Segmentation with nuclei and not so great membrane staining

cellprofiler
segmentation

#1

Dear Cellprofiler pros,
I am having huge problem segmenting my pictures with DNA (as primary) and membrane (as secondary) objects.
My membrane staining is not very sharp, however with the resolution limitation of our imaging platforms, this is hard to improve.
I am trying different ways to preprocessing the membrane marker and trying different parameters in the “identify secondary object module” but nothing seems to help much: my membrane label is not well recognized.
I am attaching here the images and the pipeline I used.
Could you help me improve my segmentation?
Thank you!

Sorry for some reason the image files cannot be uploaded directly in here.
I put them under separate links.

segmentation.cppipe (13.2 KB)
https://drive.google.com/file/d/1_2iSpl67yOKuOC7WACdlvRAVxtklZ_x9/view?usp=sharing https://drive.google.com/file/d/1Dhx5NhGONN0QHCgkXF7I0rjTnuuOqYUp/view?usp=sharing https://drive.google.com/file/d/1gA7jkPs9BQjnHL_RI56ZnDHpnGYmGIKk/view?usp=sharing


#2

Hello Julia,

I had a look at your pipeline which seems to be ok. One thing you can try to do Global and Otsu thresholding strategies for both identifying primary and secondary objects. This might help a little bit. But this is maximum you can achieve with such image resolution. Please let me know if you need any specific help?

Thanks


#3

Dear Habbasi,
Thanks for getting back to you. I do have a question regarding “IdentifySecondaryObjects” module. For Distance-B method, should I invert my membrane marker to make it higher value inside the cells? From the description I cannot really tell.
Thank you!