Lower threshold ignored in IdentifyPrimaryObjects

The IdentifyPrimaryObjects module ignores now the lower threshold and segments objects below the set threshold forcing me to have a threshold module prior to segmentation. This is likely a bug that needs fixing as it worked just fine in the past.


Can you upload the pipeline and image set that generated this issue for you, and let us know what CellProfiler version you’re using? It will be difficult for us to track this down without that information. Thanks!

Hi Beth,

Thanks for looking into this. I’m using CellProfiler 4.07 and it happens with any image I’ve tried. I can’t upload the last specific example as I have now done Threshold → Imagemath to replace that function. But I can upload a simple example and make a toy pipeline if you wish.

An example pipeline and image set would be great for us to test this out and try to reproduce the error. If you could also use some screenshots to show what you mean about how CellProfiler is segmenting objects below the set threshold and to contrast that with the desired behavior, that would also be helpful. We are image analysts, so images truly are worth a thousand words!

Hi @MarcKobudo,

I’ve not been able to replicate this thus far, the lower bound seems to be working normally. If you could send the pipeline that’d at least let us check which module settings you’re using to produce this issue. Most importantly - which thresholding strategy was it set up to use?


I was assembling two images to demonstrate the behaviour, but failed to reproduce it too. I then went to the original dataset and reproduced it again. That dataset has ~1500 images. Do you think it could be an update issue? Might changing the parameter propagates slowly when lots of images are in the working directory? I have seen this behaviour irregularly which is why I have not reported it earlier. I’ll try to capture the next instance and report back. Sorry for the bother.