Output format

Hello, I am not sure about the output format. Is the output channel 31 or 4? And should we output 'mha' prediction files?

For the CT task, the output should be a single-channel label map with label index ranging from 0 to 30, saved as an mha file, exactly same as the ground-truth files in the training set. Some of the labels ought to be left empty since there are not as much as 3*10 fragment instances in each case.

For the X-ray task, the output should be a 31-channel mask that is binary encoded to a single tiff file. This is different from the CT task because it allows overlap between labels.

https://pengwin.grand-challenge.org/guidelines/