Files in this item



application/pdfCHRISTODOULOU-DISSERTATION-2015.pdf (6MB)
(no description provided)PDF


Title:A subspace approach to accelerated cardiovascular magnetic resonance imaging
Author(s):Christodoulou, Anthony Glenn
Director of Research:Liang, Zhi-Pei
Doctoral Committee Chair(s):Liang, Zhi-Pei
Doctoral Committee Member(s):Boppart, Stephen A.; Do, Minh N.; Sutton, Bradley P.
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Magnetic resonance imaging
cardiovascular imaging
cardiac imaging
Abstract:Magnetic resonance imaging (MRI) is a uniquely flexible tool for imaging the heart, as it has the potential to perform a significant number of structural and functional cardiovascular assessments. However, the low imaging speed of MRI has limited its clinical application. The assessments that are currently performed in a clinical setting are typically done using gated methodologies, which are complicated by respiration and fail for patients with cardiac arrhythmias. This dissertation describes a subspace approach to accelerate cardiovascular MRI, freeing cardiac MRI from gating techniques and enabling whole-heart 3D dynamic imaging for multiple simultaneous assessments. This imaging approach comprises developments in image modeling, data acquisition, and image reconstruction. A spatiotemporal image model is designed to represent the particular subspace structure of cardiovascular images. The data acquisition development is composed of: a) a sampling strategy which allows integration of the subspace model, parallel imaging, and sparse modeling; b) a novel pulse sequence implementing "self-navigation" for collecting both auxiliary data (for temporal subspace estimation) and imaging data after every excitation; and c) k-space trajectory evaluation and design, replacing Cartesian trajectories which are highly sensitive to readout direction. The image reconstruction work centers on the integration of the subspace model, sensitivity encoding (for parallel imaging), and sparse modeling into one optimization problem; evaluations of strategies for regularizing the image model, adaptively enforcing model order, and for estimating sensitivity maps are also included. The approach is evaluated through simulations on numerical cardiac phantoms and in vivo experiments in human, rat, and mouse subjects. Multiple cardiovascular applications are demonstrated: cine imaging, first-pass myocardial perfusion imaging, late gadolinium enhancement imaging, extracellular volume fraction mapping, and labeled immune cell imaging. Experimental results include human cine images up to 22 fps and 1.0 mm × 1.0 mm spatial resolution, mouse cine images up to 97 fps and 0.12 mm × 0.12 mm spatial resolution, rat images at 74 fps and 0.31 mm × 0.31 mm × 1.0 mm spatial resolution (capturing wall motion, first-pass myocardial perfusion, and late gadolinium enhancement in a single scan), multi-contrast rat images (for extracellular volume fraction mapping) up to 50 fps and 0.42 mm × 0.42 mm × 1.0 mm spatial resolution, and rat images at 98 fps and 0.16 mm × 0.16 mm spatial resolution (depicting labeled immune cells). The end result is an imaging approach capable of ungated, whole-heart 3D cardiovascular MRI in high spatiotemporal resolution. Images can be obtained even for patients with irregular heartbeats, and both cardiac motion and aperiodic contrast dynamics can be imaged in a single scan. These capabilities should enhance the utility of cardiovascular MRI, allowing comprehensive evaluation of the heart.
Issue Date:2015-04-17
Rights Information:Copyright 2015 Anthony Christodoulou
Date Available in IDEALS:2015-07-22
Date Deposited:May 2015

This item appears in the following Collection(s)

Item Statistics