Release of human annotations for remaining cases

Hello!

Some of the human annotations are missing for a few of the aggressive PCa lesion cases. Are there plans to release these? Or AI-inferred lesions for ~1500 instances as is the case for the ~400 aggressive PCa lesions?

Thank you

EDIT: it would also be helpful to know how the annotations for the testing and validation were derived. Apologies if this is mentioned elswhere

Hi José,

We will not release human-expert annotations for the remaining cases. This is intentional, because it is practically infeasible to annotate all lesions at the scale of the Private Training Dataset. Hence, we encourage participants to develop methods that can account for or figure out how to use non-annotated cases in the Public Training and Development Dataset as well. At RUMC, we deal with non-annotated training cases with a semi-supervised learning strategy (Bosma et al., 2022). We have released AI-derived csPCa lesion annotations for all 1500 cases in the Public Training and Development Dataset (picai_labels/csPCa_lesion_delineations/AI), using this method. You can choose to use these AI-derived annotations for non-annotated training cases or use your own methodology for the same. Please see https://pi-cai.grand-challenge.org/DATA/ for more details.

For the Private Training Dataset, we will also provide these AI-derived annotations.

More information on how the annotations were derived can be found in our Study Design, Item 23: Annotation characteristics. If you have further questions regarding the annotations or another aspect of the challenge, please let us know.

Hope this helps, Joeran

Dear Joeran,

Thank you kindly for your response and the promptness :-) everything is much clearer now.

José