So experiments it is recommende to avoid unintentional 'cherry picking'. So do you analyze fluoresence intensity? If you do then adjusting while imaging is not advised. But if you are just looking to see for fluorescence then it doesn't really matter, you could adjust while you image, and sure not all cells are going to be usually equally fluorescent.

Hello,

With caution, I would say the first option is safer.

Please check the following review (it´s about IHC, not IHF, but many of the statements are valid for both):

Article Quantification of immunohistochemistry - Issues concerning m...

IHC, ICC, IHF and ICF are semi-quantitative techniques, which basically means that you always need to have controls (reference standards) in your experiments. Variability between different experiments is completely normal. This is called intra-observer variability and can be related to many factors such as, biological variability, minor (non-intentional) differences in your staining protocol (fixation, minimal variations in Ab concentration...). Of course, you should do your best to standardize the protocols and to have minimal variations between experiments, but they will always be there.

I have to disagree with one of the above comments stating that you should not adjust parameters when imaging. This is a wide miss-conception in many of the routine immunostainings techniques used in (research, not diagnosis) labs. What you should have is positive/negative controls and always run them in parallel (same buffers, times of incubation, primary and secondary Abs mixes...) during the same experiment. Control vs Treated, Wt vs. KO, etc... This is why the units are "arbitrary fluorescence units"... Unless you are doing a purely quantitative assay in which you can demonstrate a clear-cut known linearity relationship between the amount of staining in certain conditions and the amount of antigen (however, you would need reference standards for this too).

Once you have done this (parallel conditions) you can (and in most of the cases you should) adjust the acquisition settings for different days of experiments. What you shouldn´t do, is modifying this parameters for different conditions belonging to the same experiment. And you should at least have one shared condition in all your experiments (let´s call it control) .Then, in order to pool your data between different days, you should express one of the conditions as relative (normalized) to the other.

E.g.:

-KO will be used as to set your zero signal while imaging to afterwards image the WT and see if your antibody and labelling is good and specific enough.

-"Control" will be used as to normalize the "treatment" condition.

And that´s the reason why you need to run them in parallel, you cannot compare a ICF/IHF performed on a WT one day with a KO performed a different day (no matter how much standardized your protocol is). The same applies for control vs treated.

Offline (a posteriori) treatment of the images is usually advised (background subtraction, brightness correction), but I would try to standardize (so to have the same settings for everything) as much as possible these for two main reasons:

1-To minimize user intervention (save time, and potentially undesired "observer effect")

2-To allow automatization

Thresholds could be a function of your images´intensities (Which is basically how an "Auto threshold" works), in a way such:

Threshold="n times the average background fluorescence+n times the SD of different background measurements" .

However, you shouldn't use different thresholds for different conditions performed the same day of experiment (this would be bias), although you could use different thresholds for different days in the above-indicated manner. If you decide to do this, the background regions used to calculate the Th should include background measurements of every condition of your experiments.

So... long-story-short:

a) I would adjust acquisition conditions between different days but NOT between different conditions, making sure you have proper controls, b) I´d try to standardize the offline analysis and image treatment as much as possible and c) I would always try to express my data as normalized to one of the conditions in my experiment.

Open to discussion however,

Cheers,

J

If you are relying on fluorescence brightness for analysis, you're doing it wrong. Fluorescence brightness is not quantitative. Please see "The 39 Steps: A Cautionary Tale" by James Pawley (Article The 39 Steps: A Cautionary Tale of Quantitative 3-D Fluoresc...

Grab a dark frame from your camera or, even better, from a blank sample that has been treated along with your other samples, and subtract that image from your experimental images. Anything above the background level is now your data.

For what it's worth, room lighting should not affect your fluorescence levels at all. The light from a fluorescence microscope is many thousands of times brighter than room lighting, and is hot enough to burn you should you try to look at the end of your finger (I fooled around and found out). You can't control for the amount of time you've looked at a sample. A good mounting medium, such as Cytoseal 60 or gelvatol/mowiol will help with the fading (and work much better than many other commercial mounting media, even those from, e.g. Molecular Probes, that have anti-fade reagents).

To anyone reading this in the future,

I consulted with a microscopy specialist in our department and he recommended not changing the settings on the imaging step, but rather setting a baseline difference in fluorescence by comparing unstained control groups between replicates, and adjusting the data based on the difference measured.

Thanks for all your responses! I appreciate the assistance.