Mrtrix3tissue response estimation


I recently used the mrtrix single shell 3 tissue response function calculation using

dwi2response dhollander

After trying to run ss3t_csd_beta1 and getting a command does not exist error I looked into the issue and realized that mrtrix3tisse needs to be downloaded separately (I didn’t do a good enough job of reading the documentation).

I am going to download mrtrix3tissue and then run ss3t_csd_beta1, however, I read under some documentation that the mrtrix dwi2response dhollander function might be different than the current one used in mrtrix3tissue? Should I rerun the dwi2response call specifying to use the command in mrtrix3tissue instead or mrtrix or can I trust what I have already calculated and simply move forward with ss3t?

Thanks for any help you can provide!


1 Like

Hi Daniel,

Thanks for the interest in SS3T-CSD and 3-tissue response function estimation via dwi2response dhollander.

That’s correct. This is a fork, so as to ensure the SS3T-CSD code works reliably and SS3T-CSD behaviour is as easily reproducible as possible (because it depends on a lot of other functionality that might change behaviour at times). It looks like you’ve figured this out though; good to hear. :+1:

Yes, this is currently the case. This probably also changes over time. In any case, currently the latter version is the new and improved 2019 version of the dwi2response dhollander algorithm. Details and comparison to the previous 2016 version available in this abstract and explained in this talk.

It would technically work and run with the response functions you had already calculated, however, I strongly advise to rerun with the 2019 algorithm (which also takes far less time to run/compute by the way). The abstract and talk illustrate how this leads to a better 3-tissue CSD fit, with sometimes quite a marked impact on the result. So I highly recommend to rerun.

No worries at all. Hope it’s all clear now. :slightly_smiling_face: