I see two highly unusual properties of these images:
- The image is turned inside out (i.e., its axes form a left-handed coordinate system). This is an error, it is not valid NIFTI. This can be fixed by resampling the volume into a right-handed coordinate system by using
Crop volume
module (using default settings, using the entire volume’s region). - Voxel type is float. This is not technically an error, but there should be no reason for a clinical CT to use this type, and using this voxel type for segmentation is even worse (as you need to store discrete label values). You can fix this by converting to
Short
type usingCast scalar volume
module.
Slicer should be able to cope with such images (at least it should display a warning or error when loading them), but we come across with such images so rarely that it is hard to justify investing much time into dealing with them.
I don’t see any errors on Windows (the segmentation is editable without any issue), so it is hard for me to investigate. @muratmaga To narrow down the issue, could you please check if the problem is caused by the first or the second issue? Do you see the issue on linux as well?