IHC RNA quantification

Hi,

first of all thanks for the great software. I just started to use it, I hope my question won t be too stupid/easy.
I am analyzing images stained in the cytoplasm with 2 different RNA probes (red and blue)…
I would like to classify cells based on this staining… single positive and eventually double positive… even better to classify them based on highly vs slightly positive…
Thanks in advance for the help
Laura

I would guess this is brightfield then, and you have some ability to separate your nuclear stain from the blue ISH stain? If you can isolate the stain vectors, you should be able to use Subcellular detection to get spot counts, then classify however you want.

If you are using both blue for your nuclear detection and your ISH staining, that could be a problem, though. If your image is fluorescent, though, and your cells are somewhat regularly shaped, it should be fairly straightforward. More information/images would be needed to give further help though.

That looks like a decent start, though you will likely run into the same problems as others with one stain being too dark.
You will almost certainly want to look through the Estimate stain vector video above, and try using the small annotation area method to set all three stain vectors while using the Image Type of “Brightfield Other” (Image tab on the left). Brightfield Other will let you handle two stain vectors at once in the Subcellular detection command.

Note that you can use one set of stain vectors for cell detection, swap stain vectors, and then perform the subcellular detection. It is usually easiest to keep track and swap stain vector sets with one line scripts pulled from the Workflow tab.

The problem of determining colour vectors from an image stained with >1 stains was discussed briefly a few days ago.
You need samples of the two or three stains from single stained samples. That is, process a sample for IHC ONLY and another one with the cell stain ONLY, etc. Make a montage of the two or three images and then determine the vectors for each colour. Otherwise you are mixing unknown amount of dyes into the result and the result will not be accurate by an unknown amount.
See “Determining new vectors” here:
http://www.mecourse.com/landinig/software/cdeconv/cdeconv.html
You can use the same procedure to do this in qupath.
Also the colour choice in the image above is not ideal. I would try a cell staining with something that is not in the continuous hues of blue-violet-red. Think in RGB cube terms.