[Bug] Polygon resolution after cellpose leads to overlapping and misshapen cells

[pakiessling]

Hi Quentin,
we talked a little bit about this in Munich.
I think this might be more noticeble in our samples as we are working with big cells:

You can see the lines where I assume the shapes from different runs were stitched.
Even more apparent with my trained cellpose model:

I find the overlaps especially concerning as the transcripts / intensities will be counted twice.

I am also experimenting with porting the polygon resolution from VPT:

In this case it leads to the removal of many cells, but there are no more overlaps.

I have uploaded the sdata and my code here:
https://github.com/pakiessling/sopa_poly_experiment

Let me know what you think. I would be happy to try and help in improving this

[quentinblampey]

Hi @pakiessling,

Thanks for sending these detailed images. It seems the segmentation here is pretty difficult (is it DAPI staining?), so I think the results are less consistent than expected between the different patches, leading to these artifacts.

What would be a good solution for you?
I wanted to work on a new resolve method to avoid these overlaps (split the overlapping region between the two cells). Would that be something interesting here?

I'm planning to add a new resolve_method argument, which would be a string to decide on the resolve method to use (with, by default, the one that is currently used).