Cellprofiler can't find DAPI primary objects in images with few cells


I am checking through my images before running analysis with cellprofiler and I have noticed a smaller selection of the images where cell profiler is not able to identify the primary objects (nuclei) at all. This seems only to be the case in images with a few cells so I thought it might be something with the background staining. I tried running the pipeline with the correct illumination apply/calculate module for DAPI but it doesn’t make a difference. Anyone has a clue what is wrong??

I have attached my pipeline along with a normal imageset and a weird imageset
NIH3T3 D1 RhoA E40Q 01 org-Image Export-19_b0c0x0-1388y0-1040.tif (4.1 MB) NIH3T3 D1 RhoA E40Q 01 org-Image Export-19_b0c1x0-1388y0-1040.tif (4.1 MB) NIH3T3 D1 RhoA E40Q 01 org-Image Export-19_b0c2x0-1388y0-1040.tif (4.1 MB) NIH3T3 D1 RhoA E40Q 03 org-Image Export-21_b0c0x0-1388y0-1040.tif (4.1 MB) NIH3T3 D1 RhoA E40Q 03 org-Image Export-21_b0c1x0-1388y0-1040.tif (4.1 MB) NIH3T3 D1 RhoA E40Q 03 org-Image Export-21_b0c2x0-1388y0-1040.tif (4.1 MB) Pipeline JNO 2.6 changed for 20201012 with std.cpproj (2.3 MB)

Hi @EllenAppel,

The automatic thresholding methods use histograms to find the optimal threshold value. In unusual images with very few cells these methods can fail due to the lack of a prominent peak of staining. For this reason it’s best to carefully specify a minimum/maximum threshold to use as a ‘backup’ for when there are too few or too many cells to properly use a histogram. In your pipeline changing the lower bound for the threshold in IdentifyPrimaryObjects from 0.001 to 0.01 might help in this instance.


Hi David,

Thank you so much, your suggestion worked!
You’re a lifesaver :pray: :smiley: