Previous discussion of how to specify unsigned 32-byte integer dtype for masks here: #207
We have collaborators that are struggling with enforcing this env command when running steinbock either through an HPC cluster or through basic CLI. Additionally, for increasingly large ROIs, as well as updated 5 micron tissues modes for IMC, we are observing more instances where there are many more cells that can be accommodated in 16 byte integers (~65000).
Perhaps the default mask output dtype could be changed to unsigned 32 byte integer? I understand that for past compatibility with ImageJ, unsigned 32 byte integers were not used, but with new imaging software was well as our own visualisation application, support 32 byte masks has become the standard (and is the default is our IMC viewer for the reasons stated above).
Previous discussion of how to specify unsigned 32-byte integer dtype for masks here: #207
We have collaborators that are struggling with enforcing this env command when running steinbock either through an HPC cluster or through basic CLI. Additionally, for increasingly large ROIs, as well as updated 5 micron tissues modes for IMC, we are observing more instances where there are many more cells that can be accommodated in 16 byte integers (~65000).
Perhaps the default mask output dtype could be changed to unsigned 32 byte integer? I understand that for past compatibility with ImageJ, unsigned 32 byte integers were not used, but with new imaging software was well as our own visualisation application, support 32 byte masks has become the standard (and is the default is our IMC viewer for the reasons stated above).