Fully exploratory network ICA (FENICA) on resting-state fMRI data

Independent component analysis (ICA) is one of the most valuable explorative methods for analyzing resting-state networks (RSNs) in fMRI, representing a data-driven approach that enables decomposition of high-dimensional data into discrete components. Extensions to a group-level suffer from the drawback of evaluating single-subject resting-state components of interest either using a predefined spatial template or via visual inspection. FENICA introduced in the context of group ICA methods is based solely on spatially consistency across subjects directly reflecting similar networks. Therefore, group data can be processed without further visual inspection of the single-subject components or the definition of a template (Schöpf et al., 2009).In this study FENICA was applied to fMRI resting-state data from 28 healthy subjects resulting in eight group RSNs. These RSNs resemble the spatial patterns of the following previously described networks: (1) visual network, (2) default mode network, (3) sensorimotor network, (4) dorsolateral prefrontal network, (5) temporal prefrontal network, (6) basal ganglia network, (7) auditory processing network, and (8) working memory network. This novel analysis approach for identifying spatially consistent networks across a group of subjects does not require manual or template-based selection of single-subject components and, therefore, offers a truly explorative procedure of assessing RSNs.

Research highlights▶ Group analysis of resting-state data. ▶ FENICA identifies consistent and corherent group networks. ▶ FENICA does not need a priori template definition, visual inspection nor single-subject component selection.