To reduce the noise level, binary networks are often constructed from weighted connectivity matrices. For this purpose, thresholding of the correlation matrices is utilized. If the correlation weight between two nodes is greater than a certain threshold, this represents a relation between the nodes. A certain threshold value is used to threshold the weighted correlation matrices [51]. In the sparsity thresholding method, all extracted networks have the same density; thus, the comparison is unbiased in relation to density.
