task1 image border question

Dear SynthRAD_2023 Community,

I trust this message finds you well. Firstly, I would like to express my sincere appreciation to the entire team of specialists who diligently contributed to the creation and maintenance of this valuable dataset.

Similar to the previous question, (https://grand-challenge.org/forums/forum/synthrad2023-643/topic/clarification-request-on-task1-challenge-1448/)

https://ibb.co/MkpfyrD

https://ibb.co/9WCVzgw

I think the range of the mask is applied differently in CT and MR

when I calculate the metric, should I calculate the metric within the given mask, or should I compare it to the whole CT image and synthesize CT?

Thank you

Dear dlrkdgns1324,

Thank you for the question. The metrics will be calculated on the mask.

What you observe is the different FOV extension between CT and MRI, to account for it, we tried to consider the intersection, so the smallest surace between CT and MRI. From what I see from the shared images, of the six patients, only for 1PA047 the mask is too big compared to the MRI.

Please note that the impact should be minimal in the metrics, but if this is considered disturbing during training, you could consider excluding a couple of slices in the cranio-caudal direction (specifially the most cranial and most caudal).

Also, be aware that in the test set, we have thoroughlly revised for this effects.

I hope this may help you, but please do not hesitate to reply if something is not yet clear.

Best,

Matteo

Thank you for your answer. I'm looking forward to the release of validation data set the next day