Help with Automatic Object Identification and Background Subtraction

Dear Friends,
I am new to CellProfiler and I have a problem with automatic object identification during my analysis. My cells are transfected with GFP, nuclear staining with Alexa-568 and DAPI. I am interested to evaluate Alex-568 and DAPI intensity in GFP+ and GFP-negative cells separately. The transfection efficiency for my cells is very very low. Next, I am trying to subtract the background intensity.

I have tried to make several pipelines. Manual object identification is perfectly working but I don’t understand how to identify the GFP positive objects automatically and how to subtract the background intensity in each channel?! Please help me with this. I have attached my pipelines and one image. Please let me know if you need more information.

Thank you very much in advance. :smiley:

Best Regards,
Puru20200510 new_pipeline.cppipe (16.0 KB) Advanced Pipeline_Puru.cppipe (16.5 KB)

Hi @Puru,

I’m struggling to understand why you think you need to do a background subtraction step? In the example it looks like the staining is pretty obvious. The general approach I’d take for this problem is as follows:

IdentifyPrimaryObjects: Find “Nuclei” from DNA stain.
IdentifySecondaryObjects: Expand “Nuclei” a few pixels using the Distance-N method to generate approximate “Cells”.
MeasureObjectIntensity: Quantify green staining in “Cells”.
FilterObjects: Specify a filter cutoff based on Intensity_MeanIntensity_GFP to classify cells as positive or negative.
ExportToSpreadsheet: Export image and cell sheets. Image sheet should have the count of total and positive cells per image.

If you really need to remove background for some reason, I’d suggest looking at the ImageMath or CorrectIllumination modules.

Hope that helps.

Hi David,
Thank you very much for your suggestion. I have tried to make a new pipeline as per your suggestion but could not manage to fix the problem. I have several experiments where I can see some background staining (may be due to washing problem) for the RFP channel and that is why I was trying to add a background subtraction step. I understand that it could be unnecessary in my case but I just wanted to see if there is any difference before and after background correction. Thank you once again.

That’s fair enough, we may be able to provide further advice if you could show us a “problem” image.

Thank you for your reply. Please find attached some images that I found difficult for analysis. I wanted to automatically detect GFP+ and Negative cells, quantify DAPI intensity in GFP+ and negative cell nuclei, RFP intensity in GFP+, and negative cell nuclei, if possible subtract the background for the RFP channel. Please suggest if we can do it!

Thanking you.

20x 1.3.tif (1.2 MB) A 1.4.tif (1.3 MB) D 1.6.tif (1.9 MB)

It looks like your cells are nicely spaced and there isn’t too much background, so the outline I provided above should work reasonably well. I expect there may just be some settings that need tweaking, do you have a copy of the current pipeline?

Please find the recent pipeline I am using.

Hi, If you see example image D-1.6, there are lots of backgrounds in the RFP channel. I want to quantify RFP only in the cell nuclei. Do you think object intensity measurement in DAPI/Nuclei will help and I don’t have to worry about the background?

When you say “background”, do you mean something like “nonspecific staining” in the form of spots outside of nuclei? Those are quite bright to the point of looking like normal objects. I don’t think background subtraction would be the right approach to remove such objects. If you’re only interested in intensity within nuclei you might be able to get away with just not worrying about them, since you’ll be able to restrict measurement to just within the nuclei in using the MeasureObjectIntensity module.

1 Like

Hi @Puru

I agree with @DStirling, but just to add, if you are very concerned about those small highly intense red objects, you could identify them as primary objects and then mask our of your image prior to the pipeline @DStirling suggests.


1 Like

Thanks for your comment. I will try to fix the pipeline as per your suggestion.