antsApplyTransforms output is resampled to MNI grid, but qform/sform_code remains “Scanner Anat” — expected behavior?

1

Hi all,

I’m applying spatial normalization to fMRI data using ANTs, and I’d like to ask whether the following behavior of antsApplyTransforms is expected or if I’m missing something.

Context

  • Data were preprocessed with fMRIPrep
  • I am warping tedana-denoised BOLD data from native (boldref) space to MNI152NLin2009cAsym
  • Transforms used:
    • from-boldref_to-T1w_mode-image_desc-coreg_xfm.txt
    • from-T1w_to-MNI152NLin2009cAsym_mode-image_xfm.h5
  • Reference image:
    • *_space-MNI152NLin2009cAsym_boldref.nii.gz from fMRIPrep

Command

antsApplyTransforms \
  -d 3 -e 3 \
  -i denoised_native_bold.nii.gz \
  -r space-MNI152NLin2009cAsym_boldref.nii.gz \
  -o denoised_MNI_bold.nii.gz \
  -t from-T1w_to-MNI152NLin2009cAsym_mode-image_xfm.h5 \
  -t from-boldref_to-T1w_mode-image_desc-coreg_xfm.txt

What I observe

  • :white_check_mark: Output image has identical grid and voxel size as the MNI reference
    (dim and pixdim perfectly match)
  • :x: However, the output NIfTI header shows:
qform_name   Scanner Anat
qform_code   1
sform_name   Scanner Anat
sform_code   1

while the reference has:

qform_name   MNI_152
qform_code   4
sform_name   MNI_152
sform_code   4

Visually, the output aligns correctly with the MNI template when overlaid.

Question

Is it expected behavior that antsApplyTransforms:

  • resamples data correctly into MNI space but does not update qform/sform codes,
  • leaving them as Scanner Anat unless explicitly modified afterwards?

In other words:

  • Is it safe to treat this as a header/metadata issue only,
  • and fix it downstream (e.g., with fslcpgeom / fslorient)?

I want to be sure I’m not silently mislabeling data or misunderstanding ANTs’ handling of NIfTI headers.

Any confirmation or best-practice recommendations would be greatly appreciated.
Thanks a lot!