[macOS] Packaged version of CemrgApp does not load DICOMS

[alonsoJASL]

A full explanation of this bug can be found in here. It is unsure when is this going to be addressed by the MITK team, as they were not able to reproduce our error.

This issue is then created here to determine the workarounds necessary to load and convert DICOMs external to the DICOM plugins for our users to handle.

[alonsoJASL]

Ideas

Docker container linked to CemrgCommandLine class
This would be a docker container with either of the following libraries:

• dicom2nifti
• c3d

Building MITK/CemrgApp with a newer version of DCMTK This would be achieved by setting the CMake variable MITK_EXTERNAL_DCMTK to a more recent version of DCMTK than hte one currently in MITK v2018.04.2, which might address the problem.

[OrodRazeghi]

I put together a Docker container for the dicom2nifti python tool and pushed it to the docker hub. The test on one of the example MRI cases was successful. The syntax to test the code is as follows: orodrazeghi/dicom-converter "Dicom Folder" --gantry --inconsistent The gantry and inconsistent flags (dicom2nii options) are by default false. The dicom folder needs to be linked up with the container using the volume parameter. For example: docker run --rm --volume=$(pwd):/Data orodrazeghi/dicom-converter "Mri Heart" The resulting niis will be in a folder called NIIs @alonsoJASL if this works on your example cases, we can put this as the backup option for the dicom converter

[alonsoJASL]

Made initial tests on branch feature/alternative_dicom_reader.

Commit 37778b4: Lays out the initial changes in the CemrgCommandLine class, as well as its implementation in the AtrialScarView plugin.

Further problems:

• Orientation of the dicoms might need to be addressed. When tried a simple segmentation, the LA was not centred at akl.
• Also, the implementation in the plugin was made in the AtrialScarView::ProcessIMGS method, this could change if there's a need for a cleaner code which can be passed on to other plugins.

@OrodRazeghi, have a look whenever you have the chance. It needs thorough testing, although it might not end up taking that much time.

[alonsoJASL]

Commits: 67732e9 and 57ee030 of the feature/alternative_dicom_reader branch.

A feature was added to reorient the loaded NIFTI file to RAI configuration and resample them as isometric. This functionality has been included in the CermgCommonUtils class, so it can be used across all plugins.

At the moment, this is only true in the AtrialScarView plugin. Next stage involves testing with cases where the DICOM conversion is known to be a problem. I'll try to contact the clinicians getting this problem.

It is possible that this change might help with issue #12, where pulmonary Veins don't clip (presumably) because the orientation of the image is not right.

[alonsoJASL]

Closing issue as it has been finalised. Testing done so far indicates that the alternative DICOM reader works well. This is currently in the Development branch.