Files in this item



application/pdfNAUGHTON-DISSERTATION-2019.pdf (20MB)Restricted Access
(no description provided)PDF


Title:Diffusion-weighted MRI of skeletal muscle: Estimation of microstructural parameters
Author(s):Naughton, Noel Martin
Director of Research:Georgiadis, John
Doctoral Committee Chair(s):Georgiadis, John
Doctoral Committee Member(s):Liang, Zhi-Pei; Boppart, Marni; Gazzola, Mattia; Kersh, Mariana
Department / Program:Mechanical Sci & Engineering
Discipline:Mechanical Engineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
skeletal muscle
lattice Boltzmann method
inverse problem
Gaussian Process
Abstract:Establishing effective treatments to improve muscle quality in elderly populations is a key component of improving quality of life for the elderly. Changes in muscle’s microstructural organization impact its functional ability to produce mechanical work. But function follows structure, so the ability to accurately and non-invasively measure the microstructural organization of skeletal muscle is an important step in developing quantitative and repeatable measures of muscle efficiency. Diffusion-weighted magnetic resonance imaging (dMRI) enables the non-invasive probing of the microstructural organization of tissue by sensitizing the MR signal to the micron length scale displacements of water molecules within the tissue, which then probe the geometrical features of the microstructure. This work develops a framework within which dMRI measurements are used to non-invasively estimate microstructural parameters of skeletal muscle. This framework consists of developing a numerical solver for dMRI physics in multiphase tissues using the lattice Boltzmann method (LBM) to characterize the interplay between the microstructural organization of skeletal muscle and dMRI physics. This characterization consists of a comparison of the LBM model with analytical two-compartment models as well as the completion of a global sensitivity study of the effects of microstructural and pulse parameters on the dMRI signal. The results from the sensitivity analysis allow for the creation of a polynomial meta-model, which accurately connects microstructural parameter variations to dMRI metrics while substantially decreasing the computational cost compared to the full LBM model. Using this meta-model, the inversion problem of estimating microstructural parameters from dMRI measurements is addressed through the development of a Gaussian process regression (GPR) model. The GPR model produces a mean estimate as well as a confidence interval for each voxel, allowing confidence bounds to be developed for the prediction of any microstructural parameter. The GPR model is shown to accurately estimate microstructural parameters when compared with noise free simulations and to estimate the correct values within its confidence intervals for noisy data. Initial validation experiments show that the GPR model accurately estimates diameter and volume fraction when compared with experimental data. The LBM model is also used to simulate histologically-informed domains allowing the impact of the extracellular matrix on the dMRI signal to be characterized. Finally, a plausible hypothesis of the source of transverse anisotropy in skeletal muscle as originating from elliptical organization of fascicles is proposed.
Issue Date:2019-11-25
Rights Information:Copyright 2019 Noel Naughton
Date Available in IDEALS:2020-03-02
Date Deposited:2019-12

This item appears in the following Collection(s)

Item Statistics