Quantitative Assessment of Magnetic Resonance Spectroscopy Data Reconstruction Methods: Region-of-Interest Averaging and Spectral Localization by Imaging

Abstract

The aim of this dissertation was to compare two magnetic resonance spectroscopy (MRS) localization techniques: Fourier based region-of-interest (ROI) averaging, and the non-Fourier based spectral localization by imaging (SLIM). Unlike ROI-averaging, SLIM provides a technique for calculating the metabolite spectra of a compartmental region without the need for averaging voxels of spectral data to estimate that region. Because of this, SLIM has the potential to greatly reduce the acquisition time needed to acquire compartmental spectra. SLIM was processed over multiple k-space sizes and over an assortment of brain regions and then these results were compared to their equivalent ROI-averaged regions. The assorted k-space sizes were used to demonstrate SLIM operating with different amounts of available data, which was used to compare the process to ROI-averaging. The results of this study validate SLIM as a valuable localization tool that will shorten scan times and improve accuracy in spectral localization. The dissertation is divided into five main chapters: (1) Introduction, which addresses magnetic resonance background concepts and applications of MRS techniques; (2) Methods, which describes the processes involved in developing a programming pipeline designed to produce metabolite data for the localization techniques; (3) Results, which provides statistical measures of the localization methods; (4) Discussion, where comparisons were drawn from the datasets based on the results section; (5) Conclusions, which evaluates the thesis work and addresses possible research directions for the future.