IdentifyPrimaryObject not distinguishing between clumps of cells

Image Analysis
, ,
#1

I am trying to use Cell Profiler to count the number of cells in an image from a Calcein stain and am using the IdentifyPrimaryObjects module to do so. However, the program is unable to distinguish between a few cells packed closely together, which is fairly common in these images, and is thus returning an incorrect count.

I have tried altering the “Method to distinguish clumped obejcts” and “Method to draw lines between clumped objects” to almost every combination. Intensity for both seems to work best, but still does not separate between close groups.

How can I make this counting process more accurate? Is there another module I should use?

Thank you in advance for any help!

#2

That’s indeed the module you should be using, and the setting you should be playing with - though you could also try tinkering with some of the other declumping settings such as the size of the smoothing filter, whether or not to use a lower resolution version of the image, etc.

If none of that works, I’d start tinkering with you image; see if by doing some sort of threshold application, enhance features, morph, etc you can try to make it more obvious where the dividing lines between your cells are. You can then try feeding that transformed image to IdentifyPrimaryObjects and hopefully have some better luck.

If you try all that and it’s still not working, feel free to upload a pipeline and some images. Good luck!

#3

I would just add that illumination correction (CorrectIlluminationCalculate, CorrectIlluminationApply) can also help in declumping (at least it did in my case). Also if you upload images we can be more precise :slight_smile: .

1 Like