Closed zaccharieramzi closed 2 years ago
Hello @zaccharieramzi, thanks for finding this. As far as I can tell this feature is not used in any of the baseline examples, which might be why it was missed.
Would you be willing to write up a little unit test to make sure we don't have a regression for this function in the future? Could be a hand-coded apply_mask
with padding where you know exactly what the output should be.
Sure yes I can do that
In the end I just took the structure of the other apply_mask
unit test and adapter to make it fail on the previous version, and pass on this one.
This fixes the apply mask function for the case of single-coil data.
Currently, when using the fastmri data module, the returned k-space are all 0, because during the padding fixing phase, all mask values are set to 0.
I can provide a minimal reproducible example if needed and create the corresponding issue. This was such a minimal change that I didn't bother for now.