Estimating and simulating multiple related functional connectivity networks via the MNS Package

In many neuroimaging studies functional connectivity measurements are collected across a cohort of multiple subjects. Given such data, a fundamental problem corresponds to accurately quantifying variability across subjects. In this vignette we provide a brief illustration of the recently proposed Mixed Neighborhood Selection (MNS) algorithm which is able to simultaneously estimate connectivity networks at the population and subject level as well as quantify inter-subject variability. The MNS package includes parallel implementations the MNS algorithm as well as cross-validation functions; thereby providing computationally efficient methods through which to select regularization parameters and perform model estimation. Moreover, this vignette also introduces an algorithm from which to simulate functional connectivity networks for a cohort of related individuals. It is well documented that functional connectivity displays reproducible activation patterns across subjects while simultaneously exhibiting high inter-subject variability. To our surprise, we found there to be limited algorithms through which to simulate connectivity networks for a cohort of subjects which display these widely accepted properties. To address this, we present a simple and efficient method through which to simulate functional connectivity networks for multiple subjects.