Evaluation on healthy patients.

Dear organisers,

Thank you for preparing such a large dataset of PET/CT images, and the challenge itself. I am a bit confused about the evaluation of segmentation performance on healthy patients, who have zero masks. According to the "Evaluation criteria" section, such cases are evaluated using the false positive volume and the false negative volume, ignoring the dice score. Correct me if I am wrong, but the false negative volume is always zero for those patients, since the ground truth is zero.

The contradiction follows from the following statements: "In case of test data that do not contain positives (no FDG-avid lesions), only metric 2 and 3 will be used", and "3. Volume of positive connected components in the ground truth ...".

Hi!

you are right that in this case the FN is always 0 and therefore this value has no effect on the final score. In practice, only FP is used.

Thanks Marcel