Fig. 2

Flowchart of the salient skeleton construction. On the left the procedure to get the high salient skeletons is reported. It consists of different steps: scale free networks with power-law distributed weights were extracted from initial network of each subject, then links participating at least once in the shortest paths, starting from a fixed reference node, were recorded in the shortest path tree matrix, finally shortest path tree for each reference node were added up to obtain, for each subject, salience matrix, whose values \(s_{i,j}=1\) represent high salient skeleton. On the right, a small-world and scale-free network is represented with the corresponding high salient network