Orientation problem labels-disc M12 subjects

[plbenveniste]

I identified an orientation problem in the sub-van111_ses-M12_PSIR_lesion-manual_label.nii.gz.

The dimension is : (16, 336, 336) instead of (336, 336,16)

→ Currently investigating how to solve and if other files have the same problems

The files come from "Discs labelling Hourglass"

tagging @NathanMolinier @valosekj

[valosekj]

Good catch!

The original PSIR image looks good (it has 336, 336, 16 dimensions; which also corresponds with M0):

valosek@macbook-pro:~/data/data.neuro.polymtl.ca/canproco/sub-van111$ sct_image -i ses-M12/anat/sub-van111_ses-M12_PSIR.nii.gz -header | grep "dim"
dim		[3, 336, 336, 16, 1, 1, 1, 1]

valosek@macbook-pro:~/data/data.neuro.polymtl.ca/canproco/sub-van111$ sct_image -i ses-M0/anat/sub-van111_ses-M0_PSIR.nii.gz -header | grep "dim" 
dim		[3, 336, 336, 16, 1, 1, 1, 1]

Maybe the disc labeling was done on the reoriented image? If so, try to reorient sub-van111_ses-M12_PSIR_lesion-manual_label.nii.gz to the PSIR native orientation (AIL).

[NathanMolinier]

Indeed the images were reoriented before creating the labels and I forgot to reuse the original orientation of the image.

However, it is a good reminder that we always need to make sure that we check the orientation before treating an image. The class Image (from SCT) is suited for this task.

[valosekj]

Okay, I see! Could you please reorient the disc labels back to the original orientation and push them to git-annex?

[plbenveniste]

This was solved during the re-orientation of all derivatives file in this issue : ivadomed/ms-lesion-agnostic#1 (comment)