Pushing diffusion MRI towards new dimensions
(2020)- Abstract
- Diffusion-MRI techniques allow the non-invasive investigation of microstructural changes in living tissues. However, a detailed interpretation of the data is complicated by the fact that multiple microscopic environments with varying diffusion properties all contribute to the measured signal. To address this problem, we adapt strategies from solid-state NMR spectroscopy and magnetic resonance of porous media to design multidimensional diffusion MRI protocols that can establish correlations between distinct features of the underlying diffusion process. Inversion of the acquired data enables the quantification of tissue heterogeneity with non-parametric distributions of diffusion tensors. The size, shape, and orientation of the estimated... (More)
- Diffusion-MRI techniques allow the non-invasive investigation of microstructural changes in living tissues. However, a detailed interpretation of the data is complicated by the fact that multiple microscopic environments with varying diffusion properties all contribute to the measured signal. To address this problem, we adapt strategies from solid-state NMR spectroscopy and magnetic resonance of porous media to design multidimensional diffusion MRI protocols that can establish correlations between distinct features of the underlying diffusion process. Inversion of the acquired data enables the quantification of tissue heterogeneity with non-parametric distributions of diffusion tensors. The size, shape, and orientation of the estimated diffusion tensors have direct relations to corresponding microstructural properties of biological tissues.
In Paper I, we introduced an experimental protocol to establish correlations between diffusion tensor eigenvalues. The proposed approach was extended in Paper II, where we correlated individual diffusion tensor parameters with both longitudinal and transverse relaxation rates. In both papers, experimental validation was conducted using spectroscopic experiments on a set of specially tailored synthetic samples. Multidimensional distributions retrieved from the correlated datasets were found to provide excellent resolution between microscopic environments with distinct diffusion properties.
In Paper III, we assessed the performance of our data inversion strategies within a clinical context using in silico data. We found that the proposed model-free algorithm preserves good contrast between systems with different microscopic structures, even though its accuracy is significantly affected by high-levels of experimental noise. The algorithm was also observed to exhibit no biases at infinite signal-to-noise ratios.
In Paper IV we combined our diffusion tensor correlation protocols with MRI sequences allowing for sub-millimetre imaging of living tissues. The method was demonstrated with measurements on in vivo mouse brain and was validated using a set of phantoms emulating the diffusion properties of brain tissues.
In Papers V and VI we investigated the microscopic heterogeneity of the living human brain with spatially resolved relaxation-diffusion distributions. The retrieved distributions allowed the resolution, characterisation, and mapping of distinct microscopic tissue environments. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/1dd25d68-7047-4919-b14f-e2ab34ce3060
- author
- Martins, Joao LU
- supervisor
- opponent
-
- Professor MacKay, Alex, University of British Columbia, Vancouver, Canada
- organization
- publishing date
- 2020-01
- type
- Thesis
- publication status
- published
- subject
- keywords
- NMR, MRI, diffusion, relaxation, multidimensional diffusion MRI, relaxation-diffusion correlations, Laplace NMR, porous media, tissue microstructure, heterogeneity, anisotropy
- pages
- 208 pages
- publisher
- Lund University, Faculty of Science, Department of Chemistry, Division of Physical Chemistry
- defense location
- Lecture Hall B, Kemicentrum, Naturvetarvägen 14, Lund
- defense date
- 2020-01-31 13:00:00
- ISBN
- 978-91-7422-719-2
- 978-91-7422-718-5
- language
- English
- LU publication?
- yes
- id
- 1dd25d68-7047-4919-b14f-e2ab34ce3060
- date added to LUP
- 2020-01-05 18:37:11
- date last changed
- 2020-01-10 16:43:12
@phdthesis{1dd25d68-7047-4919-b14f-e2ab34ce3060,
abstract = {{Diffusion-MRI techniques allow the non-invasive investigation of microstructural changes in living tissues. However, a detailed interpretation of the data is complicated by the fact that multiple microscopic environments with varying diffusion properties all contribute to the measured signal. To address this problem, we adapt strategies from solid-state NMR spectroscopy and magnetic resonance of porous media to design multidimensional diffusion MRI protocols that can establish correlations between distinct features of the underlying diffusion process. Inversion of the acquired data enables the quantification of tissue heterogeneity with non-parametric distributions of diffusion tensors. The size, shape, and orientation of the estimated diffusion tensors have direct relations to corresponding microstructural properties of biological tissues.<br/>In Paper I, we introduced an experimental protocol to establish correlations between diffusion tensor eigenvalues. The proposed approach was extended in Paper II, where we correlated individual diffusion tensor parameters with both longitudinal and transverse relaxation rates. In both papers, experimental validation was conducted using spectroscopic experiments on a set of specially tailored synthetic samples. Multidimensional distributions retrieved from the correlated datasets were found to provide excellent resolution between microscopic environments with distinct diffusion properties.<br/>In Paper III, we assessed the performance of our data inversion strategies within a clinical context using in silico data. We found that the proposed model-free algorithm preserves good contrast between systems with different microscopic structures, even though its accuracy is significantly affected by high-levels of experimental noise. The algorithm was also observed to exhibit no biases at infinite signal-to-noise ratios.<br/>In Paper IV we combined our diffusion tensor correlation protocols with MRI sequences allowing for sub-millimetre imaging of living tissues. The method was demonstrated with measurements on in vivo mouse brain and was validated using a set of phantoms emulating the diffusion properties of brain tissues.<br/>In Papers V and VI we investigated the microscopic heterogeneity of the living human brain with spatially resolved relaxation-diffusion distributions. The retrieved distributions allowed the resolution, characterisation, and mapping of distinct microscopic tissue environments.}},
author = {{Martins, Joao}},
isbn = {{978-91-7422-719-2}},
keywords = {{NMR; MRI; diffusion; relaxation; multidimensional diffusion MRI; relaxation-diffusion correlations; Laplace NMR; porous media; tissue microstructure; heterogeneity; anisotropy}},
language = {{eng}},
publisher = {{Lund University, Faculty of Science, Department of Chemistry, Division of Physical Chemistry}},
school = {{Lund University}},
title = {{Pushing diffusion MRI towards new dimensions}},
url = {{https://lup.lub.lu.se/search/files/74289479/Joao_Martins_No_Manuscript.pdf}},
year = {{2020}},
}