I have four sets acquired sequentially within the same scan session (set 1: b-700, pe dir: AP, set 2: b-700, pe dir: PA, set 3: b-3000, pe dir: AP, set 1: b-3000, pe dir: PA). Thus far, I have used each b val set pair to build tractograms, they both look good on inspection. When building the (whole brain) connectome, should one include the b 3000 sets only (high angular resolution, low SNR), or all 4 sets (low b with high SNR/high b with low SNR).
If the latter, how would one do it (at what step in the pipeline would one concatenate/merge the stepwise files respective to the 2x2 sets, and what pipeline command changes would be required for the inclusion of all 4 sets)?
Thank you,
If you have reversed phase-encode pairs for all DWI volumes, your best bet is to use either the -rpe_header or -rpe_all option in dwipreproc to combine each pair of volumes with the same phase-encoding direction into a single output volume. This is preferable to keeping all individual volumes separately, since it applies appropriate differential weighting between the two volumes in regions of signal expansion/contraction. Unfortunately this is buggy at present, but the fix should go out in the next couple of days.
In terms of combining b-value shells, this should really be thought of as a question of FOD estimation rather than “connectome generation” (the effects on which are much further downstream). If you have three b-value shells (counting b=0 volumes as a “shell”), you can estimate three distinct tissue types as shown in the multi-shell multi-tissue CSD paper; see documentation here. With only one non-zero b-value shell, you can’t currently do this.