If I enable crosshair rotation, the segmentation breaks. It does not matter anymore where I segment, the segmentation is added to the very bottom of the rotated slide.
Description
Related Objects
Event Timeline
Try: E130-Daten\BlackSwan\Patients_for_Registration\0HLXYUL7_0HLXYUL7\16\resources\DICOM
I could reproduce it when rotating the axial plane in both orientations, sagittal and coronal, and then drawing a contour in the axial window. I also noticed the way too narrow contrast window. I checked the geometry/matrix and it is one of those that do not have the spacings on the main diagonal. I processed the source image (after conversion to NRRD) with a mini app I had written some weeks ago, which rectifies such images in a way that the direction matrix has its spacings on the main diagonal again. The output image is absolutely congruent to the input image, though. And now it works, both the default image contrast and segmenting in rotated slices.
I do not know how hard it is to identify and fix the actual bug, but we should consider including my mini app as a workaround for such problems. Does only work, when no direction axis is rotated - just permuted axes.
The app called RectifyImage is in the personal/kislinsk/RectifyImage branch:
https://phabricator.mitk.org/source/mitk/browse/personal%252Fkislinsk%252FRectifyImage/Modules/RectifyImage/cmdapps/RectifyImageMiniApp.cpp