Problem with identify primary objects

I posted here previously because I could not open images; that problem disappeared magically when a colleague allowed me to use his high RAM, 64-bit system. I am working with three-color immunofluorescence (DAPI/red/green) and cannot seem to identify the nuclei as objects no matter how I fiddle with the settings (I tried rescale intensity, enhance edges, and playing with illumination correction in various combinations). Each time zero objects are identified. The first part of the pipeline (loadimages, crop, colortogray) works just fine. I have uploaded a small cropped portion of the image where I am trying to identify primary objects (DAPI+ nuclei). Any thoughts would be helpful.

Hi,

Glad to hear you got it working; I wasn’t sure if you had a 64-bit machine available.

Without seeing your pipeline, I’m not sure what the problem would be, but in any case, I’m attaching a pipeline. I’ve increased the typical diameter limits from the default values, and changed the declumping method to Shape/Distance since your nuclei are fairly round. It seems to work reasonably well.

Does this help?
-Mark
2010_04_29.cp (4.71 KB)

Thank you – this works. I did not try adjusting the method to distinguish clumped objects, and that seems to have done the trick.

I very much appreciate your help. I am putting together preliminary data for a DoD new investigator grant application (due in 4 weeks, hence my hurry) and the capability to do high throughput sophisticated image analysis with my material will figure prominently in the application. To my knowledge no one has tapped archival paraffin-embedded tissues in quite this way, particularly for this disease process. I am putting a fancy computer in the budget proposal, since that is clearly critical. I am also sure that as I try to scale this up (eventually 100s of 0.6 to 1mm tissue array cores) and tap into the capabilities of cell profiler analyst I will undoubtedly be asking for more help. If you would like to talk about formalizing a collaboration, let me know.

Great to hear that it’s working!

Our platform is certainly interested in setting up collaborations with outside groups. When you’re at that point, please email our platform director, Anne Carpenter (anne@broadinstitute.org).

Regards,
-Mark