Arbitrary affine matrices for 20 train samples?

[alexandrebone]

Dear organizers,

Thank you very much for your continued work regarding this important topic of liver tumors. I noticed that for a total of 20 training samples (from volume-28.nii to volume-47.nii), identity affine matrices were given in the nii metadata. For the z-spacing in particular, this does not seem to correspond to the true characteristics of the CT scans.

Would it be possible to update someday these affine matrices? This would facilitate the development of 3D methods dedicated to liver lesion detection and segmentation.

Thank you again, best regards, Alex