Whole-gland Prostate annotations during hidden validation and tuning cohort and the hidden testing cohort

Will the annotations of Whole-gland Prostate be provided during hidden validation and tuning cohort and the hidden testing cohort?

Hi Zelin,

Whole-gland prostate segmentations will not be available for any of the cases in the Hidden Validation and Tuning Cohort (100 cases) or the Hidden Testing Cohort (1000 cases). These segmentations were automatically generated by our publicly available AI model and provided for the Public Training and Development Dataset (1500 cases) via pical_labels, to kickstart your model development cycle. If you need them for validation/testing cases (or any other case upon deployment), you can essentially use the same AI algorithm or train your own and then use that in your submitted AI container.

Hope this helps.

Thank you so much.

I need Whole-gland Prostate annotations as one input of my csPCa detection model. So, can I use your trained model for prostate gland extraction directly?

Hi Zelin,

Yup, you can use this particular trained model for generating whole-gland segmentations directly. Please check out its algorithm page to learn how it was trained specifically, and find out where you can access its trained weights and source code. Do note, that AI-derived annotations can be susceptible to errors or faulty segmentations (e.g. whole-gland segmentation for case 11050_1001070 from the Public Training and Development Dataset).

Hope this helps.