Error during unwarp_resampling

1

Hi,

I’ve been trying to implement fmriprep to preprocess my data (MB=8, TR=0.735, scans=490, 3T Siemens). While running, it seems it came across an error and crashed but I’m not quite sure how to fix it.
The protocol also includes dwi (A–>P, and P–>A), so PEPOLAR distortion correction was performed (or at least it tried to?)…

I’m sharing a google drive link that includes the logs/reports.
https://drive.google.com/drive/folders/1bgyqtjEdpgiMlf9-m82K7hsKDGA9-gSw?usp=sharing

Thanks for all your help!
-Erin.

2

Hi Erin,

DWI acquisitions would not be used in fMRIPrep.

It would help if you could answer the following questions:

  1. Are your data BIDS valid?
  2. Is this error subject specific?
  3. How much memory/cpu are you devoting to the job?
  4. What additional arguments/flags are you inputting into the fmriprep command?

Thanks,
Steven

3

Hi Steven,

Thanks for getting back to me so quickly!

  1. They are BIDS validated.
    Maybe I’m not understanding it correctly. But followed “Case 4: Multiple phase encoded directions (“pepolar”)” section from the BIDS specification guideline to put AP in the dwi directory, and PA file in the fmap directory with the appropriate json file. Sorry, I’m still a bit confused on how to utilize diffusion data for susceptibility distortion correction…
  2. Yes, it is subject specific. Only ran on one subject for testing.
  3. Devoted 4 cpus, with 10 GB mem
  4. I’m sharing my command arguments.
    fmriprep-docker $bids_root_dir $bids_root_dir/derivatives
    participant
    –participant-label $subj
    –skip-bids-validation
    –md-only-boilerplate
    –fs-license-file $HOME/fmriprep_test/derivatives/license.txt
    –use-syn-sdc
    –ignore slicetiming
    –dummy-scans 10
    –fs-no-reconall
    –output-spaces MNI152NLin2009cAsym:res-2
    –nthreads $nthreads
    –stop-on-first-crash
    –mem_mb $mem_mb
    -v

Hope this helps!
Thanks again!
Erin.

4

A few more questions:

  1. Is the PA file a full diffusion run, or just a PA reference? Also, you would want to make sure if you are using a PA fieldmap, that your BOLD run is AP.
  2. Does the fmap json have an IntendedFor field that points to the functional file you want to associate it with?
  3. If you have a fmap, why are you also including –use-syn-sdc?

Steven

5

Hi Steven,

sorry for the late reply. Turns out the error was a totally different issue.
After checking all the pointers you had mentioned, it seems to run smoothly.
I had -use-syn-sdc option for subjs that did not have dwi AP and PA fmap data, but now I see I should turn it off.

Thanks for all your help!
Erin.