i would like to know how to apply multi shell multi tissue CSD using just response functions from WM and CSF (since I work with neonatal data where WM/GM distinction is not clear).
Can something like this be ok?
dwi2response dhollander dki_bfc.mif wm_response.txt gm_response.txt csf_response.txt -voxels response_voxels.mif
dwi2fod msmt_csd -mask mean_b0_BRAIN_corr.nii.gz dki_bfc.mif wm_response.txt wm.mif csf_response.txt csf.mif -nthreads 30
note that dki_bfc is the processed dwi and mean_b0_BRAIN_corr is the corresponding mask
Yes, that can work, and is indeed the approach taken in a recent paper of ours – though this decomposition is then used primarily to drive the registration. But in general, the neonatal brain is so different from the adult that I’m not sure there’s any consensus on how best to analyse the data. The 2-tissue WM/CSF MSMT-CSD decomposition is probably the simplest approach, but there’s a lot of interest and ideas for figuring out how to extract more information from the data – and we clearly know there is more information in there. Lots of research focusing on this at the moment…