Ignacio, thanks a million for your input.
This approach I used to determine the area, which is a very important parameter for me as well. I am now left with trying to determine the basic shape of the mesh. To define the yellow areas on basis of a circularity parameter for example, can be useful, but I think it is hard to conceptualize. It is a (flexible) filamentous network, so it can be approximated to a stiff mesh composed of some assembly of polygons… but which ones? and in which proportions?
Another perspective on the problem is to imagine that one of the ROIs has a rhombus shape (4 vertices), but once you look at the filaments you see that only 3 of the vertices are involved in connecting to the rest of the mesh, and that the 4th vertex is merely a kink - resulting from the flexibility of the filament that locally bends. This means that in reality, that shape although resembling a 4-sided polygon, is structurally a triangle. See example below, maybe it makes it more clear:
Better image to illustrate.