Multiplexed analysis and measuring areas


I am using the tutorial for multiplexed analysis with my samples and it works very well. My issue is when I try to add tissue segmentation ( Tumor vs stroma) in my WSI using pixel classification, it also works great, however my cell clasification is deleted. I think i am not resolving the hierarchy…My idea would be cell segmentation using stardist, tissue segmentation ( only 3 areas in the WSI: tumor vs stroma and ignore) and cell classification, in order to calculate immune infiltration in the tumor and percentage of immune phenotypes in the stroma

Do you know how I can resolve this hierarchy?

Many thanks for your help

If you are dividing your tissue into tumor and stroma, run that first, and then run the cell detection within those regions. Do not run cell detection within the tissue detection. Possibly, don’t run the tissue detection at all if it is not needed (since you should be able to run your tumor/stroma detection on the whole image).

Otherwise, you can break the hierarchy as shown by @petebankhead in a recent post.

If you do that, the cells will not be part of the tumor or stroma annotations, but those annotations should still contain some information about the cells “below” them.

Many thanks!

I wasnt creating the appropiate geometry annotations with tumor and stroma in the first step, so my “parents” values were wrong.

Thanks again amazing community
Learning a lot

1 Like