Interlaced particles

I have a rod-shaped particles that are interlaced and I would separate them in order to count. How can I do?
Also, I would like to know their dimensions with a scale in nanometers. How of can I do put it insteaod of the number of piwels. Thanks for your answer.

(Moving this topic to the CellProfiler 2.0 Help forum)

If you have a representative image, could you upload it to this thread so we could better assess the problem?

For this one, please see this FAQ item for details on this issue.

Sorry for my late reply but I did not see before your. Thanks a lot for your answer. For my representative image, should I attach to my post?
Thanks for your answers.

Yes, you can use the “upload attachment” tab. Or give us a link to a Dropbox location, etc.

Hi Claverie,

I looked at your image (I attached it here for others’ benefit - I can remove it if you like) and here are a few comments:

*Note that the image is actually a “color” image, i.e. an RGB image with the intensities all equal so that it only appears as grayscale. I know this because it uncovered a bug in CP when I tried to use our Morph module on it! The bug is now fixed in the Developer’s version). You can save yourself some space when you export, if you export as a single grayscale plane/image, and it will process faster and without more issues

  • These objects will be difficult to count accurately. The cells overlap (bad) and they are of different length (also bad). I took a try anyway (see attached small pipeline) - but since it is hard for me to count these accurately by eye, you may be able to tolerate a significant variation if the automation benefits you enough.
  • The Morph module applies a tophat filter, which highlights objects ~30 pixels wide or somewhat smaller, which is about the width of the cells. It is not great, but helps to minimize the background variation at least
  • The IDPrimary module segments the cells, but the cells that are adjacent along parallel axes are hard to split properly
    ** The “Clumped Objects” settings could be optimized with more time, likely. (halfway down the settings of IDPrim)
    ** You could also try the Worm Toolbox in CP. It is designed to segment C. Elegans worms which have similar properties as these objects (lie parallel to each other, clump together, etc.). However, since the sizes vary substantially, the model you train will be size-dependent. So give it a try (I am not expert using the Worm Toolbox), but I suspect it won’t magically solve your issues.

The best advice I can give is to try and keep the cells physically apart, or at least in a monolayer. I realize this may not be feasible, though.

Good luck!
interlaced_test_pipeline.cp (3.93 KB)

Thanks for your answer and I prefer this image is removed if it’s possible.
Best regards

Image deleted. I was unclear and thought you just had trouble uploading it, so apologies if this caused any trouble for you.