Nuclei clumping - How to resolve?


I have a Dorsal Root Ganliga section here from mouse, stained with DAPI. Unlike cells in a dish, some of the nuclei are very close together and have irregular shapes.

When the nuclei are well separated and round, IdentifyPrimaryObjects works just fine. But for clumps with weird shapes, the program is struggling.

here is an example. Does anyone have any tips?

And I’m attaching the original too. Sorry if this is a newbie question.


I’m really not an expert, but i had to face images similar to yours.
Usually i set the declumping method on “intensity” and method to draw dividing line on shapes.
After this i tinker with local maxima value and the threshold correction factor.


That’s great advice! I’m not sure but I wonder if one of our written tutorials goes into the IdentifyPrimaryObjects settings in more detail to help in this process (you do have challenging nuclei!). Let’s see if @Minh @karhohs or @bcimini knows.

RNAscope in Spinal Cord Neurons -> Quantitation of overlap between different In Situ Probes

Thanks @Anne_Carpenter and @Valerio_Laghi. I’ll try what @Valerio_Laghi suggested, and if anyone else has suggestions, I’m all ears.


Tinkering as @Valerio_Laghi mentioned is probably your best bet, though given that these nuclei are a bit “spotty” you may want to do a bit of smoothing on them first to keep it from finding all the smaller bright bits in it (this is also where the local maxima value will help you out!)


@achamess if i remember correctly there are useful guides about the tresholding on github, but can’t find them now.
If you find yourself with too many clumped object you can try another way.
I often have to count the dapi nuclei in myotubes and to declump the object i use the morph module to “erode” and “clean” the nuclei.
Basically i reduce the nuclei dimension, but make them easier to count.