Investigating microstructure in the brain using Diffusion Weighted Magnetic Resonance Imaging
Loading...
Citations
Altmetric:
Altmetric::
Associated Records
Authors
File Licence
Editors
Date
2014
Source Title
Journal Title
Journal ISSN
Volume Title
Publisher
Melbourne Authors
Affiliation
Electrical and Electronic Engineering
Document Type
Masters Research thesis
Subject Code
Citation
Access Status
Open Access
DOI
Access URL
Open Access URL
Open Access at PMC
Linked Resource URL
NHMRC Grant Code
ARC Grant Code
Description
© 2014 Dr. Seyed Hamed Yousefi Mesri
Abstract
The morphological structure of the brain plays an important role in studying neurodegenerative diseases and disorders. Despite the great importance of the investigation of the brain microstructure, there is not any direct and non-invasive imaging techniques available to measure the micro-structural tissue parameters. Conventional Magnetic Resonance Imaging (MRI) cannot provide images of the brain microstructure due to the limitations in the image resolution. Hence, there is a need to develop indirect techniques to measure the microscopic features of the brain tissue.
Although there is not any technique developed to directly investigate the brain microstructure, studying the diffusion of water molecules in the tissue can reveal information about the underlying tissue microstructure and hence can provide an indirect method to capture the micro-structural information. Therefore, Diffusion Weighted Magnetic Resonance Imaging (DWI), which measures the diffusion of water molecules, has the capability to be used as a non-invasive method to study the microstructure of the brain and has been used in the past decade to achieve this purpose. The measurements based on DWI utilise model-based techniques to deduce the information about the tissue microstructure from the signal, which inherently suffer from bias error (as the data may not be derived from a member of the model class). However, they offer superior performance in variance minimisation, and allow the inference of parameters based on limited datasets.
In this work, we review a number of the existing model-based techniques developed to identify micro-structural parameters from DWI datasets. We present a critical analysis of different techniques in terms of their assumptions, advantages and drawbacks. In the next step, we compare the performance of different models in terms of their minimum uncertainty level. We utilise Cramer-Rao Lower Bound (CRLB) analysis to assess the minimum estimation error for different Diffusion Weighted Magnetic Resonance (DWMR) signal decay models. This includes the study of effect of optimisation of acquisition parameters on the performance of the models and identification of limiting factors to achieve higher performances. We demonstrate that, under typical measurement conditions, none of the models can provide parameter estimations with a reasonable uncertainty level. This will show up in estimation algorithms that are extremely sensitive to parameter initial conditions, for example. At the end, we present a simulation framework to model the packing of the axons in the nerve bundles and to model the diffusion of water molecules in the tissue to construct simulation datasets and utilise them to investigate the validity of the models. We demonstrate that the sensitivity of models to initial conditions of the fitting algorithm is high and none of the models are capable of providing realistic estimations in general. The results also show that the more complex models can provide better estimations of the tissue-related parameters in noise-free (simulated) conditions but have higher noise-related minimum uncertainty levels and perform poorly in practical conditions, while simpler models perform better in terms of noise-related uncertainty levels but fail to present the tissue related parameters accurately even in noise-free conditions. Our analysis sheds a new light on methods that aim to infer microstructure from DWI data, and provides a method by which to assess the practicality of proposed techniques in the future.
Notes
Terms of Use
Keywords
Diffusion Weighted Magnetic Resonance Imaging (DWI), brain microstructure, Magnetic Resonance Imaging (MRI), Brownian motion, Axon packing, Axon diameter measurement, Cramer-Rao Lower Bound (CRLB)