Help IdentifySecondaryObjects in CellProfiler with unique shaped Cells and General pipeline advice

Hi all,

I am currently working as a research assistant in stem cell research. My task is to extract as much useful/accurate data as possible over a large range of image sets. I have been working on and learning Cellprofiler for 4 weeks now and feel like I am progressing well much lacking a bit of knowledge.

Issues that I am experiencing:

IdentifySecondaryObjects is not accurately identifying the Cellbody’s and sometimes completely not finding them. I have had a play around with the advanced settings but can’t seem to get it to work efficiently across all images. I believe that it could be the method that I am using.

Alternatively I have tried to IdentifyPrimaryObjects then RelatedObjects however in some cases that works effectively but it relates 2 or 3 nucleus’s to the cell which is not useful data.

I have uploaded the pipeline that I am using and some example images. Hopefully somebody can guide me to a solution and I will be very grateful. Alternatively I am happy to take any improvement advice on my pipeline. I understand that my analysis is tricky because many of the images contain cells that are touching the borders and are odd shapes.

PIPELINE: MSC5x5ComplexPIPELINE.cpproj (693.5 KB)

Example IMAGES: (upload://uCkqTJI0mAXFxUE486TMX0xc0q1.cpproj) (693.5 KB)

Hi @sspag1,

Taking a look at your pipeline, you appear to be trying to run IdentifySecondaryObjects on an illumination correction function rather than the image itself, which had some very unusual settings in that module which produce poor results. I’m attaching a simplified pipeline which just works with the images directly and will hopefully be a decent starting point.

IdentifyPrimaryObjects includes it’s own smoothing filter, so I’ve also removed the redundant smoothing module.

You may want to take a look at some of the tutorials we have available.

MSC5x5ComplexPIPELINE.cppipe (48.7 KB)

Hope that helps!

1 Like