Live/Dead Cell Counting 3D Gel (CZI)

I am brand new to image analysis. I am attempting to analyze confocal images of a 3D gel of human cancer cells that have been treated with a fluorescent live/dead assay. I’ve tried the 3D Objects Counter in Fiji on the dead (left window below) and live (right window below) but with poor accuracy based on what I can see on the light/grayscale image (center window below) - effectively undercounting cells.

Here is the appearance of the images when I open the czi in Fiji with Bio-Formats using hyperstack view with autoscale and split channels:

Screenshot 2019-06-18 20.32.56.png

I’m wondering what tools you might recommend using to do this analysis with more accurate results? I’m willing to learn any software package that might help. I have 54 gels each with 33 images that need to be analyzed. I’ve been doing one-by-one with the aforementioned methodology but have put that on hold for now due to the questionable accuracy.


EDIT: I should probably add that these cells have been treated with an intervention which, at high doses, can impact the morphology of the cell (some break). Example:

Screenshot 2019-06-18 20.58.49.png

I’d like to add an additional question onto my initial question above: is there a way to make ImageJ do automatic threshold calculations for each slice in a z-stack, rather than looking for an average threshold for the whole stack? I ask because some of my experiments, due to the intervention, have an extremely saturated out first image and a very poorly thresholded last image with a bunch of average slices in between. It would be nice if I could have the 3D object counter do a threshold assessment for each slice.


you can have a look at this tool here:

3D Segmentation and Analysis Module

which can be found on the APEER module page: APEER - Public Modules

or try out this fiji python script:

It required this one to function correctly:

Both solutions allow to select the option to calculate the threshold slice-by-slice.

1 Like

Thank you for the suggestions. I’m still extremely new to this whole process. How do I get the APEER module working in FIJI? I’ve downloaded the the zip file and installed it via the “Install Plugin…” feature. I received a message that it was installed successfully, but then after I reboot FIJI I don’t see any such “3D Segmentation and Analysis” in the plugin dropdown menu.

Sorry for being not clear.

  • to use the APEER module just register for, login, upload your data and try it out. There is no need to install anything.
  • to use the script you need Fiji and the copy the two scripts to Fiji’s script folder and run it

I’m getting an error where it can’t seem to find the fijipytools no matter where I put it.

Are both scripts inside the Fiji/scripts folder? Can you post the error?
The tool was working fine today on my Windows 10 machine.

I had them both in “plugins/Scripts/” and the 3d_analytics-adv was showing up in the menu that way. When I move them to “scripts/” I can’t seem to find them in the menu anywhere.

I’m on macOS.

Here are some image examples of how the single threshold is impacting my 3D object counting:

Screenshot 2019-08-22 11.48.54.png
Screenshot 2019-08-22 11.49.04.png
Screenshot 2019-08-22 11.49.24.png
Screenshot 2019-08-22 11.49.32.png

@mibg Sorry I don’t have time for an in-depth reply with concrete suggestions. But in response to this:

The typical object segmentation workflow in ImageJ/Fiji is to do some preprocessing of the image first, which paves the way for thresholding to be effective. For example, background subtraction or contrast enhancement/normalization. Here is the top-level guide:

You need to run the 3d script from the script editor


I have exactly the same issue with analyzing my confocal live-dead stained images. I was wondering if you could solve your problem.

Appreciate any help.