3D Segmentation in CellProfiler with Subpar Membrane Stain

Hi everyone, I’ve been working with CellProfiler to try segmenting a monolayer of cells in 3D. I started with the tutorial on GitHub (https://github.com/CellProfiler/tutorials/tree/master/3d_monolayer), and I’ve been able to get that to work, but I have issues when I try to apply it to my own images. I think the issue is primarily with the quality of the images themselves. The nuclei are not particularly uniform, which leads to both over- and under-segmentation, and my membrane stain is really more of a suggestion of a cell boundary rather than a clear line. I’ve gotten the nuclei segmentation to a point that I’m pretty happy with, but I’m still struggling with the cell watershed. I’ve used a combination of EnhanceOrSuppressFeatures, RemoveHoles, DilateImage, and Opening to try to make my membrane clearer, but Watershed seems to often ignore my membrane and draw cell boundaries elsewhere. Any advice that people can give would be much appreciated.
The last few modules of the attached pipeline deal with bacteria segmentation and can just be ignored. The end goal of the pipeline is to relate bacteria with the cells they occupy.
Thank you!

Uploading: 20200316_shNT_LmGFP_phall_bcat_hoescht_05.tif_Bcat_processed.tif… 2020_06_24_3DSegmentation_CombinedPipelines.cpproj (829.1 KB)

It looks like my images didn’t upload. I’ve put them on Dropbox and they should be accessible:

Hi @davidmb,
I just checked at your pipeline. I tried with two changes as follows,

  1. changed smoothing scale to 8 in Enhancesuppress feature (since your interest is to get the boundary so increased it a bit)
  2. Changed slightly the thresholding parameters by relaxing the criteria. PFA screenshot.
    It looks better.
    One more suggestion would be is to try removing the last frame if possible just to ensure about your settings. Rest of the steps seems to be ok.

Hope this helps,

Hi Lakshmi,
Thanks for the notes. I’ll try that and see how it does.


1 Like