Field Maps Not Being Applied in FMRIPrep #3386
Comments
|
Can you pare your dataset down to a single subject with one anatomical, one BOLD series (all echos) and the fieldmap that should correct your BOLD file? Then could you share the file listing and the JSON files for all fieldmap and BOLD scans? |
|
I've attached the link to this here. |
|
So this also works for me. I've attempted to run with basically the same parameters: Are you using a separate B0FieldIdentifier for each subject's fieldmaps? |
No. Should I be doing so? At this point, I've just been trying to get this one subject to work for me since it seems others can get it going, so I don't know how relevent this would be for me yet. Is there some dependency I might be missing that is causing the issue? Or a problem with the code I am using to initiate docker and fmriprep? |
|
Yes, you need to use a different B0FieldIdentifier for every fieldmap in a dataset, not just every fieldmap within a subject. It's possible that the error is that, when the files are collected across subjects, they don't form a coherent basis to estimate a fieldmap and so are dropped. |
|
I have modified all of the field map and anat .jsons to include an individuating marker: fmap files: "B0FieldIdentifier": "fmap_sub-055_14" Nevertheless, the problem persists. Is there some possibility that something is being retained in memory and it is using an older setting? Some dependency missing? Something wrong with my initial input code? |
|
Try something like I can't see anything in your script. You're not using any features that would include configuration from a previous run. You can always try clearing out your working directory, but the working directory only matters after the workflow is built, and the problem is with building the workflow. What happens if you just build a dataset with one subject, as we've done with your example data? That'd be the closest apples-to-apples comparison. |
|
I have just tried the exact same dataset with one subject and pared down to one within its own BIDS folder. The issue persists. At this point, it has to be some dependency I am missing or something, right? Is that possible? Update: I've also just tried it with Singularity instead of Docker using the following script: Define memory limits for fMRIPrepmem_mb="63500" Make outputs read/writable for groupumask 0002 Define versions and pathsfmriprep_version="24.1.0rc2" echo "TemplateFlow Directory: $templateflow_dir" Check if subject number argument is providedif [ $# -ne 1 ]; then Import subject ID from command linesubject=$1 Do fMRIPrep directly on the original data without copyingecho "Running fMRIPrep command..." fmriprep_cmd="singularity run -e --containall echo "Submitted fmriprep job for: ${subject}" The issue persists with both Docker and Singularity, so the issue really must be something trivial on my end, but I can't fathom what it could be. |
What happened?
My field map files are not being used. I have submitted this issue on Neurostars, and sent a zip folder of one subject to an admin there. He was able to get the field maps to be associated with the correct runs using the exact same data without making any modifications (see thread here for more information).
He was able to get the indicator for field maps being appropriately applied:
241021-10:57:48,361 nipype.workflow INFO: Setting-up fieldmap "fmap_19" (EstimatorType.PEPOLAR) with <sub-055_acq-mbep2d19mm_dir-AP_epi.nii.gz, sub-055_acq-mbep2d19mm_dir-PA_epi.nii.gz> 2024-10-21 10:57:48,361 [ INFO] Setting-up fieldmap "fmap_19" (EstimatorType.PEPOLAR) with <sub-055_acq-mbep2d19mm_dir-AP_epi.nii.gz, sub-055_acq-mbep2d19mm_dir-PA_epi.nii.gz> 241021-10:57:48,377 nipype.workflow INFO: Setting-up fieldmap "fmap_14" (EstimatorType.PEPOLAR) with <sub-055_acq-mbep2d14mm_dir-AP_epi.nii.gz, sub-055_acq-mbep2d14mm_dir-PA_epi.nii.gz> 2024-10-21 10:57:48,377 [ INFO] Setting-up fieldmap "fmap_14" (EstimatorType.PEPOLAR) with <sub-055_acq-mbep2d14mm_dir-AP_epi.nii.gz, sub-055_acq-mbep2d14mm_dir-PA_epi.nii.gz>When I run the exact same version 21. 4.1oc2 (I've also tried 24.1.1) of fMRIPrep, SDC is not being done on any data and I am not getting any message related to field maps like he is. If it is the exact same data without any modifications (folder structure, file names, .json contents, etc. are all identical), why are field maps not being applied for me?
What command did you use?
What version of fMRIPrep are you running?
24.1.1
How are you running fMRIPrep?
Docker
Is your data BIDS valid?
Yes
Are you reusing any previously computed results?
No
Please copy and paste any relevant log output.
Additional information / screenshots
Screenshot showing SDC is not being applied.
The text was updated successfully, but these errors were encountered: