I’m having a little bit of a hard time understanding the answers to some similar questions, so please bear with me.
I’m looking at doing a fixel-analysis on tracts of interest (e.g., TractSeg), comparing the fixel-metrics (FD, FC, and FDC) between two groups in pre-selected tracts. I know that its best to keep everything in subject-space as long as possible, but all the papers I’ve read have followed the standard FBA pipeline, ran TractSeg on the population template, computed the mean fixel metrics along each tract, and then ran stats.
My question is whether its better to keep everything in subject-space in this case - so run fod2fixel in subject-space, run TractSeg in subject-space, and extract the mean metrics along each subject-specific tract.
I’m guessing this hasn’t been done because TractSeg produces results needing less manual intervention in template space compared to subject space, but I want to make sure I’m not missing something…
As a side question, I know the Genc et. al, 2020 paper recommends using a single high-b-value shell for fixel analyses (the highest I have is b=2600). Also wondering whether it would be valid to run TractSeg on my full, multi-shell data, and then generate the fixel metrics using FODs generated just from my high b-value shell. That way, I can leverage all my data.