Improving the Realism of White Matter Numerical Phantoms: A Step toward a Better Understanding of the Influence of Structural Disorders in Diffusion MRI

Abstract

White matter is composed of irregularly packed axons leading to a structural disorder in the extra-axonal space. Diffusion MRI experiments using oscillating gradient spin echo sequences have shown that the diffusivity transverse to axons in this extra-axonal space is dependent on the frequency of the employed sequence. In this study, we observe the same frequency-dependence using 3D simulations of the diffusion process in disordered media. We design a novel white matter numerical phantom generation algorithm which constructs biomimicking geometric configurations with few design parameters, and enables to control the level of disorder of the generated phantoms. The influence of various geometrical parameters present in white matter, such as global angular dispersion, tortuosity, presence of Ranvier nodes, beading, on the extra-cellular perpendicular diffusivity frequency dependence was investigated by simulating the diffusion process in numerical phantoms of increasing complexity and fitting the resulting simulated diffusion MR signal attenuation with an adequate analytical model designed for trapezoidal OGSE sequences.This work suggests that angular dispersion and especially beading have non-negligible effects on this extracellular diffusion metrics that may be measured using standard OGSE DW-MRI clinical protocols.

Robust Monte-Carlo Simulations in Diffusion-MRI: Effect of the Substrate Complexity and Parameter Choice on the Reproducibility of Results

“…This decrease of the ICVF is an effect of the presence of dispersion and deformation of the fibre bundles. However, even in the lowest resolution, the intra-axonal volume fraction achieved was over 48%, which is considerably higher than the icvf (of 20%) of a previously presented framework for generating realistic numerical phantoms for crossing fascicles [20]. By optimizing the penalization term of the strands' curvature in our framework, we expect to be able to achieve even higher packing densities-closer to the expected ones from the brain's white matter tissue.…”

“…In addition, the intracellular compartment is usually idealised as a collection of parallel hollow cylinders with constant radii or radii sampled from a distribution estimated from histological data [38,17,2,21,2]. Recent studies have suggested that such simplification cannot capture the complexity of the axonal structures of white matter, and thus its diffusion characteristics [31,20].…”

“…Because of the aforementioned problems, a number of works proposed experiments where non-trivial structures were used as intracellular substrates, e.g. dispersed axons [20], the presence of abutting cylinders [49] and arbitrarily generated meshes [33]. However, to this day, such approaches have not been thoroughly adopted by the DW-MRI community because of the high computational burden they demand and the lack of available tools.…”

Decoding the microstructural properties of white matter using realistic models

“…Dispersion can occur not only in regions of fiber crossing, fiber kissing, but also in regions traditionally expected to be unidirectional such as the corpus callosum [33]. However, 3D models are hard to construct, not only because of the lack of 3D EM data (that could represent a ground truth), but also because of the complexity of 3D axon packing [34]. Also, it could be the case that the 2D axon shapes, used in our realistic WM modeling, are elongated due to being obtained from cutting through axons that were not perpendicular to the surface.…”

Contextual Fibre Growth to Generate Realistic Axonal Packing for Diffusion MRI Simulation

“…The most common approach to this is the packing of fibres into densely packed configurations (Close et al, 2009;Ginsburger et al, 2019Ginsburger et al, , 2018Rafael-Patino et al, 2018).…”