Impact of within-voxel heterogeneity in fibre geometry on spherical deconvolution

TL;DR

This study investigates how within-voxel heterogeneity in fibre geometry affects spherical deconvolution in diffusion MRI, revealing that a single fibre response function may be insufficient for accurate fibre orientation estimation.

Contribution

It demonstrates through simulations that variable fibre geometry causes a non-constant fibre response function, challenging the assumption of a single FRF in spherical deconvolution.

Findings

Variable fibre geometry leads to a variable FRF within a voxel.

A single FRF is generally ineffective for accurate fODF recovery.

Assuming a single FRF can cause errors in tractography.

Abstract

Axons in white matter have been shown to have varying geometries within a bundle using ex vivo imaging techniques, but what does this mean for diffusion MRI (dMRI) based spherical deconvolution (SD)? SD attempts to estimate the fibre orientation distribution function (fODF) by assuming a single dMRI fibre response function (FRF) for all white matter populations and deconvolving this FRF from the dMRI signal at each voxel to estimate the fODF. Variable fibre geometry within a bundle however suggests the FRF might not be constant even within a single voxel. We test what impact realistic fibre geometry has on SD by simulating the dMRI signal in a range of realistic white matter numerical phantoms, including synthetic phantoms and real axons segmented from electron microscopy. We demonstrate that variable fibre geometry leads to a variable FRF across axons and that in general no single FRF is effective to recover the underlying fibre orientation distribution function (fODF). This finding suggests that assuming a single FRF can lead to misestimation of the fODF, causing further downstream errors in techniques such as tractography.