antsMotionCorr --write-displacement no output

[toomanycats]

Operating system and version

CentOS 7

CPU architecture

x86_64 (PC, Intel Mac, other Intel/AMD)

ANTs code version

ants-2.5.0

ANTs installation type

Compiled from source

Summary of the problem

When I use antsMotionCorr with the --write-displacements options I see no displacement output.

The input 4D NiFTI is a PET series.

Commands to reproduce the problem.

antsMotionCorr --write-displacement -m CC -d 3 -o ".nii.gz" -p Linear -t Rigid -o ave.nii.gz -a PET_4frames.nii.gz

Output of the command with verbose output.

ANTSVERSION=ants-2.5.0

ls
ave.nii.gz

Data to reproduce the problem

BRAIN_AC_ADNIrecon_2024.02.21.nii.gz

[cookpa]

I think it is --write-displacement 1. I will update the usage

[toomanycats]

I forgot to add that I already tried adding an explicit "true":

antsMotionCorr -v --write-displacement 1 -d 3 -p Linear -t Rigid -o ave.nii.gz -a ../Sample/ADNIDII-0-3028-1/BRAIN_AC_ADNIrecon_2024.02.21.nii.gz
ANTSVERSION=ants-2.5.0
verbose
average out ave.nii.gz

Still no love. I keep thinking I'm missing something obvious that I'm not seeing in my call.
Confirming there's really multiple frames:

fslinfo ../Sample/ADNIDII-0-3028-1/BRAIN_AC_ADNIrecon_2024.02.21.nii.gz | grep -P '^dim4'
dim4            6

[cookpa]

You need to do it in two stages:

antsMotionCorr -d 3 -a BRAIN_AC_ADNIrecon_2024.02.21.nii.gz -o ave.nii.gz

Then

antsMotionCorr -d 3 -o [ moco, moco_corrected.nii.gz, moco_avg.nii.gz ] \
    -v 1 --write-displacement 1  -p Linear -t Rigid[ 0.05 ] \
    -m MI[ ave.nii.gz , BRAIN_AC_ADNIrecon_2024.02.21.nii.gz, 1, 32 ] \
    -i 10x10 -f 2x1 -s 1x0

[toomanycats]

Ok thanks ! Perhaps the doc needs an example.
I'm not a skilled C++ programmer, but I find this program difficult to read. There's a lot of procedural code here:

ANTs/Examples/antsMotionCorr.cxx

Line 1746 in 592625b

ANTs::WriteImage<DisplacementIOFieldType>(displacementinv, dfn.c_str());

[toomanycats]

If want motion estimates, should I also use antsMotionCorrStats ?
I was going to compute the rigid motion myself using python.

Cheers and thanks again.

[cookpa]

Yes antsMotionCorrStats, but that only works with the CSV parameter file. It doesn't work for warp fields.

[cookpa]

Examples in the usage would be good.

There are some examples in the README.md on the main page, and others in the wiki.

[toomanycats]

I'm beginning to understand the arguments now ! Cheers.