inquiry for the nnunet baseline details.

Dear Challenge Organizer,

Could you please provide more details on how the baseline nnUNet model was configured, especially for the midRT task? Specifically: organizers_nnunet_baseline_midRT (14 Aug. 2024, 0.3557)

Did the model use all the input images and preRT delineations available, or was it limited to only the midRT T2 images?

Thank you for your assistance. Best regards, Jintao Ren

Hi Jintao,

The details for both baseline models were 3dfullres nnunet with default settings (5 fold-CV ensemble, 1000 epochs).

• For the pre-RT model, the input was only the pre-RT images.
• For the mid-RT model, the input was only the mid-RT images (no pre-RT data or delineations used).

We’ll add these details to the Tasks and Evaluation page.

Edit: Also, as previously mentioned, we encourage participants not to place too much emphasis on the absolute performance from the preliminary development phases, given that they only include data from two cases. I will say that the mid-RT baseline model demonstrates markedly better performance on the full test set (though we won't reveal what this is till the end of the challenge).

Hope this is helpful.

Best,

Kareem

Dear Kareem, Thank you for the detailed response; it is very helpful.

Best regards, Jintao