Wrong PTV margins in provided contours for planning-only challenge

There are issues with the definitions of the different PTV contours (74Gy, 71Gy, 60Gy) in the provided RTSS for the planning-only challenge, if I understand correctly they should be nested and here it is not the case, the 60Gy is too small around the prostate compared to the 71Gy. My guess is that the 60Gy was defined with 5mm margins and 10mm posterior instead of 10mm in all directions.

The contours were all created from a consensus of participants contours submitted to the main AUTO-RTP challenge. Therefore, the error does not come from applying margins directly, but from the application of margins by various participants. Some participants only used a PTVp_6000 for planning, and did not provide a structure for the inner PTVs - consequently their structures will have influenced the PTVp_6000 and not the other PTVs leaving to the overlaps. One participant submitted the PTVp differences as the PTVp regions i.e. the PTVp_7100 was actually the PTVp_7400-PTVp_7100. This also resulted in the PTVp_6000 not being as anteriorly as it should be.

The question is what to do about it. I could take the PTVp_7400 consensus (or appropriate for the other structures) and apply my own margins. Or, I could choose the "correct" participant contours and form a consensus from only then. This would however change the input contours for this challenge. I'm open to suggestions as to what the participants would like.

Don't you have manual delineations for these patients that you used for the evaluation of the initial challenge? We would prefer to do the planning-only challenge on those manual golden standard contours.

The manual contours for these patients are the hidden ground truth contours for the always open challenge, so cannot be released for this challenge.