Detection & Count cells

Hi,

I work to detecte osteon cells on picture and then count and classify them.
I did some try and I have some results but the detection is very hard because the cells are close to each other.
So firstable I wanted just to detect and count the all osteon cells and that’s why I need your help.
If you have any information that it could help me your are welcome !

just below you have the picture to analyze, my pipeline and the results.

Best Regards

Stefano
test2.cp (5.82 KB)




Hi Stefano,

Sorry about the delay on your earlier email. We had taken a look at the image and it’s actually quite difficult to process using any of the tricks that we are familiar with. We have tried using pixel-based classification but it’s not very effective.

However, there is a possibility using the attached pipeline. This pipeline runs with the latest build of CP available here. Please note: This build is not an official release, so any pipelines made will not be backwards compatible with our latest official release (r10415) and we do not guarantee compatibility with any upcoming release.

That said, this pipeline uses a filter which we’ve added to EnhanceOrSupressFeatures which attempts to detect circular features of a given size, in this case, 40 pixels in diameter. The output is then speckle-filtered to highlight the spots at the center, which are then identified using IdentifyPrimary and grown outwards using IdentifySecondary.

Again, this result is using a specific feature size, so if you want to detect additional osteons, you would have to add these same modules (EnhanceOrSupressFeatures with circles, EnhanceOrSupressFeatures with speckles, IdentifyPrimary, IdentifySecondary) for each feature size you want to look for. The final result would be the combination of all the objects found at each feature scale.

This is not an optimal solution by any means, but it’s the best we could come up with. If you have access to an image processing expert, you may want to consult with them.

Hope this helps!
-Mark
2010_10_28.cp (6.27 KB)

Hi Mark,

Really sorry for the delay but i would like to thank you.
Your pipeline works efficently and i have some good results. I’m still working on because there are some problem of detection but as you say it’s a very difficult type of detection. I will see with an “expert” of image analysis if he has some tips for me.

Thank you again.

Best regards !

Stefano

PS: if you have some new detection types that you add on cellprofiler and could help me please let me know.