Ok Jesper, let me hear the worst... We are assessing the expedient option of taking 2 soccer ball bvecs one used for topup one topdown and moving the topdown to another shell b value double the topup. Your thoughts. And don't hold back :-).
J
well, as I said it might work. I would even say it probably will work ok. You can even check if it works before you spend time on collecting any data by grabbing some of the unprocessed HCP where there are L->R and R->L data for each shell, and throw one of them away for each shell.
In general though I would personally always go for balance. It is likely that correction will work slightly less well for the higher b-values because of SNR, noise floor etc. And I would not like to have that confounded with b-value.
Jesper
Sent from my iPhone
Dear Jeff,
Hi saad. As a follow on: can topup/eddy be used if top up is one shell and topdown is another?
I think there is a chance it might work, though I have never tried it. It seems like a crazy scheme though and unless you have already collected the data I would rethink.
Jesper
Regards
Jeff
Sent from my iPhone
Hi
That’s correct. You also need to turn on the gradient non-linearity correction (-g option) which will make use of the grad_dev.nii.gz file (so you need that in the subject folder as well). You also need this if you run dtifit (—gradnonlin option).
Cheers
Saad
Dear experts,
I have a question regarding anlysis of the multi-shell human connectome data. When running bedpostx (from the command line), I have to state the following, right?
bedpostx <subject directory> --model=2
Apart from making sure bedpostx uses a multi-shell model, do I have to change any other setting from the bedpostx or probtrackx commands?
Thanks for the info!
Sincere regards,
Nele
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
