missing output files antsMultivariateTemplateConstruction2 + rerunning/what was included in template0.nii?

[gjoue]

Hi,

I ran antsMultivariateTemplateConstruction2.sh (from ANTS v.2.5) to
create study separate group templates for my two image modalities
(T1w-MPRAGE and T2w-TSE). When it was done, I saw that I had the final
template file for each modality, i.e. *_template0.nii.gz
and there were no errors printed to screen (which I had redirected to a
log file -- so I thought all jobs terminated correctly. The group
template files also look fine. However, there are discrepancies in the
number of input individual subject files read in and the different files
output for each subject. For example, in the logs, I see that the number
of _{_GenericAffine.mat files listed was the same as the number of
the input files -- but it was not the number of physical
_GenericAffine.mat files, which in turn was also not the same as
the number of template0WarpedToTemplate.nii.gz files I had either.
Why is this?}

Does this mean I have to rerun everything again or will
antsMultivariateTemplateConstruction2.sh be able to check what has been
completed and process only what is missing?

Is there a way to check what input images have been considered to
generate the template0.nii.gz?

Many thanks for any help!
Gina

[cookpa]

Can you attach the log file and also list the contents of your output directory?

It may be possible to get more information from it

[gjoue]

By log file, you mean what was print to screen during the antsMultivariateTemplateConstruction2 call and not the metriclog.txt files, right? I've attached the directory listing of antsMultivariateTemplateConstruction2's outputs and the messages it printed to screen for the call, as well as, a list of the input files I had sent to the call (for the T1w images only) in the three attached files. And here's a count of the input (169) and output files:

ANTS_ls_out.txt
ANTS_inputList.txt
bnst_43b_make_grpTempl8_T1_OLD.log
:

$ ls ../../SRC/*T1_*N3.nii.gz | wc -l
    169
$ ls T1*WarpedToTemplate.nii.gz | wc -l
    169
$ ls T1*.mat | wc -l
    144
$ ls OTHERoutputs/T1*InverseWarp.nii.gz | wc -l
    144
$ ls OTHERoutputs/T1*[e]Warp.nii.gz | wc -l
    144 

Thank you so very much!
Gina

[cookpa]

The call to AverageAffineTransform in your logs appears to have all 170 transforms, for 170 images. But then there's also T2 output files present. I'm not sure what's going on here. But I think building two separate templates in the same directory is unlikely to work.

If you really want separate templates (that may not be aligned to each other), you can do that. Or you can build a multivariate template if the T1 and T2 for each subject are aligned to each other. It's up to you, but if building a separate template, I'd start each one in a clean directory with only the input for that template.

You can resume template building using your existing OK-seeming templates as initial templates (using -z).

If you continue to have problems, I would suggest updating ANTs, as there might have been script fixes since 2.5.0. Also, we need to know the actual call to the script, as well as the terminal output.

One thing I think I recall fixing post 2.5.0 was dealing with input CSV files from Python, which use Windows newlines by default. If things are acting weird, check if there's windows carriage return characters.