Fluorescence colocalization of extracellular plaque in QuPath

Hi QuPath community
My supervisor and I are looking into amyloid plaques in Alzheimer’s disease and we have different fluorescent markers to label different parts of a plaque that we then image. Now someone recommended us to use the QuPath pixel classifier to detect these plaques all over the brain. I’ve been annotating and building my classifier for quite a while, and it seems to work rather well. Now we are interested in seeing how much the annotations for the plaque in green and the one in red overlap/colocalize. So far I’ve only been able to get out the simple “size” measurement (e.g. perimeter/area).
Would I have to split the channels (e.g. in ImageJ) and train the classifiers separately or can I just choose each channel in “View” and then annotate?
I’ve been working with 0.2.0-m2

There isn’t much support for the pixel classifier in M2. That only came with M5, and you can find a script to run saved pixel classifiers across a project here.

It might be more straightforward to use the simple thresholding tool which will let you set a manual threshold, if the area you are interested in is based on a single channel. If you run it twice, once for each channel, you could Intersect the two areas, as shown in the link.

If you create a single pixel classifier for the overall objects, you could then calculate colocalization coefficients for each object. I haven’t played with that code in a while, so I’m not sure how well it will work with M7.

If you want to get more information out of an ROI, I would also recommend looking into adding features.