Molecular Simulations of Diffusion-Tensor MRI

Date of Award

Spring 5-15-2014

Author's School

Graduate School of Arts and Sciences

Author's Department


Degree Name

Doctor of Philosophy (PhD)

Degree Type



The diffusion of water molecules in white matter is highly anisotropic because of the unique anatomic features of human brain. In the presence of a strong magnetic field gradient, diffusion of water molecules causes spin dephasing, which could be directly read out by the signal intensity in diffusion weighted MRI. A diffusion tensor MRI (DT-MRI) method with a high-sensitivity tetrahedral-orthogonal sampling pattern was analyzed. It was confirmed that anisotropy and diffusivity along and across white matters could be measured accurately.

A computational model was introduced to simulate anisotropic diffusion of water in brain white matter during DT-MRI. White matter is modeled as an ordered array of cylindrical axons where the wall of axonal membranes is a barrier to water movement. By matching the simulation results with the experimental MRI data from human association white matter, we could determine the self-consistent values for parameters describing the underlying tissue microstructure and physical properties (e.g. axon diameter, spacing and permeability). Different from conventional analytic and numerical approaches, we introduced a parameter "τ" to describe the phenomenon of water molecules detained on the membrane surface or in the membranes. The axonal membrane and τ parameter result in the non-linear coupling between water diffusive motions in orthogonal directions. This novel simulation framework may prove useful for theoretical understating of the widely observed pathological changes in water diffusivity in animal models and human disease. Additionally, in order to predict the effect of microstructural parameters on DT-MRI measurements, we used exponential, linear and other model equations to fit our result.

We also proposed three-compartment models (intracellular, extracellular and myelin sheath) to facilitate a detailed delineation of different regions of the human brains that have different diffusivities. Based on the simulation model, we could give a prospective speculation of the microstructure of that brain region without invasive biopsy or histological analysis. In the end, we further applied this model to the study of autism patients, to discover the disease substrate of the axons in this pathological state.


English (en)

Chair and Committee

Thomas E Conturo

Committee Members

Ralf Wessel, Samuel Achilefu, Anders Carlsson, Yanmei Wang, Li Yang


Permanent URL: https://doi.org/10.7936/K7154F1M

This document is currently not available here.