SICE info.txt

If you want to use SICE as a thresholding function during network construction please use computed COVARIANCE matrices
(e.g. generate covariance matrices with the "generate connectivity matrix" function)

Additionally to applying the “more traditional” thresholding methods (i.e. significance, relative, absolute), GraphVar offers pipeline construction
of binary graphs with predefined densities using sparse inverse covariance estimation “SICE”, also known as Gaussian graphical models
or graphical Lasso (Huang et al., 2010).  This method imposes a “sparsity” constraint on the maximum likelihood estimation of an
inverse covariance matrix, which leads to reliable estimation of the inverse covariance matrix with small sample sizes
(i.e., length of the fMRI time series). As compared to partial correlations this method entails the advantage that for reliable
estimations the sample size of the data  does not need to be substantially larger than the number of nodes (i.e., brain regions)
modeled in the network (i.e., the number of time points can be close to or even less than the number of brain regions modeled).
This feature is implemented using “Glasso” (http://statweb.stanford.edu/~tibs/glasso/).

Huang, S., Li, J., Sun, L., Ye, J., Fleisher, A., Wu, T., Chen, K., Reiman, E., 2010. Learning brain connectivity of Alzheimer’s disease by sparse inverse covariance estimation. NeuroImage 50, 935–949. doi:10.1016/j.neuroimage.2009.12.120


*********************SPECIAL THANKS TO Hossein Karshenas: hosseinkarshenas at gmail dot com FOR PROVIDING THE COMPILED VERSIONS OF GLASSO***************
*********************SPECIAL THANKS TO Chao-Gan Yan FOR THIS NICE IDEA***************