Fundamental and practical limits to image acceleration in parallel magnetic resonance imaging

Abstract

Imaging speed in conventional magnetic resonance imaging (MRI) is limited by the performance of magnetic field gradients and the rate of power deposition in tissue. Parallel MRI techniques overcome these constraints by exploiting information stored within the spatial sensitivity patterns of radiofrequency detector arrays to substitute for some of the spatial information that would normally be obtained using magnetic field gradients. Parallel MRI strategies have been applied clinically to increase patient comfort, enhance spatial resolution, expand anatomical coverage, and reduce image artifacts. The effectiveness of parallel MRI techniques is largely determined by the amount of spatial information that is stored in the detector coil sensitivities. This dissertation investigates the spatial encoding properties of coil arrays from three practical and fundamental perspectives. First, a novel array design is presented that enables spatial encoding in multiple directions simultaneously. Second, the impact of inductive coupling between array elements in parallel MRI is investigated theoretically and experimentally. Finally, electromagnetic calculations are described that permit computation of the ultimate intrinsic signal-to-noise ratio available to any physically realizable coil array for parallel MR. These calculations help to establish fundamental limits to the image accelerations that may be achieved using parallel MRI techniques. These limits are intrinsically related to the wavelengths of the electromagnetic fields at MR imaging frequencies. The sensitivity patterns that correspond to the ultimate intrinsic SNR also represent potential starting points for new coil designs.