Counting GFP Stained Neurons


I have been using CellProfiler for a few days now, and I am very impressed with the program so far. I counted the cells using the following modules: LoadImage > ColorToGray > IdentifyPrimaryObjects (a picture of the settings is attached)

However, there are some minor issues in counting the cells, and I was hoping someone from this forum could help me with them.

The first issue is regarding the unclear sections of my images (check the attached example). I would like to exclude them from there so the program does not attempt analyzing them. These sections include the background noise and the areas that contain cells with low color intensity. Is there any way to exclude regions with low intensity from the image? Basically what I would like to do is to exclude the noisy pixels of the image so CellProfiler distinguishes the cells easier.

The second issue is that the program sometimes counts a small portion of the same cell as another cell (just like the example in image 2). Also, sometimes the somas of the neurons are clumped together and are counted as one. Is there any way to make the program count cells that are bigger than a certain surface area as two cells? And is the problem in image 2 solvable?

The original and the analyzed images are also attached to this post.

Thank you!


Indeed, there are many approaches to exclude these regions, and you are already using some of them! IdentifyPrimaryObjects has a threshold settings to only include bright objects. You have it set to “Automatic” but if you switch this to say, Global, you will see many other thresholding methods.
Additionally, you can pre-process the image to try and remove the noise. Try Smooth module > Median filter with a filter size smaller than your somas. There are also Illumination correction modules to try and remove slowly varying background intensities, CorrectIllumination[Calc,Apply].

We call this issue “declumping”. In IDPrimary, you have “Method to distinguish clumped objects” set to “Intensity”. Your objects don’t have much intensity difference between clumped objects it seems (can you fix this on the imaging side?), so I would try rather Shape, and also Shape for the next setting. It won’t fix all your issues, but will likely help.

Settings right beneath the above “Declumping” settings allow you to change the size of the declumping filters. Adjust these bigger or smaller, depending. It’s hard for us to tell what might work best, but you could attach your pipeline (.cppipe file) and we could see more easily. Frankly it’s a bit hard for me to distinguish some of these soma by eye, so you shouldn’t expect CellProfiler to do any better than that, just so you have reasonable expectations.

Hope that helps,