By: zhixiongzh on April 28, 2021, 6:19 a.m.
Hi,
why the counts of 2 raters are so different e.g. in sub-101_space-T1?
By: kvanwijnen on May 11, 2021, 1:31 p.m.
Thank you for your message and great to hear you are participating in our challenge!
Annotating enlarged perivascular spaces (PVS) is very challenging and is quite subjective, different raters have different annotation styles. For this case rater 1 observed many enlarged PVS while rater 2 observed less enlarged PVS. For the elements that rater 1 annotated and rater 2 did not, possibly rater 2 thought that those elements were PVS that were not enlarged enough or that they were other structures. It seems like in this case rater 2 annotated more cautiously than rater 1.
You can check the segmentation masks to see what the raters annotated and where this different in count comes from (e.g. with ITKsnap or FSLeyes). The counts in csvs correspond to the number of connected components in the segmentation masks.
Hope this helps!
By: zhixiongzh on May 16, 2021, 9:57 p.m.
Thank you for helpful reply!
