Measuring Fluorescence Intensity in Confocal Microscopy images

I have some images taken using a confocal microscope of different cell lines which have undergone immunofluorescence staining using two antibodies for different markers. The green dye shows the marker, red shows a soluble isoform of the marker and the blue is dapi. I wanted to measure the fluorescence intensity of both isoforms of the marker to compare the level of expression between different cell lines as well as between the isoforms. The images are czi files, I was wondering if this could be done using ZEN 2.1 lite? Or is using tiff files in ImageJ preferable? Could you please also let me know what are the steps involved? Should I be taking the original image and then converting it to 8-bit for measuring intensity in imageJ or does it make more sense to split the image into its respective RGB channels to measure intensities of the different markers?
Example original image.tif (8.0 MB)

Hi @clare,

You can use ImageJ and tiff files.

Steps are:

  1. split Channels
  2. get nucleus on Dapi channel (one of the auto-thresholding methods)
  3. measure intensities on red and green channel in nucleus (select the redirect to red/green in Analyze Measurements and then perform Analyze Particle on thresholded Dapi)

Of course, you can do some image enhancement prior to thresholding and measuring (background substraction, filtering…)


That brings up a good point, since you don’t seem to have a fourth marker to determine cell areas, and your expression pattern does seem different in the nucleus vs the cytoplasm. If you aren’t interested in nuclear vs cytoplasmic expression, or even if you are, you could run into trouble with images like those.

Whole image analysis-> impacted by ratio of nuclear space to cytoplasmic space, assuming you mask for the cell vs background.
Nuclear analysis-> Impacted by differing expression patterns between nucleus and cytoplasm, and at high zoom, will be affected by which slice of the cell you choose as it looks like the nucleoli have more red expression.

Hopefully your change in signal is strong enough that you can detect it despite those factors, but it is probably a good idea to decide what exactly you want to measure, whole cell or nuclear expression, and how.

I would generally add the channels together, create a mask that way that removes most of the background, and then reimport the mask and calculate the mean intensity of the masked area. Better to have a cytoplasmic marker, but in your case the sum of both channels might suffice. Generally it is not a good idea to use the marker you are measuring to define the area of… the marker that you are measuring. Thus @VirtualSlide’s method of using the nucleus would give the most accurate results if you are only interested in nuclear expression.

1 Like