I have some nice images of kidney tissue, where the endothelial cells in the kidney tubules are labeled with GFP. I am trying to segment the tubule cross-sections using a pixel classifier -> create objects workflow.
Is there a way to threshold the output of the pixel classifier at something other than 0.5 in a binary classifier scenario? I want to leverage the fact that the classifier is less confident in the space between closely touching tubules to achieve better separation.
Thanks in advance for the help.