Re-assign class of cells

Dear All

I have detection measurement results from the image attached (capture 1). The cell segment for epithelium, stroma and immune cells are labelled with brown, green and purple colour after detection classification. A subset of active immune cells was later found in a subsequent IHC experiment using the same slide. I have worked out their identities and coordinates in the original IHC experiment. I am wondering if there is a way to re-assign the class of these active immune cells and visualize their location on the original image (capture 2).

Thank you

Capture 1

Capture 2

Annotations tab, select the new class, use the Set Class button below.

Thank you for your reply. This tip has solved 50% of issue.

These active immune cells were identified by self-organising map and clustering analysis using signals detected from multiple IHC experiment. When experiment is on a small scale, I can manually find these cells and re-set the class. It is not plausible when the experiment covers the whole slide with multiple biopsies. Is there an easier way to select these cells based on a list of identity and centroid coordinate?

Thank you

Pretty much anything is possible with a script and a good workflow, but I would be hesitant to try to match exact X-Y coordinates. They almost never line up between experiments.

If you want those values though, I recommend having a look at the Nearest Neighbors script here that makes use of the XY coordinates of all cells.

You might have better luck if you can generate an annotation in one image and use the Interactive Image Alignment tool to transfer the annotation to the new image, then reclassify anything within the transferred object.

Thank you for your reply.

The activation status of immune cells is normally determined by several cell surface markers so the idea of using activation annotation created from a single image (one marker) probably won’t work.

Qupath v0.20-m5 version has a “Select by measurement (experimental)” function under the tab of Objects/Select. I am wondering if it is possible to build a command or script to select cells based on a list of coordinates?

Sure, the coordinates, would just have to match up exactly between images. This is almost never the case, even if it is the same slide and same tissue.
Selecting
Classifying
Manipulating objects

Though if the cells are all exactly aligned, there isn’t really any need to check the XY coordinates, just move all of the cells to the new image (copy and paste the data.qpdata file in the data folders), and add a new set of measurements for the new marker.
Creating features