Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Nonparametric D-R1-R2 distribution MRI of the living human brain

Martin, Jan LU ; Reymbaut, Alexis LU ; Schmidt, Manuel ; Doerfler, Arnd ; Uder, Michael ; Laun, Frederik Bernd and Topgaard, Daniel LU (2021) In NeuroImage 245.
Abstract

Diffusion-relaxation correlation NMR can simultaneously characterize both the microstructure and the local chemical composition of complex samples that contain multiple populations of water. Recent developments on tensor-valued diffusion encoding and Monte Carlo inversion algorithms have made it possible to transfer diffusion-relaxation correlation NMR from small-bore scanners to clinical MRI systems. Initial studies on clinical MRI systems employed 5D D-R1 and D-R2 correlation to characterize healthy brain in vivo. However, these methods are subject to an inherent bias that originates from not including R2 or R1 in the analysis, respectively. This drawback can be remedied by extending the... (More)

Diffusion-relaxation correlation NMR can simultaneously characterize both the microstructure and the local chemical composition of complex samples that contain multiple populations of water. Recent developments on tensor-valued diffusion encoding and Monte Carlo inversion algorithms have made it possible to transfer diffusion-relaxation correlation NMR from small-bore scanners to clinical MRI systems. Initial studies on clinical MRI systems employed 5D D-R1 and D-R2 correlation to characterize healthy brain in vivo. However, these methods are subject to an inherent bias that originates from not including R2 or R1 in the analysis, respectively. This drawback can be remedied by extending the concept to 6D D-R1-R2 correlation. In this work, we present a sparse acquisition protocol that records all data necessary for in vivo 6D D-R1-R2 correlation MRI across 633 individual measurements within 25 min—a time frame comparable to previous lower-dimensional acquisition protocols. The data were processed with a Monte Carlo inversion algorithm to obtain nonparametric 6D D-R1-R2 distributions. We validated the reproducibility of the method in repeated measurements of healthy volunteers. For a post-therapy glioblastoma case featuring cysts, edema, and partially necrotic remains of tumor, we present representative single-voxel 6D distributions, parameter maps, and artificial contrasts over a wide range of diffusion-, R1-, and R2-weightings based on the rich information contained in the D-R1-R2 distributions.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; and
organization
publishing date
type
Contribution to journal
publication status
published
subject
keywords
Correlation, Diffusion, Microstructure, Relaxation, Signal inversion
in
NeuroImage
volume
245
article number
118753
publisher
Elsevier
external identifiers
  • scopus:85120305789
  • pmid:34852278
ISSN
1053-8119
DOI
10.1016/j.neuroimage.2021.118753
language
English
LU publication?
yes
id
93f6ed40-fd7c-443d-acd9-c6b368182b03
date added to LUP
2021-12-15 11:00:29
date last changed
2024-04-20 17:43:06
@article{93f6ed40-fd7c-443d-acd9-c6b368182b03,
  abstract     = {{<p>Diffusion-relaxation correlation NMR can simultaneously characterize both the microstructure and the local chemical composition of complex samples that contain multiple populations of water. Recent developments on tensor-valued diffusion encoding and Monte Carlo inversion algorithms have made it possible to transfer diffusion-relaxation correlation NMR from small-bore scanners to clinical MRI systems. Initial studies on clinical MRI systems employed 5D D-R<sub>1</sub> and D-R<sub>2</sub> correlation to characterize healthy brain in vivo. However, these methods are subject to an inherent bias that originates from not including R<sub>2</sub> or R<sub>1</sub> in the analysis, respectively. This drawback can be remedied by extending the concept to 6D D-R<sub>1</sub>-R<sub>2</sub> correlation. In this work, we present a sparse acquisition protocol that records all data necessary for in vivo 6D D-R<sub>1</sub>-R<sub>2</sub> correlation MRI across 633 individual measurements within 25 min—a time frame comparable to previous lower-dimensional acquisition protocols. The data were processed with a Monte Carlo inversion algorithm to obtain nonparametric 6D D-R<sub>1</sub>-R<sub>2</sub> distributions. We validated the reproducibility of the method in repeated measurements of healthy volunteers. For a post-therapy glioblastoma case featuring cysts, edema, and partially necrotic remains of tumor, we present representative single-voxel 6D distributions, parameter maps, and artificial contrasts over a wide range of diffusion-, R<sub>1</sub>-, and R<sub>2</sub>-weightings based on the rich information contained in the D-R<sub>1</sub>-R<sub>2</sub> distributions.</p>}},
  author       = {{Martin, Jan and Reymbaut, Alexis and Schmidt, Manuel and Doerfler, Arnd and Uder, Michael and Laun, Frederik Bernd and Topgaard, Daniel}},
  issn         = {{1053-8119}},
  keywords     = {{Correlation; Diffusion; Microstructure; Relaxation; Signal inversion}},
  language     = {{eng}},
  publisher    = {{Elsevier}},
  series       = {{NeuroImage}},
  title        = {{Nonparametric D-R<sub>1</sub>-R<sub>2</sub> distribution MRI of the living human brain}},
  url          = {{http://dx.doi.org/10.1016/j.neuroimage.2021.118753}},
  doi          = {{10.1016/j.neuroimage.2021.118753}},
  volume       = {{245}},
  year         = {{2021}},
}