Problem with Cell detection

Hi there!
I’m using Qupath for a while but since few days, i have a problem with cell detection. QuPath detects cells when there are none. I tried with severals images but each time i have the same problem even though the parameters i used have always been the same.

Hi @YBUSCAIL,

You can change the parameters in the Cell detection pane, which will influence the classification. There is a more detailed explanation of what the parameters do in here (even though these videos have been made in a much earlier version than the current one, some of these ideas still stand).

E.g. changing the threshold might help… We could give you a more in-depth guidance if you explain a bit more where/what the problem is :slight_smile:

Thanks for your answer :slight_smile:
I have already seen this YouTube chanel. It helps me very much!
I try to explain me better : i work on a TMA image. I apply the same parameter for each TMA core but the result seems to be aberrant for some TMA core.
Example : to the left i have a pretty good detection while to the right qupath detects a lot of false cells :


Could it be because of the quality of my image ?

Thank you for your help :slight_smile:

Have you adjusted for your background (Estimate stain vectors, needs to be run before opening the cell detection window)? It looks rather high, just based on the color of the empty areas vs the color the the QuPath menu bar.

Cell detection: Your Threshold for “Optical Density sum” looks really, really low. Try increasing that, and setting the Background Radius to 0 to start. If you get it working that way, then you can start adding the Background Radius, which can help exclude cells in tissue folds.

1 Like