H&E Staining of Mouse Ear Tissue - Need Help Identifying/Quantifying Neutrophils

I just found out about CellProfiler today after a rather fruitless literature search for automated methods for discriminatory cell identification and quantification. At present, I’m looking for a pipeline that will allow me to accurately identify the neutrophils (should have polymorphic, multi-lobed nuclei) in my samples. I’ve included a representative image here.

Thanks in advance for your help.

Hi, I’d like to help. To start, here is a pipeline that will separate your colored image by the stains you used, which will help isolate the structures you want to identify. Did you use haematoxylin and eosin?

neutrophil.cppipe (3.6 KB)

It would be very helpful if you would annotate your image to highlight and outline your objects of interest. Would you please point out the neutrophils and other objects around them?

1 Like

Thanks, @karhohs.

Yes, I used hematoxylin and eosin. I haven’t been able to clearly distinguish neutrophils in this tissue, as I have little experience with histopathology. I just know that the neutrophils should have polymorphic nuclei, like so:

I was hoping there’s a feature within CellProfiler that will allow me to locate the polymorphic nuclei and thus the neutrophils.

Hmmm, I’m having trouble making out anything that looks like the neutrophil picture inside your tissue sample. I do see structure in the light pink islands throughout the tissue image with a sea of dark stained objects (nuclei?) between them. I can envision a pipeline segmenting the pink islands, but we’ll need a clearly defined goal/output if we want to succeed with CellProfiler. It would be ideal to have someone experienced with this specific histology annotate the image.

@karhohs: That’s what I was thinking too, that they might be embedded in the “pink islands.” We’re aiming to quantify neutrophil infiltration to the extravascular space.
My project collaborator and I don’t have the experience to identify the structures and our experienced histopathologist is unavailable. I am curious, however, to see what information the pipeline you’re envisioning might produce.

I only have basic experience, and I might have highlighted oddly cut gland (?) nuclei, but these might be your guys.
It doesn’t look like a a lot of leukocyte invasion overall, and that (as far as I see) is primarily of lymphocytes.

In this image of appendicitis you can clearly identify neutrophils close to the epithelial layer

I imagine you could could do fluorescence IHC for a neutrophil-specific antibody. Acquire images, then H&E them, then re-acquire color brightfield. You could use the IHC as a object mask to guide detection (or find some way to “train” it, i.e. machine learning).

1 Like

Thank you for the tips, @Fabba123. That’s an interesting approach. I will pose this to my team and see if this will be feasible for our purposes.