joeranbosma
commented
2 years ago Hi Christoph,
Thanks for raising this issue! You are completely correct, the annotation for study 1000707 of patient 10691 was resampled to the wrong T2-weighted scan. I have resampled this annotation to the correct scan now, see picai_labels for the update.
Also, I have updated the imaging metadata, to correctly reflect the release is v2.0 and it was released on June 10th.
Kind regards, Joeran
After the update of labeI data (mid June 2022) is there a mismatch between T2W and resampled annotations in number of voxels for the following case?
T2W images/fold0-4/10691/10691_1000707_t2w.mha : center: -37.663136 33.162276 -70.192629 extent: 192.000000 192.000000 75.599997 axis[0]: 1.000000 0.000000 -0.000000 axis[1]: -0.000000 0.993961 -0.109734 axis[2]: -0.000000 0.109734 0.993961 size: 384 384 21 voxelgap: 0.000000 0.000000 0.000000
Resampled annotation human_expert/resampled/10691_1000707.nii.gz : center: -37.563128 33.261668 -70.203641 extent: 192.000008 192.000008 68.399908 axis[0]: 1.000000 -0.000000 -0.000000 axis[1]: -0.000000 0.993961 -0.109734 axis[2]: 0.000000 0.109734 0.993961 size: 640 640 19 voxelgap: 0.000000 0.000000 0.000000
Maybe getting the corresponding correct T2W is a bit misleading because the Zenodo page (https://zenodo.org/record/6624726) still names the updated MRI data as V1.0 from May 5, 2022.
greetings, Christoph