AFNI version missing in 3dinfo

AFNI Message Board
1

Hi AFNI team,
I am wondering the possible reason for a missing field of AFNI version in the output of 3dinfo. For example, I have the output of 3dinfo from one subject like this:

----- HISTORY -----
[lzh@node24.cluster: Thu Sep 27 18:04:07 2018] {AFNI_18.0.09:linux_xorg7_64} 3dDespike -NEW -nomask -prefix r2_despike.nii.gz r1_original.nii.gz
[lzh@node24.cluster: Thu Sep 27 18:04:37 2018] {AFNI_18.0.09:linux_xorg7_64} 3dresample -orient LPI -prefix r3_reorient.nii.gz -inset r2_despike.nii.gz
[lzh@node24.cluster: Thu Sep 27 18:04:59 2018] 3dvolreg -prefix r4_volreg.nii.gz -base 16 -1Dfile ./motion_parameters.1D r3_reorient.nii.gz

See the AFNI version (AFNI_18.0.09:linux_xorg7_64) has been printed out for the first two steps of 3dDespike and 3dresample, but this version information is missing for the last step of 3dvolreg.

However, if I check the same file from a different subject, I got the following:

----- HISTORY -----
[lzh@node6.cluster: Mon Oct 1 09:36:46 2018] {AFNI_18.0.09:linux_xorg7_64} 3dDespike -NEW -nomask -prefix r2_despike.nii.gz r1_original.nii.gz
[lzh@node6.cluster: Mon Oct 1 09:37:04 2018] {AFNI_18.0.09:linux_xorg7_64} 3dresample -orient LPI -prefix r3_reorient.nii.gz -inset r2_despike.nii.gz
[lzh@node6.cluster: Mon Oct 1 09:37:17 2018] {AFNI_18.0.09:linux_xorg7_64} 3dvolreg -prefix r4_volreg.nii.gz -base 63 -1Dfile ./motion_parameters.1D r3_reorient.nii.gz

See the version information are shown for all of the three steps performed in this subject.

I noticed this issue when trying to repeat the “3dvolreg” in the first subject shown above. In that subject, the head motion file (motion_parameters.1D) generated on “Thu Sep 27 18:04:59 2018” is different than what I can currently get with exactly the same command line and input file. I am attaching plots of one movement dimension generated back then in 2018 (the black line) and generated currently (the red line). The two curves look like each other but are not identical. This issue does not exist in the second subject whose 3dinfo shows complete AFNI version. In the second subject, the head motion estimates are identical between now and then.

headmotion.jpg
headmotion.jpg1762×866 167 KB

2

Hi, Zhihao-

That’s weird. Ummm, is it possible that the “node24.cluster” had some other (older?) version of 3dvolreg visible on it, by chance? That’s the only main difference I see between those commands (the node it was run on). Is that possible to test at all?

–pt

3

Thank you Paul!
Do you know how old a “3dvolreg” needs to be to skip it’s version footprint for “3dinfo”?

4

It seems that it went in May 3, 2017.

  • rick
5

Thank you Rick!
So do you know if there is a major change in 3dvolreg since May.2017, so that amplitudes of head motions are estimated higher than before that time?

6

On March 8, 2016, the default resampling method was changed from Fourier to heptic. You could add -Fourier and see if the new method becomes more like the old.

  • rick
7

Hi, Zhihao-

Another thing that came up while we were discussing this briefly-- FreeSurfer actually distributes AFNI’s 3dvolreg, and has for quite some time. It used to just be called “3dvolreg”, but to avoid path/version confusion, they changed its name to “3dvolreg.afni”. It is an older 3dvolreg—the compile date on that appears to be Aug 29, 2012, even in FS 6.0.0—but do you think it is possible that you might have a very old FreeSurfer on that node, before they renamed it? (This is an outside-the-box idea for hypothesizing how a different form of 3dvolreg might have come to be run on your data; it is very likely this is not the actual reason, but maybe something to check.)

As to what might have changed, it is hard to say without knowing the date of “that” 3dvolreg; there are a number of tweaks/changes from around/before 2017:
https://github.com/afni/afni/commits/master/src/3dvolreg.c

–pt

8

Hi Paul and Rick,
You guys are awesome!
It turns out that there does exist a 10 years old “3dvolreg” (2009) sitting in the FreeSurfer’s path on my node24, together with the “3dvolreg_afni”. Thank you so much for this valuable insight!

9

Hi, Zhihao-

Cool, glad that solved the mystery!

–pt