The R1-weighted connectome: complementing brain networks with a myelin-sensitive measure

Abstract

Myelin plays a crucial role in how well information travels between brain regions. Many neurological diseases affect the myelin in the white matter, making myelin-sensitive metrics derived from quantitative MRI of potential interest for early detection and prognosis of those conditions. Complementing the structural connectome, obtained with diffusion MRI tractography, with a myelin sensitive measure could result in a more complete model of structural brain connectivity and give better insight into how the myeloarchitecture relates to brain function. In this work we weight the connectome by the longitudinal relaxation rate (R1) as a measure sensitive to myelin, and then we assess its added value by comparing it with connectomes weighted by the number of streamlines (NOS). Our analysis reveals differences between the two connectomes both in the distribution of their weights and the modular organization. Additionally, the rank-based analysis shows that R1 is able to separate different classes (unimodal and transmodal), following a functional gradient. Overall, the R1-weighted connectome provides a different perspective on structural connectivity taking into account white matter myeloarchitecture.Author summaryIn the present work, we integrate a myelin sensitive MRI metric into the connectome and compare it with a connectome weighted with a standard diffusion-derived metric, number of streamlines (NOS). Our analysis shows that the R1-weighted connectome complements the NOS-weighted connectome. We show that the R1-weighted average distribution does not follow the same trend as the NOS strength distribution, and the two connectomes exhibit different modular organization. We also show that unimodal cortical regions tend to be connected by more streamlines, but the connections exhibit a lower R1-weighted average, while the transmodal regions tend to have a higher R1-weighted average but fewer streamlines. In terms of network communication, this could imply that the unimodal regions require more connections with lower myelination, whereas the transmodal regions take more myelinated, but fewer, connections for a reliable transfer of information.

Vendor‐neutral sequences and fully transparent workflows improve inter‐vendor reproducibility of quantitative MRI

“…The most important contribution of this article is the vendor‐neutral solution it provides for multi‐center qMRI by significantly reducing inter‐vendor variability. This issue has been hampering the standardization of qMRI methods for multi‐center clinical trials, 59 validation, 60,61 establishing protocols, 17 applied neuroimaging studies, 62 determining the range of parameters in pathology 63,64 and in health, 16,38 scanner upgrades, 48 and even for phantom studies 12,13 . By reducing such variabilities, the VENUS approach can bring qMRI closer to teasing out the true biological variability in quantifying in vivo tissue microstructure 65 …”

The Myelin‐Weighted Connectome in Parkinson's Disease

“…The diffusion data were preprocessed using MRTrix 3.0 49 . The pipeline for the diffusion preprocessing is described in Boshkovski et al 29 In brief, the diffusion images were first denoised 50,51 and then corrected for Gibbs ringing artifacts 52 and B1 field inhomogeneity. Then the images were also corrected for motion 53 and inhomogeneity distortions 54 using the FSL's eddy and topup tools, relying on b0 images acquired with reverse‐phase encoding.…”

“…There is a growing interest in evaluating pathology for which myelin‐specific changes in brain connectivity are suspected 26‐28 . Several myelin‐sensitive metrics have been proposed, including R1 29,30 . For this study, we chose R1 because it has been shown to be highly correlated with myelin 31,32 in a broad range of pathologies.…”

Quantitative mapping of the brain's structural connectivity using diffusion MRI tractography: a review

“…For microstructural measures, the mean measure within the fiber pathway is the most widely used. Other summary statistics such as the median, maximum and minimum have also been employed [50,538,537]. A summary statistic can be obtained in several ways according to how microstructural measures are computed along fiber pathways (as described in Section 5.1).…”

“…The choice of the best summary statistic for data measured within a fiber pathway is still open. For example, compared to the mean statistic, studies have shown that the median can be more robust against outliers and does not rely on the normality assumption for the distribution of the microstructure parameter along a streamline [50,538]. One study has suggested that the maxima and minima are more discriminative than the mean value in machine-learning-based disease classification [537].…”