I recommend the approach included in the following link and attachment.

You will commonly and incorrectly see a simple subtraction of the average background intensity signal from the area of interest signal.

However, incorporating the fluorescence pixel density from area of interest to subtract background is a better approach.

https://sciencetechblog.wordpress.com/2011/05/24/measuring-cell-fluorescence-using-imagej/

Hi Andre Daniel Paredes, I appreciate your response, thank you very much. I see that's a great method to subtract background and have a more realistic fluorescence intensity value. I'm still concerned about normalization because I want to know whether an intensity value from the histone mark present in one slide is really higher or lower than the corresponding intensity from other slide. How to normalize pipetting errors, laser variations, antibody hybridization, etc.? Maybe I can introduce fluorescent beads or I can perform a second immunofluorescence against other epitope in the same slide. That's something to figure out...

If you are one the very prudent and excellent scientists who actually thinks about how to analyze the data before collecting it, then, yes, certainly, adding fluorescent bead standard would be a great option to have a standard value of fluorescence. (I think that people here assumed that this the usual scenario where all the data have been collected without any worry about how to analyze it, and now, in the manuscript writing phase, you need some way to squeeze out the answer to your question from the existing data.) Of course, that is not completely devoid of problems, either, since it would still possible to have differences in staining intensity that is likely a far stronger contributor to differences in intensity than fluctuations of laser illumination for modern lasers. One option that would also correct for that would be protein-coated beads. You might find IgG-coated beads or you could find streptavidin-coated microbeads that you then stain e.g. with fluorescently labelled anti-streptavidin antibody. Since the bead (e.g.) streptavidin coating distribution will remain the same, their stain intensity distribution reports on the variations in staining process, except of the course for the permeabilization step that could be an issue for histone staining. You could have a third colour control by staining something in the nucleus whose quantity should remain the same.

Hey Juha-Matti Alakoskela I really appreciate your answer :) Very nice input and ideas. As a third color in the nucleus I could probably use centromere staining (by using a centromere probe for fish), we have worked that before. I would not expect centromeres change in quantity, do you have another idea?

I am one the rare people in biomedicine who has never worked with anything to do with cancer, so I am a bit unsure what changes in cancer cells, especially if they have a high mitotic rate, and very advanced mutation rate with chromosomal instability, chromosomal deletions and duplications and so forth. So I really cannot guess if there will be significant changes in the quantities of centromere proteins. Of course, most of the time the biologically significant changes in expression are >2 fold, so in principle if it changes by less, it should not matter too much. I was thinking that perhaps you could use some membrane-impermeable DNA dye, that would only gain access with membrane permeabilization, but of course the sufficient permeabilization to allow a dye of

Nice ideas Juha-Matti Alakoskela despite haven‘t worked with cancer before :) Thank you for your input. Thanks to you I have more ideas for considering before doing experiments.