How can I build a composite classifier with, say features like, Nucleus Area and Mean Intensity (Hematoxylin) without putting either of them in a subclass? I want to use this classifier to quantify positive and negative cells for a stain.
Do you want to manually set those thresholds or simply limit a machine learning classifier to using those measurements for the classification? You can create composite classifiers of ML classifiers in addition to the single measurement classifiers.
Alternatively, there is scripting, where anything is possible*.