MSMT-CSD for single shell data

Dear Mrtrix experts,

I have a dataset of some subjects having a multi-shell acquisition (b=0, 1000, 2000; 32 diffusion directions each b).
Im trying to compare the connectome of global tractography from multi-shell data with those from single shell data using multi-shell multi-tissue CSD algorithm. Therefore, Id like to set the number of shells and tissue types to one in the single-shell case for FOD estimation.

But, I don’t know how to write the script. Could you teach the script for setting the number of shells and tissue types to one in the single-shell case for FOD estimation?

For multi-shell data, we used a script like below.

dwi2response msmt_5tt DWI.mif 5TT.mif RF_GM.txt RF_WM.txt RF_CSF.txt -voxels RF_voxels.mif
dwi2fod msmt_csd dwi.mif RF_GM.txt GM.mif RF_WM.txt WM_FODs.mif RF_CSF.txt CSF.mif
mrconvert WM_FODs.mif - -coord 3 0 | mrcat CSF.mif GM.mif - tissueRGB.mif -axis 3

I have one more question.
Should we set a lmax for MSMT CSD algorithm? And how to set a lmax for MSMT-CSD?

Extract: the b=2000 shell from dwi.mif:

dwiextract -shell 2000 dwi.mif dwi_ss.mif

Extract the b=2000 shell from RF_WM.txt:

tail -1 RF_WM.txt > RF_WM_ss.txt

Run dwi2fod as before, but with the reduced DWI and RF and only for WM:

dwi2fod msmt_csd dwi_ss.mif RF_WM_ss.txt WM_FODs_ss.mif

About lmax: you typically don’t have to set it manually, but you can obtain the used value afterwards by looking at the size of the fourth dimension of the output files using mrinfo. See the table at http://mrtrix.readthedocs.io/en/latest/constrained_spherical_deconvolution/lmax.html to work out the actual lmax.

If you want to force a specific lmax, you can use the -lmax option of dwi2fod and provide one lmax per tissue type (comma separated)

Dear Dr Jeurissen,

Thank you for your kind reply!:blush:
I could find the size of the fourth dimensions of the each file, and I could find the actual lmax using the table.

Thank you.

Koji

Thank you for your reply!

Is this script the same method you used for response function estimation for SSST-CSD in your previous study (NeuroImage 103 (2014) 411–426)?

dwiextract -shell 2000 dwi.mif dwi_ss.mif
tail -1 RF_WM.txt > RF_WM_ss.txt
dwi2fod msmt_csd dwi_ss.mif RF_WM_ss.txt WM_FODs_ss.mif

”In the SSST-CSD framework, the anisotropic single fibre response function is…, the forward spherical convolution matrix C can be calculated analytically (Tournier et al., 2004). from NeuroImage 103 (2014) 414 page”

Or the methods in your previous study mean ”dwi2response fa”?

Regards,
Koji

No, this script will extract the single shell, single tissue response from the multi shell, multi tissue response, which is what I recommend if you want to perform a head to head comparison between the ‘single-tissue’ and ‘multi-tissue’ approach.