A first-pass mask from a single threshold is rarely the final form you want. Real images vary across the field, hold isolated extreme pixels, and contain regions whose shape is almost but not quite the one the analysis needs.
The typical pattern is to produce a starting mask, then shape it. Shaping uses morphology (dilate, erode, open, close, fill holesLoading...) to express the geometry you want. Combining with AND, OR, XORLoading... keeps only the intersection of two masks, or their symmetric difference.
After that, you label. Separating touching objects comes after labeling: the Cut bottlenecksLoading... label operation splits a detected event at a thin neck. Then you measure. The Operations Editor chains these steps in order, and each operation can use the outputs of the operations before it.