Does my RNAiSH require a different procedure from IHC?

Hi everyone, I’m new to QuPath and I have a question about how to count the number of positive cells on a RNAish slide. I followed the video tutorials available here: https://github.com/qupath/qupath/wiki/Video-tutorials Then counted the positive cells with " Analyze → Cell analysis → Positive cell detection". Is this correct for RNAiSH or is there a better way to do it?

Thank you very much!

Depending on whether your samples are brightfield or IF, and the version of QuPath you have, I would generally recommend using subcellular detection to count spots (see this post on cytoplasmic measurements, though spot counting is the original intent), and then classify based on spot count. The multiplex classifier might help with that, as you can separate by 0, 1-5,6-10, 11+ spots, etc. There are several versions of the classifier floating around, so make sure you find the one that works with your version of QuPath.

Other posts:



Plus other posts if you search for RNAscope or spots (plus QuPath).

Thank you so much for your reply! I’ve looked at the different threads you’ve linked and am convinced to change to subcellular detection. Also, I was wondering if you could help point out if there is anything wrong with the current steps that I’m attempting to RNAiSH processing and how I may update them?

I don’t know what steps you are using… so nothing comes to mind immediately. Almost everything is project and biology dependent.

Sorry for not summarizing the steps before. The images are of prostate biopsies. We first set up the image type as Brightfield, and then manually set the stain vectors Hematoxylin, DAB, and background. Afterwards, we used positive cell detection with Hematoxylin OD and otherwise default parameters. I understand now that subcellular detection is better, but are the other steps suitable for this purpose?
Thank you again!

That sounds about right. A lot can be done with scripting. I usually use Positive Cell Detection for my cell detection in order to eliminate “bad” cells. For example sometimes bits of dirt or something will create a very dark “cell,” and I can make those bad cells Positive using a very high DAB Nuclear threshold. I then auto-delete all Positive cells, and proceed with the subcellular detections on the negative cells.

Not sure if you are using Simple Tissue Detection for your annotations?