Clinically Feasible Magnetic-Resonance-Spectroscopic-Imaging Through Optimized Spatial-Spectral Encoding and Artificial Intelligence

Magnetic Resonance Spectroscopic Imaging (MRSI) simultaneously performs the functions of both Magnetic Resonance Imaging (MRI) and Magnetic Resonance Spectroscopy (MRS) in a single instance of data collection, but, in turn, it suffers from poorer voxel resolutions, longer scan durations, and lower Signal-to-Noise Ratios (SNR).

This work covers three potential contributions to the field of accelerated Magnetic Resonance Spectroscopic Imaging (MRSI) acquisition. In order, these are test-retest reproducibility of reduced Field-of-View Density-Weighted Concentric-Ring-Trajectory (rFOV-DW-CRT) MRSI, accelerated reconstruction through the interpolation of missing points of K-space through a Deep-Neural-Network (DNN), and the exploitation of the interaction between the pulse sequence and the DNN used for image reconstruction to propose a novel K-space acquisition trajectory which is capable of acquiring all of K-space in a single-shot. The latter of the three potentially offers the most in terms of additional factors of acceleration, but, as of writing, it is still in its infancy and will require significantly more resources to be invested before its full potential can be unlocked. Regardless, a proof-of-concept has been included as the final main chapter to demonstrate the plausibility of its practical utility.