Analysing hypertrophy and multinucleated cells in histology

In mouse liver histology hypertrophy and multinucleated cells are detected. I am now planning the image analysis project, so the image data (stained whole slide images) will be produced after I know the ia method. All ideas are welcome!

Analysis goals
Calculating the amount of cells and area of each cell by using membrane staining. Then counting nuclei and comparing to the amount of cells in order to see how much there is multinucleated cells present.

Challenges /Questions
What would be a good staing method for membrane and nuclei? Histological, IHC, fluorescence (in liver tissue fluorescence is usually difficult)?
Is it possible to perform this analysis in QuPath?
Does anyone have experience / tips / links for something similar?
Of course in histology counting sizes is a bit approximate, since the object can be cut from different planes, but if there is enough data, the relative differences would be seen. Or does anyone have any other ideas or approaches for this?

Thanks in advance!

Membrane staining for cells is challenging to analyze in QuPath as currently the “cytoplasm” is only a blind expansion. QuPath has no idea where one cell ends and another cell begins, and so cannot associate any particular staining with the correct cell. It is only checking what is nearby.

Some tricks, if the membrane staining cells cluster, would be to use a pixel classifier to classify cells within areas of heavy membrane staining, or if the cells you are studying have “tight” cytoplasms, like T/B cells, you can use the cytoplasm or nuclear mean.

Essentially, the farther the staining is from the nucleus, the less accurate your analysis will be. For cell outline based segmentation, I would probably recommend CellProfiler, though that may struggle somewhat with whole slide images. I know they have made some updates recently, so maybe lazy loading has made it in and I am out of date :slight_smile:

Another approach, especially if you are concerned about out of plane staining, is to quantify the area of positive staining using a Thresholder. That bypasses the issue of associating the staining with the correct nucleus.
Without sample images, it is hard to say any more.