MSMT-CSD on two-shell data - WM and CSF or WM and GM

1

Dear experts,

I am running FOD estimation on my 2-shell data (b=0 and b=1000). To my knowledge, the MSMT-CSD algorithm can be used for this purpose, but FOD estimation would be limited to 2 types of tissue with 2 b-values.

My question is, apart from WM, should I include the response function of GM or CSF for dwi2fod? Thanks a lot!!

It has been suggested that, in order to fit all signal, the inputs typically have to be WM and CSF as these two have the lowest and highest diffusivities. However, I have tried both WM+GM and WM+CSF, and the WM+GM result seems to look more reasonable - it appears to be fitted on the WM region more accurately (i.e. with less false positive signals on non-WM regions):

WM+GM:

image
image558×512 179 KB
image
image616×536 131 KB
image
image602×640 258 KB

WM+CSF:

image
image696×616 269 KB
image
image738×636 287 KB
image
image694×726 116 KB

Please also let me know if it is suggestive of potential issues in the upstream processes. Many thanks!!

2

Please have a look at Single shell vs. "Single tissue"
Cheers
William

3

Dear William,

Thank you so much. It seems that the pipeline that I used (dwi2response dhollander + dwi2fod msmt_csd) for my b=0 + b=1000 data (so called single-shell) was used long ago before ss3t_csd_beta1 was developed. I will replace my dwi2fod algorithm with it then!

Thanks,
Idy