Coreg error (maybe)

[zeynabalshelh]

Hi,

I'm new to preclinical imaging and I'm trying to use this software to preprocess some rat scans.

This is the command used. :

docker run -it --rm -v /home/zeynab/projects/pilot_usyd/data/sub-pilot1/:/sub-pilot1:ro -v /home/zeynab/projects/pilot_usyd/data/preprocess_outputs:/preprocess_outputs/ ghcr.io/cobralab/rabies:nightly preprocess /sub-pilot1 /preprocess_outputs --apply_STC --TR 2 --commonspace_reg masking=true,brain_extraction=false,template_registration=SyN,fast_commonspace=false

I am looking at the QC anat_inho_cor report and something clearly went wrong:

These are the other two reports generated. Where something is clearly not working correctly.

I'm not sure where to go from here. Please advise.

Thanks!

[gdevenyi]

Looks like your input image is upside down. RABIES expects images to be in RAS orientation and follow the NIFTI standards. Perhaps whatever conversion tool you're using isn't standards compliant?

[gdevenyi]

That doesn't look to me like an issue with the file. This looks like either permissions or path issues.

Please check that you've mounted your paths correctly into the container.

Our recommended configuration is singularitry, not docker, can you please try that?

[zeynabalshelh]

Hi,

I changed the permissions of the bids folder and that seemed to work? I didn't realise it needed writing permission to the bids folder.

Ok, so we're running into another issue here:

I am using the same command as before but the correction is way off.

Is this related to the orientation of the image?

[gdevenyi]

I didn't realise it needed writing permission to the bids folder.

Permissions with docker are complicated, its not really designed/intended for standard-user based use cases (again, singularity is the solution here)

As I said previously, your images are way over-sampled in terms of field-of-view, a basic assumption of registration is a (near) 1:1 relationship of the content of the images. If you look at the atlas FOV, it is just the brain, whereas you have the whole body and empty space around that. We can't supply a larger FOV atlas, so you will need to do some pre-cropping.

[Gab-D-G]

Hi @zeynabalshelh, first are you using the right template? (i.e. rat template if these are rat scans) If you are, then I would recommend trying the --anat_autobox parameter, which will crop images to remove extra non-brain tissue which is likely causing issues in your case.

[zeynabalshelh]

Thank you both!

Using --anat_autobox did not work but precropping the image did:

Thanks for your help. We will try to implement your suggestion to use a smaller FOV and isotropic data for our future scans.

[zeynabalshelh]

Hi,

I am now testing this pipeline on data collected using a Bruker 9.4T. I converted the data using brkraw. I have tried the following:

sudo docker run -it --rm -v /home/zeynab/projects/UNSW/data/bids/:/bids_dataset:ro -v /home/zeynab/projects/UNSW/data/preprocess_outputs2/:/preprocess_outputs2 ghcr.io/cobralab/rabies:nightly -p MultiProc preprocess /bids_dataset /preprocess_outputs2 --anat_inho_cor method=Affine,otsu_thresh=4,multiotsu=false

(i tried SYN, multiotsu on and off and differnt otsu thresholds).

For each one, it seems that the resampled mask is way off.

Each result from the different options I select are slightly different but they are all either masking a small area of the brain or sampling tissue outside the brain. I also tried using --anat_autobox. But I am having the same problem.

I don't know how to fix this. Please help.

Thank you!

[Gab-D-G]

Is this a rat brain? if so you need to provide a rat atlas during preprocessing, because by default you will be using a mouse brain mask (hence the tiny mask you are obtaining). You can use the SIGMA rat atlas for instance https://zenodo.org/records/10635831. If you are using your own masks, can we see the template_files.png in the preprocess_QC_report?

[zeynabalshelh]

Hi,

Thank you for your response and patience with me as I learn these steps. Yes this is a rat brain.
This is the template_files.png from the report:

Is there something I am missing? Do I need to specify to use my own mask?

Thanks!

[Gab-D-G]

Hi,
That is the default mouse atlas within RABIES, not a rat atlas. So that is the origin of your error. You need to provide your own template files with the --anat_template parameters and associated mask parameters. See the CLI --help for preprocessing https://rabies.readthedocs.io/en/latest/running_the_software.html.