#4 add basic clinical rev output

[AnilMistry]

Towards #4

[AnilMistry]

[tomaroberts]

Thanks Anil – screenshot looks great. I will review asap.

NB: the GitHub Action is failing because TotalSegmentator package calls sklearn as a package, which is a deprecated term as it should be scikit-learn. I've got a separate PR which sorts this out.

[AnilMistry]

Updates made as requested. This works locally for me. Instead of having the ct niftis and masks in the input folder, I have them in a separated directory in the repo root, and pull them in when needed, after commenting out the subprocesses.

[AnilMistry]

I have update the paths so that it uses each operators' working directory to place their outputs, then assign the datapaths as the output. This works locally with premade Nifti files for th CT and Segments.

[tomaroberts]

Made some changes to ensure it works with a folder of DICOMs. Had to correct some path issues. Also, I discovered that the PDF creation fails if one of the DICOM elements required to fill in the PDF is missing, so I've added a little function to check for this and simply output "Unknown" if it doesn't exists.

TODO:

• create new MAP and run end-to-end on DGX/headnode
• run MAP on AIDE and check PDF functions as expected

[tomaroberts]

Tested latest commit on DGX (nb: remember to change line 26 in Dockerfile to this branch name) and it successfully ran end-to-end (DICOM in/out). Next to test on AIDE.

Few things to note:

• reportlab was a mild pain, but working now
• required libcairo2-dev install within container Ubuntu

[tomaroberts]

27 should be resolved by 0da0875 – confirm in testing

[AnilMistry]

Ready for review, PDF should look like this:

[tomaroberts]

Confirmed working on AIDE (with a few tweaks to AIDE code) by Joe B:

TODO: merge, create final 0.2.0 container, test on AIDE PreProd, release on GH, release to AIDE App Store.