Just out of curiosity: how many gradient directions (at b=1200) and how many b=0 images does your acquisition consist of?
dhollander algorithm works on single-shell as well as multi-shell data, for a wide range of b-values and combinations. It always delivers the corresponding single-fibre WM, GM and CSF response functions. If the dataset has at least 3 unique b-values, you can perform MSMT-CSD with the 3 response functions to result in estimated WM FODs, and GM and CSF compartments. If the dataset is single-shell, and hence only has 2 b-values (b=0 and another b-value, 1200 in your case), the MSMT-CSD algorithm is limited to providing outputs for 2 tissue types only.
…so that idea would work, but there’s a few errors in your command there. It should be something like:
dwi2fod msmt_csd wm_response.txt wm_fod.mif csf_response.txt csf.mif -mask mask.mif
…which would indeed apply MSMT-CSD, but only with 2 tissue types. Naturally, you’d indeed always choose WM and CSF in such a scenario, and dismiss the GM response function. I’d recommend to explicitly refer to your results as “2-tissue CSD results” when you write about them, e.g. in a paper or other publication. That clears up any possible confusion for your readers.
So that’s indeed an entirely different algorithm, that does allow to get 3-tissue results from only single-shell data… and yes, it’s not released yet. If I don’t get swamped with too much other stuff, I’m aiming to release a version of it by the end of the year (I’ve been saying this more than a year ago as well, but let’s hope I can make it happen this time ).
On a vaguely related note: before that happens, I’ll also be releasing (or at least publicly trialling) an improved version of the
dwi2response dhollander algorithm by the way. I had some initial worries about not being able to make it robust enough (and I’m not the kind of person to easily sacrifice robustness; I’m quite happy with the existing algorithm’s current robustness )… but it looks like I’ve got that covered now.