Field mapping files into Bids via Heudiconv not working

Hi there,

I am using Heudiconv to transform my data into BIDS. Unfortunately, my output is variable although I do not change anything in the heuristic file. Sometimes, I get one phasediff file and two magnitude files and sometimes not.
My heuristic file looks like this:

And sometimes my output looks like this:


and sometimes like this:


Does anyone have any idea what I’m doing wrong? Can I still use the single magnitude file for preprocessing or do I need both?

Best regards,

Ahoi hoi @JuliaSchraeder,

thank you very much for your post and welcome to neurostars, it’s great to have you here.

Could you maybe add the heudiconv version you’re using and how you’re calling it (e.g. container, etc.)? Did you compare the dicominfo for sub-003 and sub-004, i.e. is there a chance that their data acquisition (and those of other participants) differed?

Cheers, Peer

If you look at the json text file, what is the manufacturer?

  • If your manufacturer is Siemens, is it possible that your PACS system is not familiar with the fact that Siemens reuses instance numbers for fieldmaps, which might lead to some images overwriting others?
  • If your manufacturer is GE or Philips, is it possible that the sequence was set up for deriving a field map (in Hz) for some participants and set up to store raw images for others? For example, see my notes on the variation in ADNI datasets and how to check these with visual inspection.

By the way, since DICOM images typically have random names, it is sometimes useful to rename your DICOM images to understand how many images you have for each series. For example, the command:

 dcm2niix -r y -f %t/%s_%p/%4r_%o.dcm /path/to/DICOMs

Will rename DICOM images based on study time (t), series number (s), protocol name §, instance number ® and instance UID (o). You can then see if the two participants had similar or different numbers of images for their field map series.

Hi Peer,

thanks for your quick reply!

I used the heudiconv version 0.9.0 and call it with docker. I compared the dicominfo file of these two subjects but the information is the same.

I asked the guys in our the brain imaging facility and we seem to have problems with the dicom images. Sometimes, field map images get lost so I assume now that this is not a transformation problem but the data I used are crappy -.-

but thanks a lot for your support!!


Hi Chris,

the manufacture is Siemens :slight_smile:

thanks for the information especially with the renaming of the DICOMs !