Characterization of harmonic phase MRI: Theory, simulations, and applications.

摘要

Motion of a human heart is a sensitive indicator of heart disease. Magnetic resonance (MR) tagging has emerged as the gold standard for visualizing the detailed motion of a beating heart. Motion is measured from tagged MR images using Harmonic Phase (HARP) MRI, which is an image processing method that allows for fast automated processing of tagged MR images. This dissertation investigates the ability of HARP-MRI to provide significant diagnostic information. The goal of this dissertation is to translate this qualitative clinical question into quantifiable engineering problems and to solve them.; In this dissertation, we characterize both the optimal and peak performances of HARP-MRI. We study the optimal performance by investigating HARP-MRI's accuracy and by deriving protocols for its optimal use. We characterize the peak performance of HARP-MRI by deriving three performance measures: ranges of measurable motion, spatial resolution, and the ability to detect small abnormalities. We also explain the presence of artifacts in the outputs of HARP-MRI, and propose an algorithm to reduce them. Finally, we use our understanding of peak performance of HARP-MRI to adapt it to track the motion of the tongue during speech.