Experimental Techniques and Image Reconstruction for Magnetic Resonance Imaging with Inhomogeneous Fields

Abstract

Magnetic resonance imaging is quite sensitive to experimental imperfections, necessitating extremely expensive electrical infrastructure and design requirements to permit high-quality experiments to be performed. By relaxing the sensitivity to imperfection, the entire system can be made less expensive and more accessible by shrinking the magnet generating the polarizing field. Decreasing the magnet size relative to the bore increases the polarizing field inhomogeneity. Moreover, current progress in MRI at ultra-high field (greater than or equal to 7T) is pushing the limits of conventional MRI methods, as field inhomogeneity increases with field strength. Hence, while many of the methods herein were developed with a small magnet in mind, they also apply at ultra-high field. The appeal of ultra-high field is increased detection sensitivity such that ever-smaller structures may be imaged in animals and humans. The primary goal of this work is to extend the current ability of magnetic resonance imaging to tolerate a large degree of spatial variation in both the transmit and polarizing fields involved. A novel method of decreasing radiofrequency pulse duration for multidimensional pulses is presented, rendering them more robust to field inhomogeneity. Furthermore, this method is leveraged to accelerate data acquisition. A new imaging sequence for quantitative determination of transverse relaxation rates is presented, which tolerates large variations in both the transmit and polarizing magnetic fields, as is often found when imaging with iron-oxide nanoparticles and/or at ultrahigh field. Finally, a computationally efficient approach for spatiotemporally-encoded image reconstruction is presented, which is inherently robust to field inhomogeneity.