Cell count

Hi there,

I’m new to CellProfiler and I’m having some diffuclities.
I need to count cells (macrophages) after a migration assay using a Boyden Chamber.
I tried to build a pipeline but the problem is that the identifyprimary doesn’t identify cells but background, and when I invert the pic with ImageMath/Invert result is the same. I don’t understand why. How can I identify all of the cells in the picture?

I attached a relevant picture as example. Cells were stained with Eosin and Methylene Blue, but picture is in gray scale as we only have a black-and-white camera. Small circles we can observe on pic are migration-membrane-pores and should not be interpreted as cells.

I would really appreciated any help.

Thank you,

Inverting the image prior to cell identification is the right approach. I would suggest changing the threhsolding method (perhaps Otsu, 3-class with the middle class set to background) and adjust the threshold correction factor upwards as needed. Perhaps adjust the min/max diameter to 20 (min) to 50 (max) as well.


The only thing I would add to Mark’s comment is that if you could add nuclear marker (e.g. DAPI, Hoechst), it would be much easier. I had tried a few image pre-processing tricks (Enhance Dark Holes in EnhanceOrSuppressFeatures, e.g.), but a different marker which does not adjoin other cells would help a lot I think. Perhaps even instead of Eosin?

Many thanks for your advice. I gonna try a Dapi staining.


I stained my cells with Dapi (1/1000), instead of Eosin, but I still have some difficulties to identify all of the cells. And, unfortunately, pores are also stained.
You will find enclosed two pictures I obtained, as well as the best pipeline I was able to build. Please, could you give me some advice to improve my pipeline?

Many thanks,

Pipeline_Dapi20x_2013.06.26.cp (8.13 KB)


Glad you could add DAPI. Unfortunately it doesn’t seem to help? The staining is very bright, and the pores obviously soak up the DAPI which is a big problem, but regardless, the DAPI seems to be staining the whole cell? The blue haze around each cell appears as big as it was in brightfield, no? Perhaps macrophages are hard to stain (I don’t know) but I bet you can use less stain and improve your signal to noise.

One other comments, regarding ColorToGray: You should Split, rather than Combine. Your raw image is only in the blue channel, so when you Combine RGB, you end up combining 2 blank channels (RG) with the single channel and so your final intensity values are 1/3 as large as they ought to be. Better, you should try to save directly as grayscale – the image files will be smaller too.

Have you gotten better staining now?
Good luck,