Sometimes it is necessary to detect objects of different types independently. E.g. nuclei and cells. And then we want to work in Harmony image analysis with the cells-nuclei combination. Currently there is no tool to assign "nuclei" to the "cell" based on largest overlap. We can do something similar (with a bug) with an ABB. It would be good to think of how we can generalize the creation of such a relation.
That should then ideally also contain options to assign regions from one population (e.g. nucleus) to the other region (e.g. cell). Currently we don't have any tools for this (except for an interesting combination of ABBs by Olavi ;) ), as the nuclei might not completely overlap with one cell, but even extend into another cell.
Applications is always, when an approach with ROI does not really work. E.g. limiting Find Spots to a cell when the objects might attach from the outside to the cell and the cell detection is not perfect. Or independent detection of nuclei and cells in an assay with multi-nucleated cells.