Hi MRtrix community,
I am jumping in this thread as it is relevant to my work as well.
I am starting to develop a pipeline for a DWI dataset acquired at 7T. This is a multi-shell protocol, however collected as separate single-shells acquisitions. Specifically, the sequence includes: one b=1000 shell (12 encoding directions + 1 b=0 volume), one b=2000 shell (27 directions, of which 3 b=0 volumes) and one b=3000 shell (53 directions, of which 5 b=0 volumes). Additionally, 6 b=0 volumes were acquired with opposite phase encoding direction.
If I understand correctly, multiple shells should be combined together and fed into relevant commands as one single dwi.mif file (correct?). Given the considerations of @rsmith, I am wondering what is the best approach to follow? I can think of three alternatives:
-
Concatenate the DICOMs as @jdtournier suggested
-
Concatenate the already converted .mif files
-
Do some pre-processing (e.g. dwidenoise) on the separate shells and then combine (?)
Regardless of the specific method, is there a specific order how they should be concatenated? (e.g. all b0s stacked at the beginning of each shell?)
Based also on what I read in another post (Troubles with tractography on 7T diffusion data), I checked whether the intensity fluctuates between the different shells. Turns out it does a bit (mean b=0 intensity within same brain mask = 624.135, 619.602, 617.853 for b=1000,2000 and 3000 respectively).
Is this considered problematic? I tried to scale it by fixing the mean b=0 intensity to 1000 as suggested in the mentioned post, but this did not solve the problem…
Sorry if the answers to these questions may seem very obvious to you, but I never handled neither multi-shell nor 7T data before, so I am trying to define the most appropriate way of proceeding.
Thanks in advance!