Single shell DTI-data

Hello
Any good advice for analysis of single shell DTI-data (I’ve got b0 and b1000). I’ve done the preprocessing (dwifslpreproc and masking). Should I go back to FSL, or is there a pipeline through MRTrix? I’m interested in both WM and GM, and ROI- analysis is of interest. I’ve enjoyed the BATMAN-tutorial, but since I don’t have multishell data, I have to find a new way as I’ve understood it.
New to the game, so any clues or good advice are welcome and appreciated!
Cheers!

I guess, that depends kinda on what you want to achieve. If you want to derive DTI parameters, like FA, MD, RD, AD, you can chose to do that with FSL or MRtrix. The main advantage of using MRtrix, is in my opinion, the ease of several tools integrated in the pipeline (including FSL).

Basically, I would follow this tutorial till step 4 (thus stop after step 3, the bias field correction), and then perform dwi2tensor.

Thus in order:

  1. dwidenoise
  2. mrdegibbs
  3. dwifslpreproc
  4. dwibiascorrect ants
  5. dwi2tensor (or you can switch to FSL and run dtifit)
1 Like

There is actually a page on the documentation about using single-shell data. However, you can do better than this these days, since even single-shell data effectively contains 2 shells: the b=0 volumes qualify as a ‘shell’ in this context (we need to update the docs to reflect that… :flushed:). So you should be able to use your data in more or less exactly the same way as outlined in the BATMAN tutorial, with the only change being the dwi2fod call, as outlined in this post and referred to in this post.

Also: bear in mind that the original CSD method was developed for single-shell data, and is still usable here. It’s not as clean as what you can get with the recommendations above though, but it is nonetheless an option (invoke with dwi2fod csd).

2 Likes

Thank you :slight_smile: