Upon trying out eddy I would ask

​for some details​

:
I was wondering if eddy would work

​ properly​

with DWIs only

​ from​

​ 1

 set of phase encoding direction

​ (let's say AP)​

but with opposite sets of B0s

​ (AP and PA)

?

​ ​

In my

​ demo​

dataset I have 1B0 and 30 DWIs, but sampled on the half sphere. So this is not like the recommended setup according to http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/EDDY#If_you_haven.27t_already_acquired_your_data

There is a good chance it will still work will with your data. As we have gained more experience of using eddy we have seen many cases where it has worked on half sphere data. Just run topup on your b0’s, putting the AP first in the 4D file you pass to topup. Then run eddy with the same AP b0 as the first volume in the 4D file you pass into eddy. The index file should just be 31 ones.

In the minimal processing pipeline of HCP data paper (M.F. Glasser et al. / NeuroImage 80 (2013) 105–124) the DWI preprocessing starts with an intensity normalization of the B0s, would it make a big difference omitting that step

​ f

or my own data with multiple B0s?​

​

I am not sure what this is.

Also as I see there is

​1 processed/

corrected *.nii

​ ​

(..\ConnectomeData\

​*

\T1w\Diffusion\data.nii.gz) and bvec

​ ​

for a subject from the 'connectome in the box'. How can I achive that with my own data (with opposing PEs in B0

​s​

and DWIs of course), should I use the --resamp lsr option? From a previous thread it looks like yes (https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=fsl;feea3a93.1510), but then how

​can 

I

​get 

one bvec

​ file​

? Or it will be provided via eddy because of the --resamp lsr flag?

Since you only have one PE-direction for each dwi you cannot use --resamp=lsr. Just use the default. The bvec file is not produced by eddy. It is something you need to supply. If you are using 5.09 you will get a file with rotated bvecs as part of your output.

I hope that answers your questions.

Jesper

​Regards

,
Szabolcs David
UMC Utrecht

​​