Dear Daniela,
>
> I would like to confirm that I have correctly used topup to correct for distortion in my epi data, using fieldmaps, which will be used for subsequent applytopup
I’m sorry for sounding pedantic, but can I first of all suggest that you only use the word “fieldmap” for actual maps of the off-resonance field? I think it will help you avoiding misunderstandings.
A “fieldmap" can either be measured directly using a dual echo time sequence, or it can be calculated by for example topup. I.e., it is the output of topup that is a “fieldmap”. The input images to topup are just regular EPI images with different phase encoding.
>
> Here are the commands I hope to use for the first couple of steps.
>
>
> #STEP1: merge together fieldmap with phase encoding direction PA and AP to prepare for topup; output boldmerge:
> fslmerge -t boldmerge.nii PAfieldmap.nii APfieldmap.nii
>
> #STEP2: runtopup to prepare for applytopup:
> topup --imain=boldmerge.nii --datain=/acquisition_parameters_merged.txt --config=b02b0.cnf --out=topupresults
>
> My specific question is about the order of inputs in the file acquisition_parameters_merged.txt. Is what I have shown the correct order, since the input to my merge step was PA then AP (my understanding is that the -1 in 2nd position, row1 is PA and the 1 in second position, row 2 is AP)?
>
> 0 -1 0 1
> 0 1 0 1
>
You have specified that the phase-encoding is in the y-direction, and that is correct. As for the signs it doesn’t really matter if you specify
0 -1 0 t
0 1 0 t
or
0 1 0 t
0 -1 0 t
as long as you are consistent between topup and applytopup (or eddy). Of course one of them is “correct” in that the field has the correct sign, and the easiest way to check that is to look at your output field. Re-run topup with --fout=my_field and take a look at my_field.nii.gz. If you see hotspots (high field) along the ear canals and around the sinuses, then it is correct. If you see coldspots (low field) in those places, then it is sign reversed.
I put t instead of 1 for the fourth column above. Again, it doesn’t matter what t is as long as you are consistent. It will only act as a scaling factor of the resulting field. BUT, some FSL software (for example eddy) checks to see if it has a reasonable value, so it is a good idea to use a reasonable value. I suggest using 0.05 if you don’t know the true value.
Jesper
> I ask because I inherited a script in which they reversed these, which may be an error in that person's script; otherwise perhaps instead my logic is off somewhere. I would like to be sure!
>
> Thanks,
> Daniela
|