Get the prediction probabilities for cell type classification


I have trained a cell classifier to classify my detected cells into 3 types: cancer cells, lymphocytes, and stromal cells. My question is: is there a built-in way to get the prediction probabilities for each cell? For example:
cell 1: 0.14 (cancer cell), 0.76 (lymphocytes), 0. 1(stromal cell)
cell 2: 0.92 (cancer cell), 0.01(lymphocytes), 0.07(stromal cell)

cell n

Looks like the output of cell classification does not have this feature, and when I create the classifier there is also no choice allows to include this statistic.

Any suggestion? Thank you!

No, some classifiers will set a single probability value within the object (visible alongside the measurements) but not all. There is no way for any of the built-in classifiers to add their prediction probabilities to the cells… although most classifiers don’t give a very probability-like output in any case.

Thank you for the reply! My thought is that there should some kind of probability value when classifier assign type to each cell. In my case, I am only cares about those cells with high probabilities as they are most likely to be correct, given that I don’t have much pathological experience to manually tune the classifier.