Label unique value count?

Label unique value count?  

  By: melhzy on June 2, 2022, 4:19 a.m.

Hi communicty,

I ran a few label unique value count, but figured most values are 0s, which is different as what I had expected. I should be able to see lots of 0s, 2s, 3s, 4s, and 5s. Please see the details in the below screenshot. Did i mis-use the label data?

 Last edited by: melhzy on Aug. 15, 2023, 12:56 p.m., edited 4 times in total.

Re: Label unique value count?  

  By: anindo on June 2, 2022, 9:07 a.m.

Hi Z. Huang,

Thanks for your query.

As listed in the README.md of the picai_labels repo (screenshot attached below) and the study protocol, out of the 1500 cases shared in the Public Training and Development Dataset, 1075 cases have benign tissue or indolent PCa (i.e. their labels should be empty or full of 0s) and 425 cases have csPCa (i.e. their labels should have lesion blobs of value 2, 3, 4 or 5). Out of these 425 positive cases, only 220 cases carry an annotation derived by a human expert. Remaining 205 positive cases do not have annotations for now (AI-derived annotations (Bosma et al., 2022) for these cases will be provided at a later date). In other words, only 17% (220/1295) of the annotations provided in picai_labels/csPCa_lesion_delineations/human_expert should have csPCa lesion annotations, while the remaining 83% (1075/1295) of annotations should be empty.

Hope this helps.

 Last edited by: anindo on Aug. 15, 2023, 12:56 p.m., edited 1 time in total.

Re: Label unique value count?  

  By: melhzy on June 2, 2022, 3:54 p.m.

Thanks, I was able to get the pixel-level label counts properly.

Here's annotation count for each patient and study:

 Last edited by: melhzy on Aug. 15, 2023, 12:56 p.m., edited 2 times in total.