Removing irregular background patterns from DAPI images


I am trying to count the number of DAPI stained nuclei on polymer spots. Some of these spots autofluoresce, creating an irregular background and false positive objects in IndentifyPrimaryObjects.

I am using OTSU>Three Classes for my thresholding and have played around with the threshold correction factor but I lose some nuclei which have a dimmer fluorescence when I increase this.

FilterObjects based on Eccentricity has removed many artifacts but I am still getting some.

I have attached a selection of images which mostly contains ones for which I get background errors but also few images that don’t have any background issues.

Currently using CellProfiler 2.2.0 on Win10

Current iteration of my pipeline also attached.
Any help would be greatly appreciated.
Kyle.Doherty_1.cppipe (12.1 KB) (10.9 MB)

Hi Doherty

Are you still having problems. This is a common question that I have answered couple times, so I suggest look at the threads. But simply I would look into enhance speckles. CellProfiler also has a great tutorial on their webpage.


Hi Lee,

Yes I still have the issue but I will look into enhance speckles.
Thank you

Hi Lee,

I have had a look for previous posts again but I couldn’t find anything useful, perhaps I have missed them.
Looking on CellProfiler website I thought CorrectIllumination might be useful but I haven’t found an improvement using it.

Regarding Enhance Speckles: do you mean for me to use this to enhance my nuclei relative to background artifacts?

Many thanks,

Hi Kyle

I would go with otsu two classes, because you have only a foreground and background. Lower the threshold correction factor to increase the number of identified cells.
Below is the enhancement of nuclei prior to identification/segmentation


Edited pipeline
Kyle.Doherty_1.cpproj (649.8 KB)


1 Like

Thanks Lee,

That is very helpful.


1 Like