What the result of two shells (one 0, one 1500) using msmt_csd should look like?

[Yufeng_He]

Dear Community,

I’m generating FOD for my two shells(0,1500), 60 directions data, using the MSMT-CSD method.I want to check my generated fod map to make sure everything is ok. But I suddenly realised that there are very few tutorials on what the results of two shell analysis using MSMT-CSD should look like.The tutorial on Andy’s book and BATMAN, which used data from larger than two shells, obtained results similar to Figure 1.But I find this very different from the results I generated. Anyway I want to make sure that my results are normal. If it is wrong, is there something wrong there? I hope to get replies from all of you. Thanks.
This is the code I used to generate the WM,GM, CSF response function, and FOD.

# use dhollander
dwi2response dhollander preproc_unbiased.mif \
wm_response.txt gm_response.txt csf_response.txt -voxels voxels.mif 
# Upsample
mrgrid  preproc_unbiased.mif regrid -vox 1.25 preproc_unbiased_upsample.mif
# FOD MSMT,only WM and CSF
dwi2fod -force msmt_csd preproc_unbiased_upsample.mif \
-mask mask_upsample.mif \
wm_response.txt wmfod.mif \
csf_response.txt csffod.mif
# check result
mrconvert -coord 3 0 wmfod.mif - | mrcat csffod.mif - 2_tiusses_combin.mif
mrview 2_tiusses_combin.mif -odf.load_sh wmfod.mif

This is the result picture of BATMAN.

And,this is the picture of my result. I would like to know if this result is normal?

On top of that, I’m confused about the resulting wm_fod.mif and the cdf_fod.mif produced after dwi2fod is about? I checked both results using mrinfo and found that they have different volume counts. （The volume number of wm_fod is less than the volume number of my raw data，10b0 + 60 1500).

Also I noticed that ss3t_csd_beta1 is available on MRtrix3Tissue for single shell data. I wonder which method is right for my data?

Anyway hope to get everyone’s input and if anyone has any insight please let me know. Thanks.
Yufeng

[bjeurissen]

I don’t see anything wrong.

@Yufeng_He

mrconvert -coord 3 0 wmfod.mif - | mrcat csffod.mif - 2_tiusses_combin.mif

Maybe better to convert them as:

mrconvert -coord 3 0 wmfod.mif - | mrcat csffod.mif $(mrcalc csffod.mif 0 -multiply -) - vf.mif

This adds an extra color channel for the missing GM (and sets it to zero), such that you get the same RGB style image (albeit without GM compartment (green)).

@Yufeng_He

On top of that, I’m confused about the resulting wm_fod.mif and the cdf_fod.mif produced after dwi2fod is about? I checked both results using mrinfo and found that they have different volume counts. （The volume number of wm_fod is less than the volume number of my raw data，10b0 + 60 1500).

Functions of the sphere (such as the FOD) are represented using spherical harmonics coefficients (see Spherical Harmonics — MRtrix3 3.0 documentation for details).

Isotropic compartments (such as CSF and GM) only require 1 coefficient, as they don’t have to model any angular dependency (hence the one in the final dimension of csffod)
Anisotropic compartments (such as WM) typically require 45 coefficients (corresponding to SH order 8), hence the 45 in the fina dimension of wmfod.
.

[Yufeng_He]

@bjeurissen

I don’t see anything wrong.

Thanks for your reply, that’s great. I’m happy I’m doing the right thing.

@bjeurissen

Maybe better to convert them as:

mrconvert -coord 3 0 wmfod.mif - | mrcat csffod.mif $(mrcalc csffod.mif 0 -multiply -) - vf.mif

This adds an extra color channel for the missing GM (and sets it to zero), such that you get the same RGB style image (albeit without GM compartment (green)).

The results look better.Like,

@bjeurissen

Functions of the sphere (such as the FOD) are represented using spherical harmonics coefficients (see Spherical Harmonics — MRtrix3 3.0 documentation for details).

Isotropic compartments (such as CSF and GM) only require 1 coefficient, as they don’t have to model any angular dependency (hence the one in the final dimension of csffod)
Anisotropic compartments (such as WM) typically require 45 coefficients (corresponding to SH order 8), hence the 45 in the fina dimension of wmfod.

Thanks for your information, I will carefully check the document about the Spherical Harmonics.