Unable to perform cell detection on fluorescent TIFF files

I’ve been using QuPath to analyze my DAB staining for some time now and have always had great success with my positive cell detection. I am now trying to analyze fluorescent images but it wouldn’t pick up on any cells. When I select an annotation to quantify, it begins the process, then quickly finishes without ever having identified a single cell even though I selected the correct image type, “fluorescent”.

The following post helped me to identify an issue with my pixel sizes: Trouble with Positive Cell Detection in Tiff files - #6 by Research_Associate. As you can see, the scale bar in that image is in mm rather than um, and it turns out this was the case in my images as well. However, what I can’t figure out is how to correct this issue. In my case, when I click the image tab, it says my metadata is not changed.


Can anyone please offer some insight into how to correct the pixel size of my images?

Thank you in advance!!

In addition to needing to be very careful about the threshold with fluorescent images, you will need to set the pixel size correctly. I do not know what correctly is for your particular image, but the code looks something like
where x and y are the length and width of a pixel. You should be able to get that from the original image from the microscope, or ask your core facility for help. The pixel width and height entries in the Image tab can also be double clicked to set the pixel size manually, though the script will be far more useful for a project.

Thank you, I see how to manually change pixel sizes now.

Would you happen to know if there is a way to filter cell counts from one channel to the next? In other words, if I would like to quantify how many of my green cells are also red, how can I do this? It seems that QuPath will only allow a single quantification at a time, but doing it sequentially would be most useful.

Thank you!

hi @lcm3aj ,

you can create a classifier for the green cells, then one for the red cells and finally use “Create composite classifier” to use both!



1 Like

Most of what it sounds like you are describing is covered here in the docs:

1 Like

Thank you, @romainGuiet and @Research_Associate !!