Identification, quantification and outlining of cytoplasm

Good morning and thank you for putting to everyone’s reach such a great software!

I have gotten a lot of very helpful tips from previous blog entries but I still have a couple of pending questions. I am trying to quantify cytoplasmic vs. nuclear staining and am attaching sample pictures as well as the project I have designed.

a) The lymphocytes I am analysing tend to have shapes that are not always circular and I have found that using propagation as means to identify secondary objects yields better results then Distance - B (that most times disregards any cytoplasm that is not circular). Are there any caveats to this approach?

b) In the attached example, the cytoplasmic outline in blue sometimes fills up the whole cytoplasm (for e.g. rectangular looking cell in the bottom left), and in other instances the outline does not fill the cytoplasm but just outlines it. It is really important for me to be able to accurately quantify the cytoplasmic intensity. Will such 2 cells be evaluated differently by the software? Which outlining should I be aiming for? Would you suggest doing anything else differently?

Many thanks in advance for your support!
p

IFNg.cpproj (951.1 KB)

Glad we’re able to help!

A)I don’t think so- propagation is generally what we recommend as a starting point. As long as you’re happy with the results, then it’s probably the right method for you!
B)It looks like in the ones where it’s just a hollow outline it’s because your nucleus is actually too big- take a look at your IdentifyPrimaryObjects output and make sure the bright nuclei aren’t getting a “halo” of empty space around them.

Good luck!

Thank you for your reply!

I assume from your response that the optimal outline to accurately quantify the cytoplasm is when the cytoplasm is completely filled in blue (such as cells A, E and F) and not simply outlined (such as cells B, C and D). Is that correct?

Indeed, it seems like the brighter nuclei are preventing their cytoplasm to be filled with blue (such as cells B, C and D). I tried modifying some of the settings of IdentifyPrimaryObjects to no avail. What do you suggest I do to prevent the empty space ‘halo’?

Many thanks!

Here is the pictures with the labeled nuclei:

The navy is showing what’s being called your cell “area” so yes, you want it to be filling your cytoplasm for your purposes. To get a tighter threshold on your nuclei I’d try adjusting the threshold correction factor in IdentifyPrimary or switch to a different thresholding method (adaptive perhaps) or algorithm (maybe Background or Robust Background). Good luck!

Thank you! I’ll try it out!

Good evening!
I tried modifying the parameters for the IdentifyPrimaryObjects but it didn’t improve the thoroughness with which the cytoplasm is detected. I also noticed 2 other things:

  • the output of the IdentifyPrimaryObjects is pretty accurate and does not create a halo of empty space around the nucleus
  • the areas that are not labeled as cytoplasm are the ones that differ slightly in intensity from other areas in that same cytoplasm (either brighter or dimmer) (for example, in the pictures attached above i- notice the discontinuously labeled cytoplasm of the cells that is above left of A and how that co-incides with a weaker staining there and, ii- how the B cells have none of their bright cytoplasm labeled).
    Any clues how can I improve this?
    Many thanks!

Aha! I realized the issue- that blue was NOT in fact a fill in of your cytoplasm- I couldn’t tell just from the screen shot. What’s causing it is that the edges of your secondary objects are being segmented very noisily; try adding a smoothing of ~3-5 pixels in the IdentifySecondary objects module (right now you have smoothing off) and it should fix your issue.

Worked like a charm! Thanks a bunch!