Epidermal Cell Counts - Make nearby particles coalesce

Hello All,

I’m creating an analysis pipeline to take 400x micrographs of epidermal impressions. The outlines of epidermal cells adjacent to stomata are not easily detected due to the drying of the nailpolish I used to make the impressions. When I threshold & make the image binary, the cells adjacent to the stomata come out as lots of little particles.

I would like to coalesce these particles so that the particle analysis counts them as one.

Any suggestions for a method to clump these particles together?

Here is a link to the original image, the image that particle analyzer counts from, and the results of a particle analysis.

Epidermal Cell Counting

Thanks in advance,


If it helps, here is a text version of my protocol:
Load image
Threshold > percentile threshold method
Make Binary
Binary > Open
Binary > Fill Holes
Analyze Particles
Size (pixel^2) = 800 – Infinity
Circularity = 0-1
Show Outlines
Include Holes

1 Like

Hello Karl,

Without being an expert it is hard to say where the borders of the difficult cells are. Maybe you can try using the Trainable Weka Segmentation plugin. I would convert the image to grayscale and start by using just a few image features.

Another option, if you can live without counting those cells, would be to set the particle analyzer so it skips small regions.