Reconstruction algorithms for MRI

Abstract

This dissertation presents image reconstruction algorithms for Magnetic Resonance Imaging (MRI) that aims to increase the imaging efficiency. Algorithms that reduce imaging time without sacrificing the image quality and mitigate image artifacts are proposed. The goal of increasing the MR efficiency is investigated across multiple imaging techniques: structural imaging with multiple contrasts preparations, Diffusion Spectrum Imaging (DSI), Chemical Shift Imaging (CSI), and Quantitative Susceptibility Mapping (QSM). The main theme connecting the proposed methods is the utilization of prior knowledge on the reconstructed signal. This prior often presents itself in the form of sparsity with respect to either a prespecified or learned signal transformation.