FSL Topup taking 5 hours?

,
1

Hi,

I am trying to run FSL top-up to create a field map as per the website below (method 2). I collected two EPI sequences. One with 4 volumes in P>>A direction (dist map) and the other is my task EP sequence with 637 volumes in A >> P.

https://lcni.uoregon.edu/kb-articles/kb-0003

I merged the dist and task EPI sequences and created a datain file.

In the terminal window I am typing…
topup -v --imain=merged_scan.nii.gz --datain=datain_file --config=b02b0.cnf --fout=my_fieldmap --iout=se_epi_unwarped

This does not error and seems to be running but is taking a very long time (over 5 hours) and the process isn’t finalizing before I am logged out.
Is this typical? If so, is there some way to restart the processing based on where I was when I was kicked off the computer (rather than having to restart every time). Is there some way to speed up processing?

Or does anyone know of any alternative processes to create a field map that might be faster? it would probably be best to create the fieldmap without fMRI prep (I think this is possible), but wanted to check the output before running.

2

Yes, it is normal that topup takes a long time when you feed it with so many volumes. Try to feed in topup the 4 volumes in PA an the first 4 volumes of the functional AP serie for example. It would go even much faster if you just give one volume in AP and one volume in PA.

Of note HCP recently moved to a strategy to select the best b0 for topup in order to decrease the processing time and also to get rid of volumes artefacted with motion, something that you may want to try to optimize your process.

Why would you want to create a fieldmap out of fmriprep? You don’t plan to use fmriprep output?

3

Ok, thanks. I wasn’t aware you could do that. Will cut down the number of brain volumes in the task image. Thanks so much!