Journal
NEUROIMAGE
Volume 231, Issue -, Pages -Publisher
ACADEMIC PRESS INC ELSEVIER SCIENCE
DOI: 10.1016/j.neuroimage.2021.117849
Keywords
Diffusion MRI; Spinal Cord; White matter; Microstructure; Multi-fascicle models; Model selection
Funding
- Danish National Research Foundation (CFIN)
- Danish Ministry of Science, Innovation, and Education (MINDLab)
- VELUX Foundation (ARCADIA) [00015963]
- European Research Council (ERC) [679058]
- Lisboa Regional Operational Programme (Lisboa 2020), under the PORTUGAL 2020 Partnership Agreement through the European Regional Development Fund (ERDF)
- Fundacao para a Ciencia e Tecnologia (Portugal) [LISBOA010145FEDER022170]
- European Research Council (ERC) [679058] Funding Source: European Research Council (ERC)
Ask authors/readers for more resources
This study reevaluates the assumptions of the standard diffusion model by collecting comprehensive MRI data and proposes three model extensions to address the issues identified. The analysis identifies a three-compartment description as the optimal model, with the third compartment exhibiting slow diffusion and a minor signal fraction. Failure to consider the presence of this compartment can severely misguide inferences about white matter microstructure.
Information about tissue on the microscopic and mesoscopic scales can be accessed by modelling diffusion MRI signals, with the aim of extracting microstructure-specific biomarkers. The standard model (SM) of diffusion, currently the most broadly adopted microstructural model, describes diffusion in white matter (WM) tissues by two Gaussian components, one of which has zero radial diffusivity, to represent diffusion in intra-and extra-axonal water, respectively. Here, we reappraise these SM assumptions by collecting comprehensive double diffusion encoded (DDE) MRI data with both linear and planar encodings, which was recently shown to substantially enhance the ability to estimate SM parameters. We find however, that the SM is unable to account for data recorded in fixed rat spinal cord at an ultrahigh field of 16.4 T, suggesting that its underlying assumptions are violated in our experimental data. We offer three model extensions to mitigate this problem: first, we generalize the SM to accommodate finite radii (axons) by releasing the constraint of zero radial diffusivity in the intra-axonal compartment. Second, we include intracompartmental kurtosis to account for non-Gaussian behaviour. Third, we introduce an additional (third) compartment. The ability of these models to account for our experimental data are compared based on parameter feasibility and Bayesian information criterion. Our analysis identifies the three-compartment description as the optimal model. The third compartment exhibits slow diffusion with a minor but non-negligible signal fraction (similar to 12%). We demonstrate how failure to take the presence of such a compartment into account severely misguides inferences about WM microstructure. Our findings bear significance for microstructural modelling at large and can impact the interpretation of biomarkers extracted from the standard model of diffusion.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available