Posted by
Olivier Burri on
URL: http://imagej.273.s1.nabble.com/Measuring-Fluorescence-density-of-nuclei-tp5008116p5008871.html
Hi Neethu and all,
First a couple of questions:
- What kind of microscope are you using? CONFOCALS are usually a no-go for 'fluorescence intensity' measurements. If all the cells are acquired at the same distance from the coverslip and the same volume is taken, you might have a chance, but I have not yet seen data confirming this.
- Have you kept all imaging and sample preparation parameters identical over each of your 1500 images?
- How were your stainings done?
As a good starter, I'd suggest you have a quick read at:
*Seeing is believing? A beginners' guide to practical pitfalls in image acquisition*
http://jcb.rupress.org/content/172/1/9.fullThe following paragraph should give you some information:
*Quantification of images—why is it useful and when is it appropriate?*
Fluorescence signal is very linear in WIDEFIELD microscopy and correlates well with fluorophore concentration, up to a point which makes it OK for semi-quantitative or relative measurements. If you expect a 5 fold or more difference then it should be OK.
Also remember that DAPI only stains AT and not GC, so depending on your biological question, this could affect any interpretation you would like to draw from your experiments. And with a quick search I have not found any papers claiming that light microscopy yields information on DNA concentration from DAPI stainings. The closest I have found, and am likely to believe, is using FACS.
*Critical Aspects in Analysis of Cellular DNA Content*
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2976661/Also, when working on thresholds to measure intensities, it’s good practice to use a 'neutral' stain to detect the area. Setting a threshold on the intensity to then measure that same intensity biases the results. One stain marks the area independently of the stain from which you would like to extract the intensity information
In the end, fluorescence microscopy like the one you are performing is good for answering "Where is this?", not so much for "How much of this is there?"
Finally, if you still want to go about with this, an idea would be something in the lines of:
1. Blur your image significantly to smooth out uneven DAPI patches.
2. Apply a "Find maxima" filter to get a single point per nucleus.
3. Enlarge these points to get a surface of a fixed size that fits inside the smallest nucleus you can find.
Measure these surfaces (No threshold) on your non-blurred image (You can also measure it in your blurred image). Integrated density and average density should yield the same when you divide the integrated density by the area.
Make sure you have controls and/or at least two conditions to compare! Just absolute differences will mean nothing unless you have something to compare them to. Something like fibroblasts vs. tumor cells where you KNOW you will get a difference. You can thus also assess the natural variation to expect from a given sample.
This was by no means a message to discourage you, but just to inform you of the potential dangers and critiques that using DAPI quantification could bring.
All the best!
Oli
Olivier Burri
Engineer - Image Processing
& Software Development
EPFL - SV - PTECH - PTBIOP
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html