Radiofrequency Pulse Design for Magnetic Resonance Imaging at Ultra-High and Ultra-Low Field Strengths

Abstract

Magnetic resonance imaging is currently experiencing groundbreaking advancements at high- and low- magnetic field strength extremes. Ultra-high field scanners at strengths of 7 Tesla and above are producing groundbreaking clinical and scientific advances as a result of the improvements in image quality that high-field systems can provide. At the ultra-low field extreme, imaging systems are being developed with magnets strengths in the tens of milliteslas, which are low-cost, portable, and capable of bringing advanced medical imaging to new populations with critical need. At both of these cutting edges of the MRI field, advanced radiofrequency pulse design allows us to push the limits of image quality and encoding. In this thesis, we present new software tools, optimization methods, and applications of MR physics for the design of radiofrequency pulses. We first present a comprehensive open-source software toolbox for RF pulse design. Using this software package, we demonstrate for the first time that the Bloch-Siegert shift may be used to localize excitation, and develop an RF pulse class using this principle, which enables the elimination of the expensive and bulky slice-select B0 gradients used in conventional imaging systems. Finally, we develop stochastic methods for RF pulse design by the Gerchberg-Saxton phase retrieval algorithm, demonstrating that inexact alternating projections may be used to explore the feasible points of the problem and find low-cost pulse solutions. We demonstrate the impact of using this stochastic pulse design algorithm on image quality at 7T.