generate_motion_statistics assumes rotations are last three columns of movement parameter file and other issues re FD/scrubbing

[fcp-indi]

generate_motion_statistics assumes rotations are last three columns of movement parameter file and other issues re FD/scrubbing

[RanjitK]

Hi Maarten,

Thanks a lot for your suggestion. I would replace my FD awk calculation with python code and also discuss the mcflirt and use of afni 3dcalc issue in our developer's meeting. We were not using nibabel to save the nifti file because it changes the header information of the nifti file. Also, 3dcalc is much faster than doing it with python. So we decided to keep it. Thanks for your points and helping us improve the cpac workflows.

[maartenmennes]

Hi Ranjit,

thanks for following up. What header information does it alter then? I did some quick comparisons and if I use the following:

temp_image = nb.Nifti1Image(data, affine=img.get_affine(), header=img.get_header())

then I can't see any change in the header (using fslhd and 3dinfo, except for the identifier code).

Looks like you are right regarding the speed of 3dcalc, quick test:
3dcalc: 6.2 seconds
nibabel save: 10.4 seconds