Novel Techniques for Measuring Cardiac Shape and Mechanics with Magnetic Resonance Imaging
Type of Degreedissertation
MetadataShow full item record
Magnetic resonance imaging (MRI) is recognized as a non-invasive cardiac imaging test that provides the most complete, detailed and precise images. The rapid technical development of MRI makes it possible to evaluate cardiac function and morphology reliably in a good spatial and temporal resolution for clinical daily use. In this dissertation, three aspects of cardiac MRI are investigated: 1) the measurement of right ventricular (RV) functional parameters such as volumes, mass and ejection fraction (EF) from cardiac MRI sequences, 2) a geometric explanation for an observed preservation of left ventricular (LV) EF in the presence of reduced circumferential strain in patients with hypertension, and 3) a novel technique for accelerating the acquisition of cardiac image sequences by exploiting the redundancies in static regions in an image sequence. While the measurement of LV functional parameters has been extensively studied, the measurement of RV functional parameters has received relatively little attention. A novel protocol for RV volumetric analysis from routine clinical cardiac MRI is proposed and validated. Hypertension is a disease where arterial pressure is elevated and the heart compensates by increasing LV wall thickness. Previous clinical studies have observed that in hypertensive patients, circumferential strain is reduced while LVEF is normal. We derive a simple, analytical framework to predict LVEF with wall thickness, radius-to-wall thickness ratio, longitudinal shortening quantified from cine MRI and circumferential strain measured by tagged MRI over a wide range of heart conditions. Our work suggests that the preserved LVEF in hypertension despite the presence of reduced circumferential strain is primarily due to the geometric effects of concentric remodeling – relative wall thickness increases without greater LV mass. The heart takes approximately 10% of the cardiac MR image. Its surrounding tissues are imaged to avoid aliasing. We propose a new imaging method, Noquist with arbitrary dynamic region (NADR), to accelerate the acquisition by reconstructing the pixels in those surrounding tissues once for the entire image sequence. The experimental results demonstrate that NADR is capable of a 56% reduction in k-t space data with negligible reconstruction error.