How to do automatic quantification of colony's circulairty

I have questions regarding to measure clonogenic colonies’ circularity. I am working with stem cells which form colonies, but they form various sizes and shapes. I found a way to stain them with CFSE dye (which can retain fluorescent signal for a long term). According to the pictures, some of the cells’ in the colonies are very condense (like attached pic #1), but some are not (like attached pic #2). Therefore, it is very hard to do consistent circularity measurements for each colony.
I want to find a way to do automatic measurement of colony’s circularity because I have large amount of data need to quantify like 100 colonies per day, so it is hard for me to do the manually count for each colony. Do u have any suggestion about how to do better measurement of circularity by using Cell Profiler?
Or do you know Cell Profiler can do a feature like doubled individual cell’s diameter to bring cells in the centre core of colonies into contact with adjacent cells. Then, I can outline the colony to calculate it’s circularity?


CellProfiler can help certainly. But what would really help would be if you could get a DNA/nuclear marker. De-clumping your cells is the main problem I see.

Also, you do mean circularity of individual cells, right? Not some gross measure of the whole cluster/colony of cells?

Please see my attached pipeline attempt. The procedure is to
(1) ID the “nuclei”, or brightest portions of the cells
(2) Grow these out (IDScondary)
(3) Measure Shape
(4) Display the chosen shape parameter (I chose FormFactor, but others might work better, like Compactness or Eccentricity)
(5) Classify the objects by choosing a threshold.
The counts will appear in the Per_Image output tables

It would definitely be helped if we could segment the nuclei first with, say, another marker. But these are live cells that you need to track over time?
Hope that helps,
DL_circularity.cppipe (10.4 KB)

Thank you for your reply!
We are interested in measure the circularity as a whole clumping colony, not as a individual cell.
As you see from the images, colony was made by lots of individual cells. Therefore, we would like to measure area, perimeter or circularity as a colony, not as a individual cell.

Thank you!

Oh! I guessed wrong then. Have you adapted your pipeline to suit the colony and not the individual cells then?


Hi David,

I am also interested in this problem. I need to quantify the shape of each yeast cell in a screen whether it’s unbudded, budded or budded with anaphase. Unbudded cells would be circular, budded cells would have a small bud do not have a nucleus and budded with anaphase cells would have an elongated nucleus and two nearly identical cells that are merged. I already extracted FormFactor, Compactness, and Eccentricity, however, I can’t decide how I should threshold them just based on these measurement. As you said, I am first segmenting the nuclei and then the cytoplasm, both in RFP, and a GFP marker is localized to the Cell Wall that might also be useful for inferring shape. Do you have any suggestions?

Best regards,

This is exactly the sort of thing CP Analyst is designed for, FWIW- you can take your measurements and images, sort your cells into your categories, and it will calculate for you what values do the best job of classifying your cells (and how many cells of each type there are in your experiment if you’d like). I’d definitely recommend checking it out- it used to be much more difficult to get started with, but if you use ExportToDatabase to make a SQLite file and tell the module to make your properties file for you it’s now super straightforward.

If you’re running CP 2.2 or greater (I don’t remember which version you said previously that you’re running) you can also use the WorkspaceViewer tool to overlay the measured values on top of the circled cells so that you can estimate the values by eye.