Empty stripes after skeleton analysis with skan

[EdwinHernandezG]

Hi @jni and @GenevieveBuckley

After analyzing the attached big skeleton using skan, the labeled skeleton showed empty stripes without any data.

skeleton_postprocessed_4.zip

The pipeline to analyze it is here:

skeleton_analysis.zip

Thanks in advance for your help

[GenevieveBuckley]

Thanks Edwin!

It was really fun to see what you're working on the other day, btw. It makes me happy to know people do use the code I write

[jni]

thanks @EdwinHernandezG! I'll play around with this soon. After stewing on it for a bit, I think probably some labels were not translated for some reason: map_array will default to 0 if a key is missing. So maybe my idea of how __array__ generates the label image is incorrect.

[GenevieveBuckley]

Were we definitely using int64 for everything?

[jni]

Interesting question. I don't think it should be relevant, since the missing pixels were in the middle of the volume (and thus probably under 2e9?), but we should absolutely check.