Files in this item



application/pdfCERJANIC-DISSERTATION-2020.pdf (6MB)Restricted Access
(no description provided)PDF


Title:Advancing the intravoxel incoherent motion model as a tool for quantifying brain microvascular health
Author(s):Cerjanic, Alexander
Director of Research:Sutton, Bradley P
Doctoral Committee Chair(s):Sutton, Bradley P
Doctoral Committee Member(s):Liang, Zhi-Pei; Cohen, Neal; Llano, Daniel
Department / Program:Bioengineering
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Magnetic Resonance Imaging
Diffusion weighted imaging
microvascular blood flow
intravoxel incoherent motion
Abstract:Quantification of the state of the microvasculature of the human brain is currently only feasible via invasive histological methods using light microsocopy. In order to correlate the state of the microvasculature of the brain with functional measures such as blood flow or cognition in humans, noninvasive techniques are required. One model for measuring microvascular blood flow is the intravoxel incoherent motion (IVIM) model of pseudodiffusive blood flow. In this work, a set of IVIM-derived microvascular biomarkers are pro- posed to quantify the mean microvascular blood velocity and mean microvessel length via diffusion weighted imaging (DWI). Very long mixing time stimulated acquisition mode diffusion weighted magnetic resonance imaging pulse sequences were developed and implemented to acquire the long diffusion time data required to derive the proposed microvascular biomarkers. To accelerate the computationally burdensome iterative reconstructions used in this work, a novel GPU accelerated toolkit, PowerGrid, was developed to implement the reconstructions. Speed ups in excess of 17x were obtained via PowerGrid on commercial GPUs. To address the challenges of fitting IVIM parameters reliably, two strategies were examined. The first was the application of a combination diffusion tensor-IVIM model as applied to white matter and deep gray matter in the human brain. The second was a novel estimator using a reducing variance at the cost of increased bias by applying bootstrap aggregation from field of machine learning. The bootstrap aggregation estimator was used in a hypercapnia paradigm to attempt to demonstrate the proposed biomarkers in vivo. Finally, future work towards further validating and developing the proposed biomarkers in humans is described.
Issue Date:2020-12-02
Rights Information:Copyright 2020 Alexander Cerjanic
Date Available in IDEALS:2021-03-05
Date Deposited:2020-12

This item appears in the following Collection(s)

Item Statistics