Intensity measurement problem with CellProfiler

Pipeline JNO 2.2.cppipe (18.4 KB) NIH3T3 RhoA WT 20X DAPI 13.tif (732.0 KB) NIH3T3 D1 + RhoA WT 20X GFP 13.tif (1.3 MB) NIH3T3 D1 + RhoA WT 20X RFP 13.tif (1.1 MB)

Hi everybody !
It’s my first time with cell profiler and I think I have a problem with something in my pipeline or with my way to analyse…
I want to analyse the localisation of my protein (HAtag) in my cells (nuclei, or cytoplasm) so I used the “mean intensity” to analyse that. But, the Mean intensity of HAtag is stronger in the nucleus than in the cytoplasm, whereas looking at the images we can see that it is clearly the opposite. You can find attached my pipeline, two photos for the example. What do you think about it?
Thank you for your help !!

Hi @Justinenouj,

I took a look at your pipeline and made a quick overlay of your segmentation against the stain you’re measuring:

One thing you can immediately notice is that some cells have much more expression than others, which may have an impact on the averages. More importantly, it looks like the cells which are ‘positive’ for staining still contain large areas with minimal staining, which may suggest an issue with either the stain or the segmentation of the cell compartment. These extra dark areas are probably the reason why the averaged Cell intensity comes out slightly below that of the nuclei.

If you wish to properly compare nuclear and cytoplasmic staining, you may also want to use the IdentifyTertiaryObjects module to create cytoplasm objects, since the Cell objects currently include the nuclei as well.

As another point of interest, you don’t need to run the MeasureObjectIntensity module individually for each image/channel you wish to measure. Selecting multiple objects or images will measure everything at once.

Hope that helps

1 Like

Hi @Justinenouj,

@bcimini and I also took a look at your pipeline today. We noticed a couple of additional points that you may want to consider:

  • For many of the pixels inside the nuclei in your images, the intensity values are approximately the same as the intensity values in the cytoplasm. While they appear darker than the cytoplasm, much of that effect is due to an optical illusion created by the immediately adjacent bright perinuclear intensity.
    – You can get a sense for these intensity values by hovering your cursor over the image in the CP viewer window and looking at the bottom of the window for the "Intensity: " reading.
    – For this reason, there may truly be very little difference between the mean pixel intensity in the cytoplasm and the mean pixel intensity in the nucleus.
  • We also noticed that your nuclear segmentation often includes a few pixels from the bright perinuclear staining. The Workspace viewer shows this nicely - you can see bright green pixels inside the pink nuclear outline:

    – These very bright pixels will increase the mean pixel intensity within the nuclei, especially since the nuclei are relatively small.
    – You may want to use the ErodeObjects module to erode the nuclei by 1-2 pixels in order to avoid including this bright perinuclear stain as part of the nucleus (if you think that’s accurate for your biology)
  • Finally, if you are interested in quantifying the presence of the bright perinuclear pattern, the MeasureObjectIntensityDistribution module may be more useful to you.



Hi @pearl-ryder and @bcimini !
Thank you for your Help !!
My aim is to know the localisation of my protein (HAtag) in my transfected cells. It seems to be around the nucleus but I don’t know how to quantify that. I will try to use the “ErodeObject”!

The MeasureObjectIntensityDistribution is a very good idea but I can’t find a good setting to quantify the presence of the bright perinuclear pattern. As you can see in the picture, it is not very good…

Which kind of settings I have to use ? I used this one :

Thank you a lot for your help !!


Hi DStirling!
Thank you for your answer ! Can you tell me how you did your overlay? I would like to add the shape of nuclei and cells on my images with the intensity distribution but I don’t know how to do? Thank you a lot

Sure, that’s done using the OverlayOutlines module. You may also want to run the RescaleIntensity module on the background image beforehand in order to adjust the brightness for display.

Intensity distribution can be a little more tricky to display nicely. The MeasureObjectIntensityDistribution module itself can produce a heatmap image, otherwise you may want to try the DisplayDataOnImage module.

1 Like