Thank you Jason.

As Jason said, Keep exposure constant so as to capture the less fluorescent cell. Take data from maximum number of cells to as to minimize the bias.

Thank you Parminder.

Try to determine the exposure setting in your most brightly fluorescing sample. You may have to survey your samples before imaging to get an idea of which sample this is. It's always possible to bring up the contrast in an unsaturated sample, but not possible to decrease it in an oversaturated one!

Not trying to sound too blunt about this but state your research problem more clearly. 

If all you care about is capturing the most number of cells in the linear range, then set it accordingly.  If you care about the dynamics, then reducing field size and focusing on a few cells may matter more.  You may have to lead up to such experiments to determine if it's even possible to keep settings fixed without oversaturation, ie., how fast is signal changing, why is it changing, can I determine copy number of insertions, does it matter, how long is the expt., etc...If you care to compare one experiment to the next, then lots of other things in combination can come into play...

To help make a clearer decision, just be rational, which can be defined as being able to reason, and "reasoning is a simulation of the world fleshed out with our knowledge" (Johnson-Laird).  From P and Q -> P, infer Q    (Goebel),

That is, given a final state (P), that is approached from an initial state (Q), infer conditions of the initial state and imagine...

Doing so will help illuminate the values and standards you use when making your decisions.  This works best if you believe that to be scientific is to be curious.   

"When the magician pulls the rabbit from the hat, the spectator can respond either with mystification or with curiosity. He can enjoy the surprise and the wonder of the unexplained (and perhaps inexplicable), or he can search for an explanation.  Suppose curiosity is his main response—that he adopts a scientist's attitude toward the mystery.  What questions should...."

~Simon and Newell, Human Problem Solving

If you have one or two outlier cells that express much higher fluorescence than the others, and you are using a confocal or laser light system on your microscope, you can photobleach the cells of high fluorescence and then take your image capture at the 'normal;' settings that are amenable to the cells that emit lower fluorescence.

Try to avoid the cells giving you higher fluorescence and focus less fluorescent cells before taking pics. If you cannot avoid these then capture as many images as you can and include all of the cells to avoid the bias.

As Jerry mentioned, it really comes down to your biological system the experimental question you're asking with it. Without knowing more I have several suggestions based on my experience with API DV systems. 

Remember that the DeltaVision will scale the displayed image from the brightest pixel.  While that my make labeled cells appear dim of even non-fluorescent, this is only an artifact of the display.  The other labeled cells will have useful intensity values.  You can check this using the data inspector tool. 

I would set the exposure using your brightest sample/cells such that they aren't saturated and then use that during the rest of your acquisition. You can use the exposure finder and provide a desired maximum intensity and the system will determine the optimal exposure for you.  For display after collection you can always stretch the histogram to bring up those dimmer cells. Altering the brightness and contrast only affects the displayed image and not the intensity data you're interested in as long as the gamma remains 1.0.  

One other point with measuring the intensities.  You'll want to calculate background subtracted, integrated-intensity values for each cell, not just an intensity value for a small ROI using SoftWorx data inspector or measurement tools.  While more time consuming, it is the valid and the most reliable measure when comparing intensities.

Thank you Matthew and Scott for the valuable suggestions. Currently I am following the same.