✨ Potential idea: repopulate certain Nifti header fields

[sgiavasis]

Related problem

Certain unused Nifti header fields can sometimes be overwritten by a tool somewhere in the pipeline.
For example, the TR value can be stored in pixdim4 and then overwritten with something like the number of TRs/timepoints later on, depending on the tool.

C-PAC relies on the BIDS sidecars for the TR and other usual Nifti header information, not the Nifti headers, so processing is not impacted by this.
However, users who may rely on older tools that do not use BIDS may need the Nifti headers intact when running those tools for post-processing analysis.

Proposed feature

We could possibly implement a feature where we simply make sure or copy over the original Nifti header to the final pipeline outputs, as appropriate - as long as this information remains accurate by the end of the pipeline.
At the very least, the TR can be held stable.

Acceptance criteria

• TR in pixdim4 or other Nifti header fields still present and accurate.
• Other fields that may be relevant are still present in the Nifti headers, as we assess them.

Alternatives

No response

Additional context

No response

[birajstha]

Following AFNI tools found to reset pixdim[4] to the Time Step value in 3dinfo.

1. 3dTproject
2. 3dcalc
3. TCat

raw file (pixdim4=0.8) -> split -> TCat = preserved (pixdim4 = 0.8)
desc_preproc (pixdim=1) -> update_pixdim=0.8 -> split -> TCat = reset (pixdim4 = 1)

[birajstha]

Looking into output of CPAC for a particular sub, ses, scan, whose native TR and pixdim[4] = 0.8. It was obvious that this happened at different places in the pipeline.

File Name	Status
sub-PA001_ses-V1W1_task-facesmatching_run-1_space-MNI152NLin2009cAsym_desc-head_bold.nii.gz	"0.0, needs updating"
sub-PA001_ses-V1W1_task-facesmatching_run-1_space-MNI152NLin2009cAsym_reg-noGSR_desc-preproc1_bold.nii.gz	"1.0, needs updating"
sub-PA001_ses-V1W1_task-facesmatching_run-1_desc-preproc_bold.nii.gz	Already 0.8
sub-PA001_ses-V1W1_task-facesmatching_run-1_space-MNI152NLin2009cAsym_reg-GSR_desc-preproc2_bold.nii.gz	"1.0, needs updating"
sub-PA001_ses-V1W1_task-facesmatching_run-1_desc-mean_bold.nii.gz	"None, needs updating"

So, @sgiavasis suggested to build in functionality to check the pixdim4 field of input raw bold files, then at the end, check the pixdim4 and update it if it has changed.