Hi Mrtrix3 Experts,
I am at the end of the analysis pipeline and had a few questions about how to go about statistical analysis of my data.
My project is around looking at recognition memory pathways and how they differ for different BDNF genotypes.
I have a sample of 61 participants and for the behavioural analysis of our data, we have used Bayesian analysis to see how recollection and familiarity scores differ in the different BDNF genotypes.
I was wondering if there is any way to pull out the FD, FDC and FC measures from Mrtrix3, so I can do a Bayesian analysis on how pathways (that are significant) differ in different BDNF carriers and if there is a difference in pathways for recollection and familiarity.
I recently bumped across this paper which is “Development of white matter fibre density and morphology over childhood: a longitudinal fixel-based analysis.”, specifically Table 3. Although it is a longitudinal study I think this way of statistical analysis is very interesting and something I would like to do in my study.
As one of the authors in this paper is Robert Smith, I am sure I would get a clear idea about it here on the MRtrix3 community page.
I was wondering if there is any way to pull out the FD, FDC and FC measures from Mrtrix3,
I suspect that what you’re actually looking for is not just “pulling out” values, which you can do in a number of ways (e.g. mrdump command, using Matlab read function, converting to some other image format that you can read), since you would then have one value per fixel per participant, which is likely an overwhelming volume for such an analysis.
More likely what you’re looking for is calculating the e.g. mean values of those metrics within specific bundles of interest. This involves deriving a fixel mask for each such bundle, after which you can use mrstats with the -mask option to compute statistics across just the fixels within the mask. The derivation of such masks would benefit from both some technical develepment and more centralised documentation, there’s likely quite a few threads on this forum describing such, but it’s typically some combination of tckgen / tckedit, tck2fixel to get a streamline count per fixel, and mrthreshold to produce a binary fixel mask.
“Development of white matter fibre density and morphology over childhood: a longitudinal fixel-based analysis.”, specifically Table 3.
You would need to harass @sgenc regarding the Bayesian repeated measures GLM, as I did not myself work directly with those data.