Image/Segmentation Origin and Orientation Issues

[juthoffSheffield]

Hello, firstly thank you for supplying this code it is performing very well on some test cases we have.

I did encounter one issue I would like to flag - I have supplied a supplemental jupyter notebook, .py file, and screenshot using the test data provided (subject_1).

From those files you can see there is a difference in the image origin between the lvsa_.nii.gz file (indicated as the 4D raw data in ReadMe), the enlarged (i.e. lvsa_ES_enlarged_SR), and the segmentation (i.e. PHsegmentation_ES).

At first I expected all non-'original-nii' images to have an origin of (0,0,0..) as the IRTK:headertool function is called with -reset which should set the origin and orientation to zeros and identity matrix. However, this does not seem to be the case for the origin.

Along these lines, it would be assistive to have saved a transform to allow users to easily register the segmentations created back into the original image's origin and orientation space.

4DsegmentOutputOriginOrientation_Issue.zip

Cheers,
Johanna