Measuring Cell Fluorescence using ImageJ

pSer ImageJ setup

Image J can be downloaded for free from here .
This guide can also be downloaded as a complete PDF here: Measuring Cell Fluorescence using ImageJ

Here is a very simple guide for determining the level of  fluorescence in a given region (e.g nucleus)

  1. Select the cell of interest using any of the drawing/selection tools (i.e. rectangle, circle, polygon or freeform)
  2. From the Analyze menu select “set measurements”. Make sure you have AREA, INTEGRATED DENSITY and MEAN GRAY VALUE selected (the rest can be ignored).
  3. Now select “Measure” from the analyze menu or hit cmd+m (apple). You should now see a popup box with a stack of values for that first cell.
  4. Now go and select a region next to your cell that has no fluroence, this will be your background.
    NB: the size is not important. If you want to be super accurate here take 3+ selections from around the cell.
  5. Repeat this step for the other cells in the field of view that you want to measure.
  6. Once you have finished, select all the data in the Results window, and copy (cmd+c) and paste (cmd+v) into a new excel worksheet (or similar program)
  7. Use this formula to calculate the corrected total cell fluorescence (CTCF).
    NB: You can use excel to perform this calculation for you.CTCF = Integrated Density – (Area of selected cell  X Mean fluorescence of background readings)

  8. Make a graph and your done. Notice that in this example that the rounded up mitotic cell appears to have a much higher level of staining, but this is actually due to its smaller size, which concentrates the staining in a smaller space. So if you just used the raw integrated density you would have data suggesting that the flattened cell has less staining then the rounded up one, when in reality they have a similar level of fluorescence.

How to Cite this if you wold like to:

I used this method in these papers:

McCloy, R. A., Rogers, S., Caldon, C. E., Lorca, T., Castro, A., and Burgess, A. (2014) Partial inhibition of Cdk1 in G 2 phase overrides the SAC and decouples mitotic events. Cell Cycle 13, 1400–1412 [Link]

Burgess A, Vigneron S, Brioudes E, Labbé J-C, Lorca T & Castro A (2010) Loss of human Greatwall results in G2 arrest and multiple mitotic defects due to deregulation of the cyclin B-Cdc2/PP2A balance. Proc Natl Acad Sci USA 107: 12564–12569

But you can also find a similar method published here:

Gavet O & Pines J (2010) Progressive activation of CyclinB1-Cdk1 coordinates entry to mitosis. Dev Cell 18: 533-543

And here:

Potapova TA, Sivakumar S, Flynn JN, Li R & Gorbsky GJ (2011) Mitotic progression becomes irreversible in prometaphase and collapses when Wee1 and Cdc25 are inhibited. Mol Biol Cell 22: 1191–1206

And my apologies to any others that I have not mentioned.


Tags: , , , ,

181 responses to “Measuring Cell Fluorescence using ImageJ”

  1. Winnie says :

    Hi, I find your post very relevant for my project. However, my cell nuclei were stained by DAPI, and my protein-of-interest was stained with Alexa Fluor 488. I am trying to quantify the amount of this protein within the nucleus. Could you give me some advice on this? Thanks!

    • Alice says :

      Hi Andrew,

      thank you very much for this fantastic post, it was really useful.
      I am just wondering why I am getting negative values of CTCF and if so how I should interpret them.
      Hope you can help.


      • Ariel says :

        I had the same problem even though my background is really low and my fluorescence of my treated cells is low relative to the vehicle treated cells, it’s still A LOT more fluorescent than the background.

  2. ScienceTechBlog says :

    Hi, this method works with any flurophore, be that DAPI, 488, 555, FITC, Texas Red etc etc. If you want to only measure the nucleus of a cell, then when you draw your region, just limit it to the nucleus. This method is good for gathering data on individual cells, and I normally find that a minimum of 25 is needed to get any meaningful stats (but more cells is always better). If your more interested in the whole population, then you might be better off using a threshold method which could give you the average intensity across the whole field very quickly. Just make sure what ever you do you use identical acquisition settings for every sample. Overall, quantifying IF images is perhaps a little subjective and open to user input, but it can still be a useful tool. I hope that this is of some help. If you still need further clarification let me know.

    • Winnie says :

      But in my case, should I separate the RGB stack into individual fields, and then measure the green and blue channel staining intensity separately? With blue channel of DAPI alone staining as background, and then use the formula as shown by you?
      Thanks for the help, I am a complete beginner in this!

      • ScienceTechBlog says :

        Hi, you need to measure each channel separately. The background measurement should be taken from an empty region on the same channel, just adjacent to the sample of interest, in your case the nucleus. You can, then also measure the DAPI staining (in the same way) if you like as well, but I’m not sure if this is really necessary. If your only interested in determining if your protein (488 channel) increases/decreasesthen you only need to measure that channel. So in summary yes you need to separate the channels, and measure them independently! Good luck

    • Clara says :

      Hi, in case you are interested in the whole cell population, how exactly would you do the threshold method? Thanks!

    • Gagandeep Singh says :

      Hi Science TechBlog, Thanks for your adding the protocol.
      I am interested in measuring the intensity of staining across the whole image (breast tissue).
      How do I account for background readings in this case?

      Kindly advise.

      Thanks in anticipation.


  3. Katherine Philips says :

    Is there a way to use ImageJ to calculate the mean fluorescence in a given field? Or will it only work for single cells?

    • ScienceTechBlog says :

      You could use this same method but instead of selecting a single cell, you could select the whole field using for e.g. the rectangle tool.

      • Jen S. says :

        Hello. I tried to do just that. I just selected a rectangle around the whole image and then a smaller rectangle around a region with no fluorecence.
        I set the scale in the image prior to doing these measurements.
        Here is the output that I got.

        1 2028497.343 73.115 148313083.953 34890653
        2 4476.089 3.965 17747.078 4175

        Where the first column is the area, the second column is the mean, third column is intden and the last column is rawintden.
        I followed your instructions, copied and pasted into Excel, used the equation you listed and am getting very high values. Maybe that is correct?? (Here is the equations I looked at: Intden – (area2 x mean1) = CTCF

        The units for CTCF would be area (e.g. um^2) correct?

        On another note, I also have confocal stacks (single dye) that I would like to look at the intensity of. Essentially I would like to compare the intensity of different stacks for a (hopefully) a simple comparison of what is ‘lighting up’ in the image. Hopefully that is clear. Not sure if you have any suggestions on this or not.

  4. Bhatia says :

    how doyou change your RGB picture to 16bit pixels only picture?

  5. Tamara says :

    This is so useful! Thank you very much!

  6. Dani says :

    Hi I am a complete beginner at using imageJ. Would I be able to use this method for comparing the intensity of alexa flurophore 488 inside the nucleus with the intensity at the periphery of the cell in order to see if the protein had moved to a different location in the cell? If so how would I go about doing this?

    Any help would be appreciated. Thanks 🙂

    • ScienceTechBlog says :

      Hi Dani, yes you should be able to use this method to do what you want… just include an extra step for the cell periphery. my suggestion is to try drawing around the total cell and then the nuclear fraction, take reading for both, correct each for background as per formulae and then divide one by the other to create a nuclear:cytoplasmic ratio, which over say 20+ samples you hopefully will be able to get some statistical relevance.
      Good luck

    • saeid says :

      Hi Dani,

      Have you done your measurment? I am a beginner with imageJ and have almost the same situation, could you please help me by telling the way you have used to do your measurments.

      Thanks a lot,

  7. Kelly says :

    Hi Andrew, I wanted to ask if you have ever used size exclusion when doing your analysis in ImageJ. If you have, how do you set the scale for your maximum size?

  8. Cedric says :

    Hi Andrew,

    I was using Imaris before to quantify my confocal images and beside the value of the fluorescence of a defined area, it was giving me also the size of the area, is it possible to evaluate that in Image J? (by giving him a scale in um before of course).
    Thanks in advance

    • ScienceTechBlog says :

      Yes you should be able to do this in imageJ. First set the scale of the image by going to the menu item Image>Properties.
      Then go set the type of things you want measured by going to the menu Analyze>Set Measurements. Here you can select a whole heap of different parameters to measure area being one of them !
      Good luck

  9. Cathy Chang says :

    this is what I searching for, thanks for sharing!

    Greeting from Taiwan


  10. Marli says :

    Hi Andrew. Thank you so much for your helpful post. I am a beginner with ImageJ, so I was hoping you could answer a question for me. My cells are stained with Bodipy (specific to lipids) and DAPI for the nucleus. I’m trying to measure the average fluorescence of the lipid droplets in the cytoplasm. How do I subtract the fluorescence emitted by the DAPI? Thanks!

    • ScienceTechBlog says :

      You should do the quantification on the image before you compile all the channels together. Most microscopes should acquire each channel as a separate image. If you have a good microscope then it should cut out most of the bleed thru, also I find acquiring the Dapi last helpful, as the blue light can often bleach the other channels a bit.
      Hope this helps

      • Marli says :

        Thanks! That does help a lot! I have the green fluorescence images, so I should be able to do that. Do I need to convert the image to an 8-bit black and white prior to starting the measurements? Also, if all the cells are different sizes, do I have to subtract the area of the nucleus since I’m only interested in what’s in the cytoplasm?

      • ScienceTechBlog says :

        Not sure it has been along time since I came across a camera that took colour pics, most are greyscale. It should be fairly simple to test if conversion makes a difference to the reading but my guess is that it won’t. As for the nucleus subtraction that is really up to what it is exactly that you want to measure. I would start with the simplest setup and see if the data corresponds to what you see visually. Then add complexity if needed to improve the readout quality if needed.

    • Lam Nguyen says :

      Dear Marli,

      I had the situation here, i also want to count the lipid droplets, did you get through with that? Please give me your suggestion, thank you

      Lam Nguyen

  11. Mate says :

    Hi Andrew,
    I found your article very useful, thanks a lot for posting on this topic! I’m also a beginner in ImageJ.. I would like to calculate nulcear/cytoplasmic fluorescence intensity ratios on my cells. I suppose, I’ll use a simple circle tool, and put down circles in the cytoplasm and in the nucleus in case of many cells and than calculate the average. This might be very simple, but I don’t see how I could set the size of the circle tool to be constant. Otherwise I need to create circles by hand all the time, and if the size is not the same, then the whole thing will get far away from being representative..
    Thank you so much for the answer!

    • ScienceTechBlog says :

      I have not yet figured out how to set a constant size, but there are two work arounds. First you can move the circle that you set up on the first image around the first image and take all your readings. Next copy and paste that circle onto the next image, then hit the delete (backspace) and it should just delete the pasted image but leave the circle, so you can then repeat the same process.
      This is ok for a few images but its easy to accidentally delete or loss the circle tool.
      Thus I think you maybe better off, making the readings independent of size by using the formulae in the guide to correct for both background and size of the area that you select. This way can draw any size circle, or even perhaps select the whole cell, and then then subtract the nucleus. I suggest having a bit of play to see which method/s give you the best/most accurate looking data.
      Good luck

      • Mojca says :

        You can achieve a constant size of your circle by running the ImageJ Macro “Circle Tool”. Then you can specify a set/constant circle size (20 pixels or whatever) and every time you click anywhere in your picture, your circle size will always be the same.

  12. Jon says :

    Hey, great article. Just had a quick question, I am comparing fluorescence between different samples and using the entire field. So the area fr each sample is the same. When I go to calculate my corrected total cell fluorescence, i get some negative values. Is there a way I can adjust for this?

    • ScienceTechBlog says :

      Hi, this method is specifically designed for single cell analysis, so applying it to a whole field probably won’t work and thus the negative values. If you just want a whole field then you are better off using a threshold over the whole field as this will be faster and probably more accurate… but won’t give you individual cell info.
      Good luck

  13. Marcella says :

    I am a beginner in imageJ and i have a question. Am I able to use this to quantify the fluorescence in a given field? I am not interested in individual cell data. Will this method be accurate and if not do you have any suggestions on what I could do?

    • ScienceTechBlog says :

      This method is specifically targeted at single cell analysis, and is probably not the best method for whole field quantification. I would look into the use of thresholds to do what you want to do.
      Unfortunately I am not able to help you out with that method as I don’t know how to do it…sorry but good luck

      • Gagandeep Singh says :

        Hi, I am also interested in quantifying the fluorescence in the whole field. I have a protocol that uses threshold function. But I am wondering if I would need to correct the background reading in this case as well?
        Any ideas?



  14. Kanya says :

    Hi Andrew, thanks for your post. It’s very useful. In my case I want to compare the intensity of my interested protein which is localized at the septum between wild type and mutant strain. The area of selected is very small approximate 12-15. Therefore, should I keep the same size of area in each cells ? There is much difference between multiply by 12 or 15. Or do you have any suggestion?
    Thanks in advance

  15. priyankar says :

    hi, I am a new user of image j. I want to analyse the fluorescence intensity of two cells. but while following your procedure, I can not find the “set measurement ” option under analyse menu. please help

  16. priyankar says :

    hi, i have another question. can this procedure be applied for calculating intensity from two separate images

  17. Gabriela says :

    I want to measure fluorescent intensity in tissue. Do I measure small areas of the tissue and then calculate the CTCF for each one of them? The values I am getting are double what I would expect.


  18. kiran says :

    I was wondering if you change the LUT levels of the images, does the integrated density or the mean fluorescence intensity changes? Or are the LUTs just to enhance the contrast in the image

    • ScienceTechBlog says :

      I normally do not change or alter the images at all. This is the best way to ensure that you don’t introduce any artefacts into your measurements. Its also critical that you acquire all your images with identical exposure times etc.
      However, i’m pretty sure that altering the LUT levels will not have an effect on measurement readings. You can probably try this for yourself, take a reading pre and post and see if it changes things.

  19. kiran says :

    Thank you for the reply. I was also wondering if I want to compare multiple images for fluorescence intensities, should the area of the cells be the same?

  20. allison says :

    Thanks for the informative article. I’m trying to use imageJ to quantify intensity of primary cilia in two different channels. In this case, i would like to keep the area I have selected constant between the two channels while measuring intensity of each color separately. Do you have any idea on how to do this? Thanks!

    • ScienceTechBlog says :

      Hi you can copy and paste an area between two images. You just have to hit the back space key ‘once’ when you paste into the new image. I have not figured out a better way and if you have a lot of images then it can be very annoying when you make a mistake. Thats why I developed this method to account for differences in area size between to images.

  21. Lucy says :

    I, like everyone else, am new to Image J. My situation is a bit different than others. I have cells that i am co-localizing microtubules and virus. I have individual images from the two different channels, however I want to get the intensity of the areas where they are merged so I need to have both channels together. Is there any way I can only quantify the yellow areas.

    • ScienceTechBlog says :

      Hi Lucy,
      Unfortunately this method is not suitable for what you want to do. You need to look at co-localization. There are plenty of plugins for ImageJ that should help you do this. A quick google pulls up several. Also many other microscopy software programs now have co-localization built in.
      good luck

  22. Bobin says :

    HI, Can this method be used to compare the fluorescence intensity to find the ratio between YFP and CFP and to calculate FRET with single cell images taken through blue channel and yellow channel separately.

    • ScienceTechBlog says :

      It can take a look at the following papers for the method.
      Gavet, O., & Pines, J. (2010). Activation of cyclin B1-Cdk1 synchronizes events in the nucleus and the cytoplasm at mitosis. The Journal of cell biology, 189(2), 247–259. doi:10.1083/jcb.200909144
      Gavet, O., & Pines, J. (2010). Progressive activation of CyclinB1-Cdk1 coordinates entry to mitosis. Developmental cell, 18(4), 533–543. doi:10.1016/j.devcel.2010.02.013

      Although there are several FRET specific plugins [Link] for ImageJ which maybe worth a look at as this could simply the process.

  23. Diego Bucci says :

    Hi, I found your method very interesting. I’m trying to measure different fluorescence from tyrosine phosphorylated cells. My only concern is that when I apply your formula for adjusted intensity, it comes out a negative number… my areas are usually around 4000 and the back intensity around 5000.. Any idea?

    • ScienceTechBlog says :

      That is very strange. Not quite sure what is going on there, without seeing the images, I would guess that perhaps your images are either too over or under-exposed resulting in a high background or under-estimation of your reading of interest respectively.

  24. Schahin says :

    This was so helpful. I have quantified the intensity of my cells. But, I was wondering what will be the unit, as the digits I got as a result are so high. Thank you.

    • ScienceTechBlog says :

      The units are arbitrary, and don’t have a physical meaning. Its all just relative 🙂
      Thus I label the graphs as “Total Cell Fluorescence (Arbitrary Units)”

      • Schahin says :

        Thanks for a quick reply. I also want to measure intensity of different concentration of proteins. What should I use as a background, also should I measure background each time? the proteins I am measuring are micro arrayed as a spot on the slide.


      • ScienceTechBlog says :

        The background reading should be an area close to your sample of interest. It should be taken and matched for each image and fluorescence filter. For micro arrayed dots, this method may not be the most efficient. There maybe more automated solutions out there ?

  25. Somayeh says :

    your post is really helpful. for measurement in the box for image J I have IntDen and also another parameter RaWIntDen, what is that?

  26. Andy says :

    Hi, your post is really helpful. Here I have some questions regarding to cell area meaurment by using ImageJ. My cells are stained by phalloidin and DAPI, and I want to meausre the area of these cells and then comparing the morphology differences between my two samples. I am currently using ‘threshold method’, however, I have find that finding the region of interesting (ROI) through adjusting the ‘threshold value’ is a littile subjective. I am wondering what is the most proper way to measure the cell area in ImageJ? and Do I need to keep the thoreshold value constant during the meaurement in order to make the results comparable between my two samples? Thanks in advance Andy

  27. Rahul says :

    CTCF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings)

    In the formula above, what does ” Mean fluorescence of background readings ” mean? Is it the mean value of the background image or the integral density of background image. Little more clarity on this would be really helpful to me.


    • ScienceTechBlog says :

      Normally you should take several background readings for each image, so the mean of the background readings is the mean of those.

      • Rahul says :

        so, i should take mean of integral densities for background images ?


      • Rahul says :

        do we take the average of the integrated density of background reading or the average of the mean gray values of background readings


      • ScienceTechBlog says :

        You use the mean for background not IntDen, as you will multiply the average mean by the area of the cell to give the Background- IntDen of the cell. If you used the original IntDen reading this would be for the small circle area you selected, which is not what your wanting. Hope that helps

  28. Kitty says :

    Hi, this was really helpful. I just quantified some data using this. I was wondering what the difference between integrated density and CTCF is? I have been unable to find definitions for each. I used a premade excel spread sheet that calculated this number for me.

    • ScienceTechBlog says :

      Integrated Density – Calculates and displays two values: “IntDen” (the product of Area and Mean Gray Value) and “RawIntDen” (the sum of the values of the pixels in the image or selection). “RawIntDen” is only available in ImageJ 1.44c or later. “IntDen” and “RawIntDen” values are the same for uncalibrated image.

      CTCF = Integrated density corrected against the background.
      Hope this helps

  29. Javier says :

    Hi, thanks for your very helpful post. I have one doubt though. I am measuring fluorescence intensity in nuclei to compare ratios of two different proxies. I have z-stacks of the image. In order to measure the CTCF I do a Sum Slices projection. Woud it be better to do Average Intensity or Maximum Intensity projections instead.

  30. twinkle says :

    hello, imageJ is certainly a great solution to many imaging processing techniques. I have a question, I am using a dye which accumulates in healthly mitochondria and on introducing drug the dye releases out of the mitochondria. can I use fluorescence quantification for control image against treated image? if so , how? I would be really thankful for this help.

  31. NweIke says :

    Hi Andrew,
    Thanks so much for this helpful post.
    Please, I would like to ask if I can use the fluorescence intDen measurement to quantify the amount of a drug molecule (ofloxacin) in tissues.
    Want to determine the absorption route of this drug molecule using sliced animal tissue.

    Thanks for your help.

  32. Ash says :

    Hi Andrew,
    Your blog is really helpful. I too need some help with imageJ to quantify fluorescence of my dye on tissue samples. I have grey scale fluorescence mosaic of my stained tissue from confocal microscope. I need to quantify the amount of dye present in the samples. Could you provide some help?
    Thank you

    • ScienceTechBlog says :

      Hi, sorry for the delay, been super busy. From the sounds of it, you maybe better off using a threshold approach. The method I posted here is really only for individual cells from numerous independent fields. If your not concerned with individual cell quantification, and just want a total field quantification then thresholding will be much faster.

  33. François says :

    I succed to use your clear method to quantify GFP of a 2D image. Thanks for This. But I would like to quantify it in a 3D image of a cell which is perfectly spheric. I have roughly 50 to 120 stacks for a cell. Should I do the quantification stack by stack or is there any other method to do this in a volume on the 3D project of my cell ?
    And I would like to know if with a 3D image of a confocal laser microscope, imageJ integrate density concidering that the image is made of pixels or does it concider that the image is made of voxels ? because in my case, a volume cannot be made of pixels, and so I cannot sum the quantification of each stack. Thanks a lot for your help.

    • ScienceTechBlog says :

      Great question, and I wish I had a good answer. What I have done in the past and I have seen other publish this also, is to just take the best single plane/slice and quantify that. You could possibly flatten the stack (max projection) but I am not sure what algorithms are used by ImageJ to do this, and if it would be consistent enough to allow quantification. You could also try this, or switch to a better (but costly) 3D program like Imaris. Sorry I cannot be of more help

      • François Barbier says :

        Thanks a lot for your answer. It is approcimately what I thank. I’ve found a lab where people use imaris. I think I will try it. It is probably possible with imagJ or fiji, but as you said, the result given by this software is not sure, and for a publication, it would turn into a problem. bye. François.

  34. François says :

    Sorry, in my precedent message I wanted to say slice and not stack. thanks

  35. Daniel says :

    Hello, I’m working with FISH images, in green fluorescence of bacteriais like E.coli (populations). This method is very nice, but i think that the threshold is more easier. I don’t know how to do it! Could you give me some advice?


  36. Blaise says :

    Hello ! Thanks a lot for the great guide you made !

    But I’ve a little problem with imageJ. In my case, I have images with 2 differents colors and I want to determine the fluorescence intensity of a specific area for both color. The problem is that I’m not able to keep the same area between my 2 stacks…so how could I copy the area on my green stack for example and paste it to my red stack (so that the side and the position of the area are the same) ?
    Sorry for my bad level in english, hope it was “understanble” enough.

    Thanks !

  37. rupert grint says :

    Have you ever thought about including a little bit more than just your articles?
    I mean, what you say is valuable and everything. But imagine if you added some great
    graphics or video clips to give your posts more, “pop”!
    Your content is excellent but with pics and videos, this site could
    certainly be one of the best in its field. Fantastic blog!

  38. Sreeja C Sekhar says :

    Very nice and informative blog.How to measure co-localization in cells using Image J

  39. Rahul Pal says :

    How do I calculate nuclear-cytoplasmic ratio in ImageJ?

  40. Feng Li says :

    I read your blog, it is very helpful for my study. in my case, I wish to get density from whole section. accroding to your instruction, I am confused about your formula:
    CTCF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings)

    in my case, I get the mean intergrated density 2666304.833. in addition, I get background density 57145.16667. area of selected tisuse is 30406. it means that CTCF=2666304.833-30406X57145.16667?
    or 2666304.833-57145.16667?


  41. Yu says :

    Hi, thanks for doing this for us beginners. I have a problem when using ImageJ. It is that I want to meature the fluorescense in more than thousands of pictures. So is there any methods for me to do it easier, instead of doing it one by one?
    Thank you so much!

  42. silky2010 says :

    hi Andrew,

    in the results window, is there a way to rename each entry? this helps me keep things organized.

  43. footka says :


    I am investigating the proliferation of my cells and am using Ki-67 marker and DAPI. By looking at my cells, Ki-67 is presented in forms of ‘dots’ in the nucleus. My question is can i still use your approach? or (to be more accurate) I need to count the fraction of cells stained with the marker? (how to do that?) Another question is about the negative control cells. I understand you have used the background analysis (to calculate the CTCF) but do I skip this when I have the negative control sample? I think, I could circle the cell/s of interest (measure their area) and subtract it from the cell/s in negative control sample to get a better result. Would you say this is a correct procedure?

    BTW, very informative blog

    Thank you in advance for help

  44. ScienceTechBlog says :

    Hi, I am new to imagej and appreciate any help. The circle tool no longer seems to be included in the download and I can’t find it anywhere on the website of imagej. Can someone please email the macro to me or tell me where I can download it?

  45. Bill Tang says :

    Hi there,

    Would you please advise how to quantify fluorescence in nuclei and cytoplasm separately?



  46. Devi says :

    Hi Andrew,

    Thanks for sharing this blog. I was wondering that in this formula “.CTCF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings), why couldn’t we multiply with average Integrated density of the 4 background readings instead. I am trying to understand the logic behind this formula. Please share ur thoughts. Thanks,


    • ScienceTechBlog says :

      Hi Devi, the background reading is to be applied over the size of your region of interest (i.e. cell) thus the integrated density of the background is not what you want to apply to this area. Think of it this way the IntDen = Area X Mean for your cell. You then want to subtract the background for the exact same area of your cell, therefore you just substitute in the mean(background) into the IntDen formula so that you end up essentially with 2 IntDen values one for the cell and one for the background. The CTCF is really just the IntDen(cell)- IntDen(background).
      Hope that helps

  47. Priscilla says :

    Hi Andrew,
    I applied a dye to an epidermal strip and I wanted to observe the fluorescence. I when I oppened the file with the image (the one that i got from the confocal microscope and lsm file) I saw I got three images (all of them are 8-bit) one in one i coud see the the green staining (it has written in the upper part of the image 1/3 (CH2)), the other image I could see some red spots (it was written in the upper part 2/3 (CH3)) and the last image was grey (3/3(ChD)). My question is which of these images should i use to measure the fluorescence or shoud I do something else?.


  48. vivek singh says :

    hi, I want to measure intensity of each bacterial cell individually in a particular image. i m doing image-adjust-threshold to make boundary around each cell, but then instead of getting intensity of each cell diiferently, i am getting intensity of all the cells as 255. i dont know where is the problem.

  49. Frank says :

    HI, I’m attempting to use your method to quantify an effluxed volume of dye on a strip of filter, having created a normal curved based on known pipetted volumes and concentrations on filter paper strips. Was able to get a R=0.94 for the normal curve. Does this seem like a reasonable approach? Has something like this been done before? Do you know any references i could use for this method?

  50. Jenny Baker says :

    Hi, this is great however I have some concerns about my results. Using your calculations my data has given me a goldilocks graph i.e. it’s exactly what I’m looking for. I’m always suspicious of perfect when it comes to cell work. Can I just double check something with you?

    I have several surfaces on which I am culturing cells and some of them are better preferred so I end up with a divide between small-medium cells on some surfaces and medium-large on others. The images were all taken at the same exposure level and my phosphomyosin staining is bright/dim in both populations. Being a complete dunce at times when it comes to math, does your calculation correct for two cells that have the same level of brightness but one is twice as big as the other? I don’t want to run the risk that the reason I have a large CTCF in my “large” cell population is simply because there is more fluorescence area wise in the picture and a small CTCF in my “small” cell population because there is less area wise.


    • ScienceTechBlog says :

      Yeap, you divide by the area of the cell, so this corrects for size !

      • Jenny Baker says :

        Brilliant. “Perfect” data it is!

      • Didi says :

        Your blog is really helpful. I too need some help with imageJ to quantify fluorescence of my single neurons. I have GFP drosophila neurons from confocal microscope. I need to quantify the intensity and lenght of GFP in the each axons in each samples. Then I have to compare with control experiment. Please help me,can I use this method for my article? Could you provide some help?
        Thank you a lot.

      • ScienceTechBlog says :

        Should be able to. See reply to romanr above.

  51. mrcc says :

    Hi, I’ve been trying to understand the difference between IntDen and RawIntDen in ImageJ. My understanding is that IntDen should be the product of area and mean intensity in that area. RawIntDen should be the total intensity in an area. Logically, these should be the same, but in my samples, (with a minimum threshold set but no other transformations done), IntDen / RawIntDen = 0.0077. I cannot figure this out! Ideas?? Thanks so much …

  52. romanr says :

    thank u for all the help here! I have never used ImageJ before but I would like to use it on my project as well. I have stained RBP in kidney slices with green fluorescence. Afterwards I have scanned all the stained slices and took pictures. Now i would like to measure the mean Intensity of RBP (green Fluorescence) in the proximal tubuli and compare it between th different slices. Is it possible to measrue the intensity of a few tubuli and compare them?

    • ScienceTechBlog says :

      In theory yes, this method is best used for application where you need to know the specific fluorescence in a defined area e.g. a specific cell compared to another specific area.
      As its a manual process its quite time consuming, so it really is only useful when you have as small sample size (less than 100). If you are using it on larger datasets then you might need to consider a different method.

      • Khaled says :

        Your post is really helpful, I am measuring fluorescence intensity in kidney tubule cells.
        In my project, I want to measure the intensity of cells from different image and I want to compare 3 kinds of tubule cells in fish kidney. In other hand, we can’t separate the cells in kidney tubule to intensity measuring and should be measured in kidney tubule that made a pipe like shape. The manner of immunofluorescence staining of this cells show a basal staining and central region of kidney not show fluorescence feature.
        I have 2 main questions:
        First: Is it possible to measure the intensity of this tubule shape?
        Second: Can I also measured the middle region of tubule (don’t immunostained) together immunostained region?

      • ScienceTechBlog says :

        You can use the free-form tool to draw around any shape you like, so in theory with this you can measure what ever shape you want.
        Good luck

      • Khaled says :

        Dear Andrew,
        Thanks for your attention
        the Graphic icon is available in plugin menu. this icon show RGB profile plot (and intensity) graph for immunofluorescence image.
        Could you provide some help to use this plugin and graph and table list description ?

  53. Devi says :

    Hi Andrew,

    I have been using ImageJ for quantifying my images using the CTCF method and it is working fine. Thanks for all your help.

    Now am expanding my expts and its becoming laborious to do ImageJ analysis on independent images. So, using Metamorph (microscopy analysis software) with our core facility’s help am writing an algorithm to automate this image analysis. This algorithm independently goes through each image and measures the total cell containing area, and based on that area calculates the ‘Threshold total Integrated density of that particular cell containing area’ and total no. of cells/nuclei. In order to get the average Int. density/cell I divide the ‘Threshold total Integrated density of cell containing area’/Total # of nuclei in that same field.

    Now, when I treat my cells with certain drugs and oxidizing agents it also increases the autoflouresence which could also be a contributing to Int. density numbers. So, I subtract the average autofluorescence Int density/cell calculated by similar logic as I stated above. I wish I could attach a snapshot of my excel sheet then you could have seen the calculations.

    My questions are the following:

    1) Is this method seems logical to you?
    2) I think I need to incorporate the background intensity from each field in this calculation as well, how should I incorporate that in these calculations? I have the ‘Threshold cell area’ and if I get ‘average bkgd Int. density/cell’; should I multiply these two factors and subtract it from ‘Threshold total Integrated density’…. l am confused a bit for this part of calculation?
    2) Is there any way we could automate the whole process in ImageJ itself?

    Sorry for the lengthy post.

  54. Anne says :

    Hi Andrew,

    Thanks so much for the tutorial it works great! I’m trying to distinguish between cytoplasmic and membrane florescence – do you have any suggestions on how to separate the two?

  55. Gonzalo says :

    Hi Andrew, I was using your way to quantify mean cell fluorescence in fixed sample treated with a fluorescent ligand, and I am very happy with it; it is simple and reliable, thanks a lot! Now I would like to ask if you know any simple way to quantify the amount of vesicles within the cell? I have to reanalyze my samples to if the reduced ligand uptake is concomitant with a reduction in total endocytic events (amount of labelled vesicles). It will be great if you may help me on this. Thanks in advance.

  56. Ural Bikkul says :


    I am trying to measure the signal intensity of my immunostained 50 nuclear just on the nuclear periphery. Is there a option to do that? Much appreciated for your help.


  57. Ian says :

    Hi Andrew,

    I hope this is a quick question; I really like this method, though may not be using it correctly. I’m imaging fluorescence in intact plant roots (so not fields of cells, but one root/image) treated with a hormone or not. So I traced areas and took a background measure for each image. And then I calculated the CTCF as in the protocol and found a difference. i’m worried I’m overestimating that difference since the untreated roots have a bigger area than the treated ones do. Based on the formula, hugely different areas would bias the equation. i’m wondering if you think a ‘correction factor’ by taking the mean area for each treatment and calculating the multiple between them, then multiplying the treated sample CTCF value by that multiple to at least eliminate some of the bias due to area. Or is this seeming like I’m bending your method too far/beyond what it’s useful for? Thanks in advance.

  58. Regina Phalange says :

    Hi Andrew,

    Thanks a lot for your post. It is very helpful.

    Do you know what is the difference between IntDen and RawIntDen? Which one is more appropriate for measuring total fluorescence intensity?

    Thanks in advance,

    • ScienceTechBlog says :

      As far as I can tell they are exactly the same.
      As long as your consistent with which one you use I don’t think it really matters
      Sorry I can’t be of more help.

      • Regina Phalange says :

        Thanks for your quick reply, Andrew.
        I have been using RawIntDen to calculate the total fluorescence intensity of my cells, but I realised the values for RawIntDen and IntDen are completely different. Whereas the RawIntDen is around million (range of 0.5 to 3), the IntDen is around 200-800. And I couldn’t find any relationship between the values. Any thoughts?

      • ScienceTechBlog says :

        Sorry my values are always the same for both, so not sure why you are getting different ones.

    • mrcc says :


      I asked this question upthread a few months ago, and at some point I figured it out. As far as I can tell, the difference is the unit of measurement used to determine the area. I decided that for IntDen it is um but for RawIntDen it is pixels. The values you get should be related to each other by the conversion factor between um*2 and pixel*2 – at least mine were. For my purposes, this meant that I could use either column, and I would agree that either is ok, as long as you’re consistent.

      Hope this helps,

    • Mohanad says :

      IF the image dimension was changed then the Int.den will be different from the Raw int.den, but if you using the same original image then int.den and raw int.den will be the same

  59. Regina Phalange says :

    I am using the sum of 20 slices from confocal micrographs as input. Could it be that the reason for getting different values for RawIntDen and IntDen?
    Thanks a lot!

    • ScienceTechBlog says :

      Sounds like it.
      For quantification you should really use the raw files from the microscope. For this I find it much faster to take a lower magnification image. Low mag lens generally have a greater depth of field meaning that you don’t need to take slices. If this is not possible, then you should really only quantify on the strongest slice from each image

      • Regina Phalange says :

        Thanks very much for your comment, Andrew. I am using the raw files for quantification. I am exporting the slices to a multi-tiff file and them combing them into a single image (per channel) using ImageJ (Image -> Stacks -> Z project -> Sum of slices). My concern now is whether I should use the RawIntDen or the IntDen for the analysis since I do want to consider the fluorescence intensity from all the slices.
        Thanks a lot for your help and congratulations on your blog. It is really great!

  60. says :

    Wonderful article! That is the kind of information that are meant to be shared across the web.
    Shame on the search engines for no longer positioning this post higher!
    Come on over and discuss with my site . Thanks =)

  61. web hosting provider cnet says :

    Wow, this article is nice, my sister is analyzing these things, therefore I am going to
    inform her.

  62. Rob Simpson says :

    Hi there I was wondering if you could help me out, sorry for the long winded question.

    I’ve been using this method to measure the intensity of a protein in the nucleus of cels, on the red wavelength. I wasn’t initially going to analyse my images in this way so I cannot guarantee this wavelength’s settings, i.e brightness will have remained the same as other people had been using the microscope while I took a large amount of images over a period of a few days.

    I understand if i was to analyse the pictures using the threshold option on imagej my results would be inaccurate.

    But am I right in thinking that taking a background reading in using this method I’ll still be able to obtain accurate results supposing the brightness settings had changed on the microscope?

    Thanks a lot for this article, it’s been a big help to me so far.

  63. Queeny says :

    Hi, Thank you very much for this method, it is extremely helpful for my analyses. However, my colleague thinks that I am wrong using this method as there is a gradient of background across the image that the formula does not take into account. He is also concerned that because the pictures are not thresholded, the data I have got is not valid. As you have used this method so many times, have you come across some issues with background gradient? I personally think this is a great method as I measure the intensities by tracing the neurons. Please help!!! I am new to this and a little bit confused. Thank you very much again

  64. Runjun says :

    hi, thanku u very much. Found very useful method and i have analyzed my fluorescence images. But i am confuse in one step: i have got fluorescence intensity of the whole image considering different locations as well as considering background signal. But, i have not divided the image into channels. will it be correct? i will be very grateful if u get idea of this.

  65. Lydia Wu says :

    Hi, I’m trying to find the intensity of fluorescence in GFP tagged neuron cells in worms. I used the formula and got backgrounds from around the worm but get a negative number. If i get a couple backgrounds from a dark/non-fluorescing area in the worm, i don’t get a negative number. Am I supposed to get a background of the black areas IN the worm AND outside/around the worm? Why do I get a negative CTCF when I use use background areas outside of the worm? If you could get back to me about this quickly, that would be great! Thanks for your time!

    • ScienceTechBlog says :

      Not sure why your getting negative numbers, I have never worked with worms before.
      I normally try and take areas close to the cells that I’m measuring, so I guess this would be most similar to the dark non-fluro areas from inside the worm.
      So I would use these.
      I always like to compare the results with what I visually see on the image to make sure that the values correspond to what is observed.
      Good luck

  66. Saeid says :


    I have applied a dye to plant root cells for a specific time. The dye is incorported in the membrane and gets inside of the cell through endocytosis. I want to compare the rate of endocytosis by taking the ratio of the intensity of the dye inside of the cell to the membrane in wild type and mutant plants.
    I would appreciate it if you could please help me with this.


  67. Anoek says :


    First, thank you for the great article!
    I have a question. The CTCF formula you use, should make sure that background noise is gone. But when I use different sizes to measure the same nucleus, I get different CTCFs. Do you have a solution for this?

    For example: (N=nucleus=measured cell, B=background)
    Nucleus 1
    N 206524 124.852 25784909 25784909
    B 206524 7.258 1499012 1499012
    B 206524 7.755 1601585 1601585
    B 206524 8.060 1664644 1664644
    This gives a CTCF of 24196532.92

    Nucleus 2
    N 417416 79.921 33360468 33360468
    B 417416 7.071 2951421 2951421
    B 417416 7.407 3091955 3091955
    B 417416 8.588 3584728 3584728
    This gives a CTCF of 30151095.51

    This shouldn’t happen, right?
    Do you have a solution for this?


    • Anoek says :

      Sorry, I didn’t make it clear in my post, the 2 measurements are from the same nucleus (so no nucleus 1 & 2, but both nucleus 1)

    • celldivisionlab says :

      The size of area you select does have an effect on the readings. If you take a larger area around the nucleus (even if you can’t see anything), there is still likely a reading, which will alter the values slightly. I personally prefer to take a slightly larger area on all cells to minimise this. Its also a good idea to count lots of cells to help reduce the effects of this variability and ensure that your final numbers are close to correct.
      ps This method is really best suited to situations were you really need to assess specific individual cells in a population. For example, different stages of mitosis. If you’re only interested in the nucleus and every cell is the ‘same’ then use the threshold method as it is much faster and more accurate.

  68. Madelin says :

    Hi Andrew,
    Please give me ur advice
    I tried to use your method, and I have some questions
    in most of my experiment the integrated density equals to raw integrated density??
    Except one or two experiment?? raw integrated density is 6 digit number like (666666), while integrated density is 2 digit numbers (13)
    and I dont know what does these mean?

  69. abhay says :

    How to get IOD value of a whole confocal image. Please reply.

    • celldivisionlab says :

      This method is not suited to 3D images. It is really only useful for a single plan. You could possibly do a maxprojection of your stack, but this is not advised, as it can alter the intensities and give misleading results. If you want to do analysis on 3D volumes, you will need to look at more advanced options, if you can get your hands on a copy of Imaris, that would be the best option.

  70. Francisco Roberto Quiroz says :

    Hello Andrew. Your method to quantify fluorescence using ImageJ is very uselful to me. I would like use your references “Burgess et al (2010) Loss of human Greatwall results in G2 arrest and multiple mitotic defects due to deregulation of the cyclin B-Cdc2/PP2A balance. PNAS 107: 12564–12569” in my next paper. However I could not find the technique description there. Please would you help me?…Sincerely FQF (

    • celldivisionlab says :

      Hi sorry, unfortunately due to space constraints we did not include a full method in the PNAS paper.
      However, you can now quote our recent paper in Cell Cycle, which does list the full method.
      Hope that helps
      McCloy RA, Rogers S, Caldon CE, Lorca T, Castro A, Burgess A. (2014) Partial inhibition of Cdk1 in G2 phase overrides the SAC and decouples mitotic events. Cell Cycle. 2014 Mar 6;13(9):0–12.

  71. Neda says :

    Thank you so much, very useful 🙂

  72. Hanis says :

    hi. i took confocal images of the zebrafish head, to see if my drugs (tagged with rhodamine) cross the blood vessels and enter the tissues. however, I am stuck on how I should analyse the image. Should I substract the image at time:2hour – time: 10minutes? but I’m not sure if the result really represent what is happening. Your suggestion is highly appreciated. Thanks!!!!

  73. Rennie says :

    Thanks Andrew, this is useful. I wish to measure intensity of immunohistochemical staining. Does it work in a similar way? I tried using your method, and found that the integrated density is lower in ROI but higher in surrounding unstained region. Can you advise why would this happen? Thank you so much!

  74. Yasmin says :

    Hi Andrew,
    I’m taking images of several live sea anemones, lying down, using Z-slices.
    I’m wondering what is the best way/equation to analyze my results, since I’m assuming that the thickness of different anemones will affect the readings.

    I’m a beginer in ImageJ, I must add.


    • Yasmin says :

      I thought to stack all the slices as sum of intensities, and take measurements of ROI’s. But then divide the IntDen to its 3D area (since I know number of slices and their thickness) and then use the rest of your equation. Will that be a correct thing to do?


  75. koko says :

    I was wondering how did you go about measuring droplet spreading versus time using image j. Analysis of spreading versus time is imperative for my research work. Your help will be much appreciated.

    • ScienceTechBlog says :

      I’m guessing you can measure the size (diameter) of the droplet at each frame, which will then give you enough data to determine spreading speed. You could also you one of the track particles plugins and use the leading edge of the droplet as your track point.

  76. Redqueen says :


    I’m quite new at fluorescence imaging and would like some advice. I want to stain my cells for CYFIP1 and Phalloidin and determine if there is any difference in Phalloidin fluorescence in CYFIP1 mutant cells compared to control cells, in a quantifiable way if possible. Is this method stated, or the threshold method, an appropriate way to do this?


  77. Lindsay says :


    Could you help me understand why there is a need to include the area instead of just using the mean grey value from the cell minus mean grey value from background? Does the mean measurement not take the area into account anyway?

    Thank you!

    • ScienceTechBlog says :

      The size of the cell affects the integrated density, therefore you have to correct for the background intensity for the size of the cell or ROI that you are interested in. Otherwise your values will be slightly skewed by cell size artefacts.

  78. Gokhan says :

    How can I cite your method? I want to use this formula in my manuscript?


  79. sharmili says :

    hi, I am completely new in this image J software. I need to analyze my particles with this software. I have two types of picture one is blank which doesnot have any clotting of particles and another picture i have a clotting of my particles . So i have to compare with these two pictures. how do i calculate that one. i dont need only area.

    • sharmili says :

      I need another information, how to convert the particle size into a mean grey value. It will be very helpful for my research

      Thanks !!!

  80. MSBA says :

    I have been using this approach to quantify number of protein signals. I would like to ask you if your method account to different nuclei sizes.

  81. Deepak says :

    Hi, I am researcher and looking for method to measure the florescence along the cell length of a bacterium. above described methods works when you select an area or region of interest, but in case I need to find what is the fluorescence across a line drawn on an image? could we do it with ImageJ?

  82. Mohana says :

    Thank you for this intersting subject.. specially i am working now on this field…my question is how can analysis multiple object without selecting it one by one (that i am working with sf9 cell and i have like 100 cell per pic)?

    • celldivisionlab says :

      Use a threshold method.

      • iansuffi says :

        Can anyone lead me to a good threshold method protocol using image J? I have been seeing this being mentioned in this comment thread but I can’t find any good guide on how to do this anywhere! (The only link someone linked here that could potentially help me was the quantification guide link from the Genevan University but the link is dead (404) to me!)

      • Mohanad says :

        Thanks for answer, i already used the threshold, but many object are disappear and i dont get the two peaks 4n and 2n in histogram

  83. Myshnee Naicker says :

    Hi, I am looking for a method to analyse my data using Analysis 5 software. I am looking at antibody staining on oesophageal carcinoma using bright field microscopy. the method I am currently doing is converting my image to grayscale image and trying to determining intensity on my regions of interest. . the values that are generated are in the range of 50- 100 which represent high intensity. My problem is that I need an algorithm to represent my data in my thesis or a method to invert the data so it represents a larger value to report. Could anybody with experience with Analysis 5 or interpretation of my data assist me?

  84. Tushar says :

    Hi Andrew,

    Your posted methods & replies are very useful & easy to understand (specially for the beginner like me), sincere thanks for that.

    I would like to have your suggestions (e.g. brief modification of your protocol) for measuring GFP expression in cell-membrane only (in HEK cells). Could you please help me.

    Thank you,


  85. Matt says :

    The cells I am imaging are not perfectly circle (they have projections), and when I select my ROI with the square tool, I end up having quite a bit of background in the ROI as a result of the projections. the background within the ROI varies with each cell I choose. I was just wondering if this is accounted for when calculating CTCF, or if it even makes a difference.


  86. Pete says :

    Does it matter if you use IntDen or RawIntDen?

  87. John Griffin says :

    Hi, Andrew,

    Although your procedure is well-written and cites other papers, this type of quantification of fluorescence is not valid. The brightness of a region in an image does not have a direct relationship with the level of protein expressed or fluorophore adsorbed. Fluorescence levels in an image can be affected by too many factors for such quantification to work. Please see James Pawley’s excellent paper “The 39 Steps”:

    • celldivisionlab says :

      Hi John,
      Thanks for the link, and yes you are correct this method has many pit-falls, is not ideal, and it is very important to understand the limitations of this technique. However, if great care is taken during acquisition and appropriate controls are used, (including the validation using other methods), then many of the pit-falls can be minimised and reproducible data can be generated. We have validated this with several well-established biological responses such as the degradation of cyclin B and securin during mitotic exit, and been able to get results that match other methods, including western blot, PLA and FRET. Thus if you antibodies are good, your technique, a willingness to acquire images at the same time under as close too humanly possible identical conditions, and to count a minimum of 100 cells, then it is possible to get this to work well.
      All the best

  88. Felix says :


    thanks a lot for this nice protocol, very usefull!

    However, I have a couple of questions which could not be answered by the web so far.

    My application is to measure the fluorescence intensity on IF stained tissue, in a certain area around a vessel. In that case we might call your method “Corrected total area fluorescence” (CTAF)

    To have reliable results:

    – do i need to keep always the same magnification for each image?
    – when selecting the ROI in ImageJ, do I need to exclude the black (background) area inside the vessel, which has variable size alsways?

    thank you so much for your reply and help

    kind regards

    • celldivisionlab says :

      It’s best to keep everything the same as much as possible to minimise artefacts and intersample variations. For the ROI issue suggest picking one method either with or without and sticking with that. If the quantification matches what you see visually then your probably doing it right.

      • Felix says :

        Thanks a lot for the reply!

        Ok than i keep it as much standarized as I can.

        Its just that I do not really understand how the formular works.

        Does it automatically exclude all the background from positive staining? In that case it does not matter how much background is in the picture.

        But how does the software know what to consider as background, since there is no threshold set?

  89. Alicia Costa says :

    Hi There,

    I wonder whether I can use the same approach to quantify the intensity of innervation of several nuclei. I have camera lucida images, so the background is paper white and the staining are black lines. The size of the areas changes with the nuclei and also the same nuclei have different areas in different anterio-posterior regions.
    What would be the best way to quantify “grade of innervation” per each nucleus?

    Thank you very much!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: