Counting double stained images

Hey everyone,

I’m just starting to use CellProfiler, and I need to count cells. The slides of interest have been stained with DAPI, and two stains conjugated with Alexa 488 and 555. An example picture is attached. I’ve been tinkering with a few pipelines I’ve found related to counting in this forum(example percent positive and one called PipelineF_DLogan) , but I’m having an exceptionally difficult time figuring out all the settings and the different steps in the analysis modules.

For example in the attached picture, the green labeled cells (counted by dapi) number around 22, while the red labeled cells are around 8 or 9. Would Cell profiler be able to help me do this sort of analysis?

thanks a ton


Certainly CellProfiler can analyze this image. Please see my attached pipeline to get you started. Drag this onto your pipeline area on the CP window (left side) and then drag your image(s) into the Images module.

A few key notes:
(1) Your image is a multi-plane TIF. This is somewhat more complicated to deal with than plain grayscale, individual images. You must extract metadata in the Metadata module for CellProfiler to know how to split apart and assign the “planes”. In fact, what you call “red” is not what CellProfiler assigns as red. (These are just pseudo-colors anyway.). But if you could export these images as 3 grayscale images, this procedure would be very much simplified!
(1a) So, in Metadata, click on BOTH update buttons. One is “Update metadata” in the middle of the settings and the other updates the table at the bottom. This will extract the metadata from the file, and allow you to see what metadata distinguishes the 3 channels from each other. You should see that the “Frame” and the “C” metadata both are 0,1,2 for the various channels. I use this info in the NamesAndTypes module.
(2) NamesAndTypes: Use the metadata to assign your channels. I just called them “Ch[012]” but you can reassign them here to whatever you like.
(3) IdentifyPrimaryObjects: Here we are “segmenting” each object/cell separately. I used all the same settings for each of the 3 modules – one for each channel. I didn’t spend a lot of time optimizing, but you could improve this by unchecking the “Automatic” for Thresholding Strategy and many more option appear.
(4) Your images seem to have a horizontal rastering, or shift. Is this a laser scanning imaging device? If this could be optimized the quality of the image analysis would be much improved I think.
(5) I didn’t add any Measure or Export modules, but that depends on what you want to analyze. If it is just cell counting, you don’t need any Measure modules per se.

Hope that helps!
DLpipeline.cppipe (10.3 KB)