Modeling Left Ventricle Wall Motion Using Tagged Magnetic Resonance Imaging

Abstract

A two-parameter computational model is proposed for the study of the regional motion of the left ventricle (LV) wall using tagged magnetic resonance imaging (tMRI) data. In this model, the LV wall motion is mathematically decomposed into two components, an isotropic deformation of the LV wall tissues along the short axis of LV and a non-uniform rotation of the tissues along the long axis of LV. The deformation and rotation parameters are determined by fitting the model to tMRI images of the short-axis planes of the LV wall. To validate this model, the tMRI images of the LV wall at the midventricular, apical and basal levels from eight subjects including healthy human, healthy and diabetic rats were studied. The result showed that this model is very effective in studying the LV wall motion and function in both small animals and human. With this model, the torsion, strain, and strain rate of the LV wall tissues can easily be calculated analytically at different phases of a cardiac cycle. It was found that the ratio of the torsion at endocardium to the torsion at epicardium is a constant during the cardiac motion even though the torsion varies with time significantly. The value of this constant of motion equals the ratio of the end diastolic radii of the LV wall at endocardium and epicardium and, therefore, is approximately the same for rats and human. This dissertation also includes a study of effects of exercise training on diabetic heart. In that study, global cardiac functions were measured using high field MRI and it was found that the global functions of diabetic heart could be improved by the training.