Testing and validation data

[kleingeo]

For either the original Miccai or the subject-based datasets for either 2019 or 2020, was any processing done to the images when running through the public and/or hidden test sets? I have noticed some oddities in some segmentations like odd labelling and random labelled voxels (few voxels with a totally random class in odd parts of the images). Were these kept for testing or was there any type of processing?