Magnetic Resonance Gradient Echo Phase Imaging as a Means of Detecting Alterations in the Tissue Microarchitecture of the Human Corpus Callosum

Susceptibility sensitive MRI GRadient Echo (GRE) imaging produces two images, one of magnitude and one of phase. Most imaging studies ignore the phase image, despite the fact that the phase image has stronger contrast in some regions of the brain and therefore can show more detail than the magnitude image, primarily because there is no standard for acquisition, post-processing and display. Additionally, the source of this contrast is not fully understood. Therefore, a better understanding of phase images has the potential to elucidate structures that are currently not visible in the magnitude image, which could be useful in a variety of clinical applications.;In order to better understand the phase image, this dissertation first used computer simulation to model the expected phase signal in a white matter fiber tract. A two compartment susceptibility model was programmed using fiber tract specific information from histopathology studies on the human corpus callosum. This model was built on the discoveries of previous studies which indicated that bulk susceptibility may be the dominant source of phase contrast. The strength of this model was assessed by comparing predicted phase shifts for the computer-modeled corpus callosum fiber tracts with values taken from filtered phase images of a healthy subject brain. Additionally, T2* values were computed from the model and compared with measured values. To assess the effect of filtering on the phase, experiments using a homogeneous spherical phantom were performed. Finally, the ability of phase to classify mild Traumatic Brain Injury patients with persistent behavioral symptoms with no visible microhemorrhages, a population expected to have altered white matter microstructure, was explored using quantitative region-of-interest analysis.;This work shows that computer simulations using a simple two compartment susceptibility model that assumes a susceptibility of myelin to be equal to that measured for cholesterol can reasonably predict measured phase in healthy human subjects in the corpus callosum. In addition, these simulations predict statistically different phase for each of five region of the corpus callosum, which are also observed in the MRI susceptibility sensitive GRE phase images. Simulation also predicts T2* values within a factor of two of what is measured experimentally from the magnitude images. When images are compared using a consistent filtering method, measurable regional differences across the corpus callosum can be detected. Finally, this work shows that regional differences between two patient populations (mild Traumatic Brain Injury with persistent behavioral symptoms and healthy age-matched controls) can be detected in the absence of visible microhemorrhages, suggesting potential clinical application for susceptibility sensitive phase imaging.