Table 1. Overview of different approaches for disentangling filamentous networks.

Input	Method	Features					References
		curved filaments	filament-specific	intensity-based	automated	parsimonious
image	texture filter	−	−	+	+	+	30
	linear programming	−	−	+	+	+	31
	rotating grid	−	+	+	+	+	32
	filament tracing	+	+	+	○	○	33, 34, 35
	filament tracking	+	+	+	+	○	36
	open contours	+	+	+	+	○	23,37
network	rule-based decomp.	+	+	−	○	○/−	38,41
	filament cover	+	+	+	+	○/+	current work

Two main classes of approaches to analyse the filamentous structure of networks can be distinguished, based on whether they operate on image data or on extracted networks. Irrespective of the class, the existing approaches vary in their capacity (+) or inability (−) to detect curved filaments, identify individual filaments, and to include information about the intensity/thickness of filaments. Further, the amount of manual user input as well as the number of parameters required by the algorithms can be feasible (+), laborious (−), or depends on the specific variant of the algorithm (○). For the network-based approaches, the number of required parameters may be different for the extraction of the network from image data and the consequent decomposition of the network into filaments (separated here by/).