Can the wrong values in these fields of NIFTI files affect the preprocessing results in AFNI?

AFNI Message Board
1

Hi AFNI experts,
The dicom files in my fMRI experiment were transformed into the .nii format by MRIconvert firstly before preprocessing.
I found some fields in the .nii files generated by different versions of MRIconvert are different, though the inputted data are exactly the same.


nifti_tool -diff_hdr -infiles ver1.nii ver2.nii
  name                offset  nvals  values
  ------------------- ------  -----  ------
  datatype              70      1    4
  datatype              70      1    512
  pixdim                76      8    -1.0 1.5 1.5 1.49995 2.0 0.0 0.0 0.0
  pixdim                76      8    1.0 1.5 1.5 1.49995 4.0 0.0 0.0 0.0
  slice_duration       132      1    0.0762
  slice_duration       132      1    0.05
  quatern_b            256      1    -0.005177
  quatern_b            256      1    0.030369
  quatern_c            260      1    -0.985187
  quatern_c            260      1    0.168696
  quatern_d            264      1    0.168696
  quatern_d            264      1    0.985187
  qoffset_y            272      1    -77.154175
  qoffset_y            272      1    102.401825
  qoffset_z            276      1    2.340701
  qoffset_z            276      1    -61.537697

The files ver1.nii and ver2.nii are generated by the different version of MRIconvert with the same data inputted.
Whether these fields with the different values can affect the results of some steps in preprocessing in afni?
For example, whether the difference in the values of the field ‘slice_duration’ can change the 3dTshift results?
Thank you in advance!
Yu

2

Hi Yu,

It is hard to know the affect of the quatern and
qoffset values (why would the offsets differ? that
is strange). But it looks like the TR has changed
from 2.0 to 4.0 between them, as well as that
slice_duration, both of which are a bit strange.

You could see what either Dimon or dcm2niix_afni do
with the Dicom files, just for comparison. It would
be good to know what the correct values are.

  • rick
3

Hi Rick,
Thank you!
The only different fields are ‘pixdim’ and ‘slice_duration’ if I choose some other EPI files for comparison:


nifti_tool -diff_hdr -infiles ver1.nii ver2.nii
name                offset  nvals  values
  ------------------- ------  -----  ------
  pixdim                76      8    1.0 1.5 1.5 1.49995 4.0 0.0 0.0 0.0
  pixdim                76      8    1.0 1.5 1.5 1.49995 2.0 0.0 0.0 0.0
  slice_duration       132      1    0.05
  slice_duration       132      1    0.0762

I also tried Dimon. The outputted pixdim is ‘1.0 1.5 1.5 1.49995 2.0 0.0 0.0 0.0’ and the slice_duration is ‘0’.
Then I tried 3dWarp and 3drefit to modify these nii files. Some brik files were generated.
I want to know whether 3dWarp and 3drefit work well.
Do you know how to display header field diffs between two brik datasets?
Yu

4

Hi Yu,

Why are you warping them (3dWarp)? That does not
seem like a good idea, both for the extra blur in the data
and because you are talking about running 3dTshift,
which after a warp would be inappropriate.

  • rick