Count labeled cells with high intensity


#1

Hi,
I just start using cell profiler and currently I am struggling a bit.
I labeled my cells with two different dyes. However, I have a little bit background of one dye in nearly all of my cells. But some cells are labeled really bright and clear. I want to count the cells, which are labeled with both dyes, but just the ones, which are labeled clearly.
My pipeline right now is:
ColorToGray
IdentifyPrimaryObjects (first dye)
IndentifyPrimaryObjects (sec. dye)
RelateObjects
FilterObjects
ExportToSpreadsheet
Is there a way to count just cells with a special intensity? I tried to adjust the threshold manually, but I always get the same results for the different thresholds. Furthermore, I changed the color balance in ImageJ and I was wondering, if settings I adjust with ImageJ stay in cellprofiler?
Many thanks in advance,
Lea


#2

I think you can add the MeasureImageInnsity then you should be able to filter based on the measurement.


#3

Thanks you! I tried it and now my pipeline looks this way:
ColorToGray
IdentifyPrimaryObjects (first dye)
IndentifyPrimaryObjects (sec. dye)
MeasureImageIntesity
RelateObjects
FilterObjects
ExportToSpreadsheet
It shows me the total intensity and mean intensity etc. In FilterObjects I selected measurements as filtering mode, but I cannot choose Intensity as a filtering method. So I still do not know how it works. Can you help me again?
Many thanks in advance,
Lea


#4

Hi,
I used ClassifyObjects after MeasureImageIntesity now. In the next step, I can filter by the classified Objects. It works, but I am not exactly sure, what I am doing.
Why do I have to select Number/Object_Number by the Category to classify? And what are bins? When I filter Bin-3, it seems that CP selects the cells with the highest intensity. How is a bin defined? Sorry for the stupid questions, but I didn’t understood it completely yet.
Thanks,
Lea


#5

You can take a look at the help for a module by clicking the ? button next to each setting. But you can see even more help if you select the module in the pipeline and click the ? button under the pipeline window. Does that help you understand the bins?


#6

pipeline.cpproj (80.7 KB)
Hi,
Unfortunately, I didn’t understood the comments in the ?. And it still don’t work well. The amount of filtered cells in always: NumberOfCountedLabeledCells- NumberOfBins+1
So for example CP counts 200 cells in the image and the number of bins is 4, than the selected number for Bin4 is 197. When the number of bins is 3, than there are 198 cells in bin 3.
Does somebody know what I am doing wrong?
I attached my pipeline.
Many thanks in advance,
Lea


#7

Hi,

I suggest to try first to see what is the difference between the clearly-stained cells and the noisy ones. To reach this goal, you can make a simple pipeline, in which you measure both the intensity and the texture of that specific channel (“Red” in your current pipeline).
You probably need just 1 image that includes both cell types, segment a few good cells for each type (clear vs. noisy) and measure them. I recommend to simply export the result into spreadsheet and use Excel, for instance, to plot the measured values of the 2 populations of cells.

Once you have an idea what is/are the criteria(s) to differentiate between the 2 cell types then we can go back to your main pipeline, build a better filter to exclude the noise cells.
For example, knowing that the noisy cells will have high integrated intensity AND low sum variance of channel Red, you can just use a single FilterObjects module to do this exclusion, no need for ClassifyObjects

Hope that helps.


#8

Hi Lea,

I think some of your issue was that you were using MeasureImageIntensity rather than MeasureObjectIntensity, so you weren’t actually getting individual measurements of each cell in your channel of interest.

If you go back and fix that, you should be able to use “intensity” in FilterObjects to get the cells you care about.

Good luck!


#9

Hi,
Yes, Thank you it works with MeasureObjectIntesity.
Thank you all for your help!
Best,
Lea