Setting correct bval/bvecs

[leotozzi88]

The bvals/bvecs can vary from subject to subject.
The conversion script should take the last 79/81 bvals/bvecs from the files downloaded from NIMS and convert those into FSL format: a matrix of size 3xnscans.
These values should be used to write new .bval/.bvec files for each participant.

[leotozzi88]

Scripts updated to fix this issue for HCP format.

[leotozzi88]

Needs to be fixed for BIDS.

[bstavel]

So, currently in the hcp to bids format, we need to take out the hardcoded bval and bvecs, correct?

[leotozzi88]

Hello Brooke, there are some issues regarding how the HCP pipeline handles our data. We should discuss how it is best to proceed.
To sum up the problem, HCP assumes the first volume is a b0, but we cut them away because they are calibration scans. This makes eddy crash. We can look for a solution with the FSL people... or keep the calibration volume. To me the most rational fix would actually be to "shuffle" the volumes, so that the last volume (which is a b0) is moved at the start. The bvals/bvec would need to be shuffled too.

[leotozzi88]

... of course somewhere deep inside the pipeline, there is a piece that finds the b0s and moves them, but I think it's not doing it based on the bvals. I have asked to the HCP mailing list to tell me where it is (I suspect I found it) and we could possibly change the code there, but Glasser told me to ask Jesper from the FSL mailing list for the details (he developed this part of the pipeline too).