| The topic of water diffusion is of great interest in current magnetic resonance imaging (MRI) research. With the observation that acute changes in the “apparent diffusion coefficient” (ADC) correlate with the onset of cytotoxic edema (cell swelling) associated with stroke onset, a thrust to understand the basis of the signal changes has been made. Questions surrounding this topic are numerous: Are cellular size changes as influential as changes in extra-cellular tortuosity? How do changes in cellular permeability affect the signal decay? Can we derive a system for detecting separate physiologic events associated with decreased neural viability? Can we tailor pulse sequences to be sensitive to selected neural events?; The recent observation of an ADC decrease with functional stimulation in humans leads to the same sorts of questions, since the origin of the signal is thought to be essentially neural (1). It is exciting to think that an MRI technique, the images of which are reconstructed to contain information on the order of millimeters, is sensitive to such sub-millimeter processes as restricted diffusion.; This dissertation is based on understanding the direct relationship between neural structure, both microscopic and macroscopic, and the diffusion-weighted MRI (or DWI) signal. A need for this work is apparent when one considers the amount of controversy regarding the source of such signals. It has been hypothesized that signals arising from both intra- and extra-cellular water combine to form the characteristic non-exponential decay curve as a function of diffusion weighting (b-value). Such a cortical model would be consistent with current thoughts about local neural changes during disease and function, and would conveniently link the sub-voxel sensitivity of DWI with macroscopic changes observed with “anatomic” methods such as T1 and T2-weighted imaging.; One may alternatively hypothesize that the source of non-exponential diffusion is somehow related to two pools of water: one in almost constant contact with boundary proteins by ionic bonds, and one which either interacts with the former “bound” pool by hydrogen bonds or proton exchange, or is otherwise free to diffuse in both intra- and extra-cellular spaces.; This thesis provides support of the latter hypothesis, as experiments on non-biologic protein samples are shown to exhibit non-exponential signal attenuation similar to that seen in our in-vivo rat studies. In addition, we develop here a quantitative analysis with the stretched-exponential model, which fits the experimental data well and is sensitive to the width of the distribution of the diffusion rates of an unknown number of intra-voxel proton pools. The significance of these results lies in their new predictions about the source of DWI changes due to tissue pathology. |
