Date of Award

Winter 12-15-2016

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Doctor of Philosophy (PhD)

Degree Type



Although x-ray computed tomography (CT) is a gold standard for pulmonary imaging, it has high ionizing radiation, which puts patients at greater risk of cancer, particularly in a longitudinal study with cumulative doses. Magnetic resonance imaging (MRI) doesn't involve exposure to ionizing radiation and is especially useful for visualizing soft tissues and organs such as ligaments, cartilage, brain, and heart. Many efforts have been made to apply MRI to study lung function and structure of both humans and animals. However, lung is a unique organ and is very different from other solid organs like the heart and brain due to its complex air-tissue interleaved structure. The magnetic susceptibility differences at the air-tissue interfaces result in very short T2* (~ 1 ms) of lung parenchyma, which is even shorter in small-animal MRI (often at higher field) than in human MRI. Both low proton density and short T2* of lung parenchyma are challenges for pulmonary imaging via MRI because they lead to low signal-to-noise ratio (SNR) in images with traditional Cartesian methods that necessitate longer echo times (≥ 1 ms). This dissertation reports the work of optimizing pulmonary MRI techniques by minimizing the negative effects of low proton density and short T2* of murine lung parenchyma, and the application of these techniques to imaging murine lung. Specifically, echo time (TE) in the Cartesian sequence is minimized, by simultaneous slice select rephasing, phase encoding and read dephasing gradients, in addition to partial Fourier imaging, to reduce signal loss due to T2* relaxation. Radial imaging techniques, often called ultra-short echo-time MRI or UTE MRI, with much shorter time between excitation and data acquisition, were also developed and optimized for pulmonary imaging. Offline reconstruction for UTE data was developed on a Linux system to regrid the non-Cartesian (radial in this dissertation) k-space data for fast Fourier transform. Slabselected UTE was created to fit the field-of-view (FOV) to the imaged lung without fold-in aliasing, which increases TE slightly compared to non-slab-selected UTE. To further reduce TE as well as fit the FOV to the lung without aliasing, UTE with ellipsoidal k-space coverage was developed, which increases resolution and decreases acquisition time. Taking into account T2* effects, point spread function (PSF) analysis was performed to determine the optimal acquisition time for maximal single-voxel SNR. Retrospective self-gating UTE was developed to avoid the use of a ventilator (which may cause lung injury) and to avoid possible prospective gating errors caused by abrupt body motion. Cartesian gradient-recalled-echo imaging (GRE) was first applied to monitor acute cellular rejection in lung transplantation. By repeated imaging in the same animals, both parenchymal signal and lung compliance were measured and were able to detect rejection in the allograft lung. GRE was also used to monitor chronic cellular rejection in a transgenic mouse model after lung transplantation. In addition to parenchymal signal and lung compliance, the percentage of highdensity lung parenchyma was defined and measured to detect chronic rejection. This represents one of the first times quantitative pulmonary MRI has been performed. For 3D radial UTE MRI, 2D golden means (1) were used to determine the direction of radial spokes in k-space, resulting in pseudo-random angular sampling of spherical k-space coverage. Ellipsoidal k-space coverage was generated by expanding spherical coverage to create an ellipsoid in k-space. UTE MRI with ellipsoidal k-space coverage was performed to image healthy mice and phantoms, showing reduced FOV and enhanced in-plane resolution compared to regular UTE. With this modified UTE, T2* of lung parenchyma was measured by an interleaved multi-TE strategy, and T1 of lung parenchyma was measured by a limited flip angle method (2). Retrospective self-gating UTE with ellipsoidal k-space coverage was utilized to monitor the progression of pulmonary fibrosis in a transforming growth factor (TGF)-α transgenic mouse model and compared with histology and pulmonary mechanics. Lung fibrosis progression was not only visualized by MRI images, but also quantified and tracked by the MRIderived lung function parameters like mean lung parenchyma signal, high-density lung volume percentage, and tidal volume. MRI-derived lung function parameters were strongly correlated with the findings of pulmonary mechanics and histology in measuring fibrotic burden. This dissertation demonstrates new techniques and optimizations in GRE and UTE MRI that are employed to minimize TE and image murine lungs to assess lung function and structure and monitor the time course of lung diseases. Importantly, the ability to longitudinally image individual animals by these MRI techniques minimizes the number of animals required in preclinical studies and increases the statistical power of future experiments as each animal can serve at its own control.


English (en)

Chair and Committee

Jason C. Woods

Committee Members

Mark S. Conradi, Alexander Barnes, Francesc Ferrer, James G. Miller


Permanent URL: https://doi.org/10.7936/K7CF9NH4