Thresholding methods

I’m having some trouble with the thresholding methods in the IdentifyPrimaryObjects module.

There are several different methods and I’ve tried manual and Otsu methods but I can’t seem to find a sweet spot with the cell counts that the pipeline produces. The counts change with shifts in threshold value and I wanted to see if there is any way to narrow down the accuracy of the counts with these methods.

I tried the manual method with a threshold value varying between 0.7 and 0.8 and Otsu with range between 0.7 and 1.0 and correction factor of 1.4.

Is there a proper way to go about these methods or any particular direction I should take for cell marker identification (ie CD8, Foxp3…etc)?

This applies for an inverted grayscale image of 4T1 cells.