Doctor of Philosophy

dissertationDiffusion tensor MRI (DT-MRI or DTI) has been proven useful for characterizing biological tissue microstructure, with the majority of DTI studies having been performed previously in the brain. Other studies have shown that changes in DTI parameters are detectable in the presence of cardiac pathology, recovery, and development, and provide insight into the microstructural mechanisms of these processes. However, the technical challenges of implementing cardiac DTI in vivo, including prohibitive scan times inherent to DTI and measuring small-scale diffusion in the beating heart, have limited its widespread usage. This research aims to address these technical challenges by: (1) formulating a model-based reconstruction algorithm to accurately estimate DTI parameters directly from fewer MRI measurements and (2) designing novel diffusion encoding MRI pulse sequences that compensate for the higher-order motion of the beating heart. The model-based reconstruction method was tested on undersampled DTI data and its performance was compared against other state-of-the-art reconstruction algorithms. Model-based reconstruction was shown to produce DTI parameter maps with less blurring and noise and to estimate global DTI parameters more accurately than alternative methods. Through numerical simulations and experimental demonstrations in live rats, higher-order motion compensated diffusion-encoding was shown to successfully eliminate signal loss due to motion, which in turn produced data of sufficient quality to accurately estimate DTI parameters, such as fiber helix angle. Ultimately, the model-based reconstruction and higher-order motion compensation methods were combined to characterize changes in the cardiac microstructure in a rat model with inducible arterial hypertension in order to demonstrate the ability of cardiac DTI to detect pathological changes in living myocardium