through three aspects, including heart volume functions, mechanics and geometries. The two groups consist of a normal subject group with young and old subjects, and a mitral regurgitation patient group with age-matched control subjects. The two groups were imaged with cine MRI and endocardial contours were semi-automatically identified with a previously developed algorithm implemented in Matlab [77]. Surfaces of all four chambers were reconstructed using the developed algorithm to evaluate the changes in volume functions, chamber geometries and wall motions during the entire imaged cardiac cycle. This procedure was used to investigate the remodeling in functional and geometric parameters of all four chambers versus age and between MR patients and age-matched controls, and the effect of valve repair surgery after 6 months.