Count cells without stained nuclei


#1

Hi,
i would like to ask if it is possible to count cells in Arabidopsis petals for which there is no dapi staining. Is it possible to count them by avoiding the module for the primary objects like nuclei because it is not possible to do this staining in my protocol ?. I ma attaching a picture of how they look like.
Many thanks and congratulations for this amazing software.



#2

Hi Rachil,

Keep in mind that the IdentifyPrimAutomatic, the module which does the main work of identifying objects, doesn’t just work on nuclear-stained images. The parameters default to those typically appropriate for nuclei since that is a very common use, but it is not limited to only that application. Basically, the module is directed towards identifying any bright objects on a darker background, be they nuclei or something else.

That said, IdentifyPrimAuto can be used to find the petals in your picture provided you are able to either (1) petals as bright as possible compared to the darker edges; or (2) invert the image (using InvertIntensity) and in the resultant image, make the borders are bright as possible compared to the darker petals. However, based on my attempts with your sample image, the edges are somewhat blurry and indistinct, making both (1) and (2) difficult. Is there any way to focus more on the edges and try to make them as clear as possible? If so, I think you’ll have much better luck.

Regards,
-Mark


#3

Hi again,

Another option is to invert the image using InvertIntensity and then use the SmoothOrEnhance module, selecting the “Enhance BrightRoundSpeckles (Tophat Filter)” option. That will enhance the edges in your images to the point where IdentifyPrimAuto can pick them out.

I should mention that the fine edges within the petals will also get enhanced along with petal edges. So IdentifyPrimAuto may have difficulty thresholding and picking out only the edges that you want. Still, with a judicious selection of settings, you might get it to work successfully.

Good luck!
-Mark