Cell expansion parameter in cell detection modifies the number of detected cells

Hi all,

I’ve noticed something that seems strange to me.
I used the QuPath cell detection with cell expansion set to 0µm to get nucleus. Then, I tried with 5µm expansion. But the total number of detected cells is different.
In my mind, this function should have detected nucleus, then tried to expand them as far as 5µm, but this parameter should not influence the number of detected nucleus ?

Thanks for your advices


1 Like

I hadn’t noticed this before, but I can replicate it if I choose a large area – i.e. large enough that QuPath will have to break the region into tiles for detection.

I haven’t checked through the details, but I expect it relates to how QuPath resolves cells occurring at the tile boundary overlaps. You should get the same number of detections when using the same parameters, but changing any parameter can potentially make a difference whenever the region is large and needs to be tiled.

Hi Pete,
Thanks for your answer. Good to know.
I tested it on a series of 86 images, and the total difference between no expansion and 5µm expansion ranged from ~1% to ~6%.