Hi Nicolò,
Sorry for the late response.
This looks like a mask issue, even though your mask inputs are reasonable there could be processing steps changing the mask.
Here are a few things I can think of
1. Can you check if your t1.mask.nii.gz looks reasonable ?
2. Since the program crashed, can you check if BDP created a temporary mask file called epi_mask.nii.gz. BDP deletes these files but may be in this case it stopped before deleting. If you do find such a file, how does it look?
4. Run BDP to save intermediate png files using the flag –png-out or –p. This will save snapshots of the masks prior to this failure. Can you check the output <name>.approx_align.png. I have to warn you that this is a very memory intensive process.