Implementation and evaluation of functional connectivity measures -­ a resting state fMRI study of cognitively impaired subjects

Abstract: Resting State Functional Magnetic Resonance Imaging (rs-fMRI) is often used to map brain networks or Resting State Networks (RSNs). The networks are estimated by calculating how similarly certain spatially distinct brain regions behave. These networks have been shown to dier in individuals with many dierent kinds of cognitive diseases. The purpose of this study was to evaluate the capability of dierent network characteri- zation measures to distinguish rs-fMRI data from subjects with Mild Cognitive Impairment (MCI) from that of healthy controls. The measures evaluated were synchronization likelihood, mean phase coherence as well as Pearson correlation. Four cohorts were delineated using CerebroSpinal Fluid (CSF) biomarkers and scoring in a word recollection assignment (ADAS-3). The networks of each cohort was compared to the same healthy aged matched controls. Both synchronization likelihood and mean phase coherence was found to have lower statistical power to dierentiate groups than correlation. However, the only measure capable of capturing the complex, non-linear dynamics of functional networks, synchronization likelihood, exhibits a pattern in reduced connectivity linked to perceived MCI progression.