DAB detection and its intensity measurement in cytoplasm using Qupath

Hello everyone,

I have stained my tissue sections using IHC and the stains are Hematoxylin and DAB. I try to detect the positivity (DAB) and measure the intensity of the area. But in my case, the positivity is mainly in the cytoplasm as the protein I am studying has mitochondrial origin. So when I use positive cell detection it gives me a percentage that does not make sense.
I’m wondering how I can detect DAB positivity correctly and measure the Intensity of the positive area. If you think there’s also other software than Qupath that I can benefit from better, please let me know. If you guide me through this I will appreciate it.

Hi @sadaffazeli1995,

Welcome to the forum.

In my opinion QuPath is a great tool for what you are doing based on what I know about it, especially if your images are whole slide.

For one, you can set the positive cell detection to be looking at the cytoplasm only (example shown below). Do you feel you are getting good cell segmentation as this is dependent on that? It might be useful to show what parameters you used and/or provide an example image if you are struggling.


Re: “intensity” measurements of DAB, I’m afraid that there isn’t a quantitative relationship between the darkness of DAB staining and the amount of antigen (the expression of your protein) so it’s not accurate to report the “intensity”. A threshold has to be set in order to say it is present or not (“yes” or “no”) but that’s all really. If you search ‘DAB intensity’ on this forum you will find a lot of discussions and more explanation about this topic.


Thanks a lot for your reply. I’m doing my master’s thesis project currently and I’m working on expression of one protein in different cancers and I first discussed with the pathologist. He graded each slice based on the DAB intensity from 1 to 4 plus positivity with naked eye. So the idea is to measure the intensity to define the intensity range for each grade and make it more quantitative. That’s why I need to measure it in the way that it corresponds to the protein expression. After your comment I got the idea why IHC technique is not good for measuring the intensity. Do you any idea how can I make it quantitative or what’s the best way to analyse my sections in this case? I’, using tissue microarrays and one slide contains 77 different cancer types.

Thanks in advance,

Also I’m wondering if the intensity does not correspond to the expression, then how come the pathologist graded them by eye? he did that based on the intensity of the brown color.

Visual grading is not quantitative measuring.
This both procedures are not directly comparable.

Despite all weakness and non-linearity’s of the human visual system, of the visual perception and of the neural processing trained human observers are (most time but not always) astonishingly good in coarse grading by including additional information (cell morphology, spatial signal characteristics, hue-saturation-brightness signal information, characteristics of illumination, variations due to influences of preparation …) beside simply judging the signal intensity …
… if they are used to the task.

But even then they are not measuring something.
They evaluate something and usually categorize it into a few distinct classes, such as
1 (“Thumb up”)
2 (“I don’t know”)
3 (“Thumb down”)

And, if the TMAs are human samples, I want to emphasize that all of those things can come into play even with fluorescent intensities, which are generally considered “more quantitative.” Except, when the genetics of your samples are not the same, even IF is no longer quantitative, as the antibodies will have different binding properties, and in the case of polyclonal, some might not bind at all in a given patient since their epitope will not be present. This problem isn’t limited to IF, or brightfield, in these cases, but to how antibodies work in a complex environment.