CellProfiler Intensity

Dear Sir/Madam,

I used CellProfiler to quantify the number of Lipid droplets are contained per cell (I used LD540 for lipid-droplet staining and my sample was Brown adipocytes). However, when I compared the intensity of my differentiated adipocytes (that have more and bigger Lipid droplets) to the intensity of the non-differentiated adipocytes (that is mostly Background), I observed that cell profiler gives me the same intensity for both pictures/samples and in the end I get same results of amount of Lipid droplets for both samples. One can clearly see that the differentiated adipocytes have more Lipid droplets and also I openned the same pictures with PhotoShop and it gave me more intensity for the differentiated adipocytes (as expected).

Could you please help me with this and suggest me a way that I can adjust my settings in order to identify the difference between the intensities?

I attach two pictures (differentiated vs. non-differentiated cells) so that you can also see the differences and check if there is anything that can be done.

Thank you very much.

Regards,
Arionas.




The differentiated image that you posted appears a lipid stain, whereas the non-differentiated image appears to a nuclei stain. Can you check the posted images and confirm, and re-upload if necessary?
Thanks!
-Mark

Both of the pictures have been stained with Hoechst (for nuclei) and LD540 (for lipid droplets). But in the differentiated one there are both nuclei and lipid droplets whereas in the non-differentiated there are nuclei and no lipid droplets (or very small lipid droplets or just yellow background). Therefore, the intensity of the differentiated should be more than the intensity of the non-differentiated picture. But when I run the analysis, it is not different.

I will upload 2 more pictures.

Thank you,
Arionas.

Here are the other 2 pictures




Hi Arionas,
Perhaps I am just misunderstanding things, but how are you distinguishing the nuclei from the droplets if you only have one image that has both of them stained, rather than one image per stain? Typically an assay like this would have one image for the nuclei and another image from the same site for the lipids, so you can tell them apart. Maybe you should post the pipeline you have thus far?
-Mark

Dear Mark,

You are absolutely right, it was my mistake. I attach 4 pictures (2 differentiated and 2 non-differentiated), for Lipid Droplet and Nuclei staining each. As I had mentioned in the first message, we would like to see a difference in the intensity of the Lipid droplets between differentiated and non-differentiated cells (since one can see by eye that the two pictures are different in terms of Lipid droplet intensity and also PhotoShop does give a different intensity) but cell profiler gives us the same readouts for both samples. Could you help me with this one? (I think I sent the correct pictures now).

Thank you very much,
Arionas.








Ah, that’s more like it… :smiley:

I’m attaching a pipeline that should get you closer to what you want. Some comments:

  • I’m identifying the cell from the lipid stain as best I can, since there’s no cell stain. But ideally, you should have a separate cell stain channel, e.g, phalloidin, tubulin, etc.
  • Some care is needed to set the lower threshold bound for the negative control image so it doesn’t identify a bunch of false positives. You may need to check this value for other images to confirm it’s OK.

Regards,
-Mark
pipeline.cppipe (12.1 KB)

Thank you very much Mark! It worked well.

Cheers,
Arionas.

Hi again Mark,

So, I ran again my pictures with the pipeline you sent me and for most of them I had again the same problem. That is, I get the same degree of differentiation (same number of Lipid droplets) for non-differentiated and differentiated cells. Apparently, the problem is not in the pipeline but - for some reason - Cell Profiler recognizes the same intensity of Lipid droplets between differentiated and non-differentiated cells (even though non-differentiated have some very small spots that shouldn’t be recognized as Lipid droplets - check attached pictures). Even if we normalize the intensity or not, we get the same results. So, maybe Cell Profiler is doing some kind of re-scaling that we do not want? For example, when we run the same pictures at PhotoShop we get a clear shift of the Histogramm between the two pictures.

If you want to test them, I attachpPictures that gave me the same amount of Lipid droplets (even though one is differentiated and the other is not).

Could some kind of re-programming of Cell Profiler help to solve this?

Thank you,
Arionas.








If the non-differentiated case has small droplets that are being erroneously identified by the 2nd IdentifyPrimary module, then you can adjust the lower diameter bound in that module to exclude them. It’s currently set to 2 pixels, so you’ll need to adjust it to something larger.
-Mark

Hi Mark,

Yes, I understand that. By doing this, we could avoid the small lipid droplets. But still, isn’t there a way that the intensity is being reprogrammed so that we get different histogramms whenever we have differences between lipid droplets? Now it seems that all pictures give the same intensity (even if we normalize or not).

Best,
Arionas.