Hey Benny,
That’s the consequence of using single-tissue CSD (following dwi2response tournier
) or 2-tissue CSD (with the WM and CSF responses from dwi2response dhollander
) on low b-value data. What you’re seeing is GM-like tissue (well, actual cortical GM) being fitted mostly by the WM compartment (and for a very small fraction also by the CSF compartment). At low b-values, this produces relatively more signal than WM-like tissue, which is the bright cortical ribbon you’re seeing.
Separately, the super-bright region at the front of the brain in your screenshot in particular, that’s further emphasised by EPI distortions.
That can happen additionally when using 2-tissue CSD + mtnormalise
at low b-values; I’ve observed this myself in some experiments, so it’s expected, yes. This in addition to other challenges with interpretation, depending on the scenario, that I hinted at in this post.
Modelling GM-like tissue in its own separate GM-like compartment, which will deal with the problem you’re seeing here, is for single-shell (and low b-value) data only possible using single-shell 3-tissue CSD (SS3T-CSD).
Cheers,
Thijs