Resting-state FMRI: canonical networks in normal children
Independent Component Analysis of resting-state FMRI studies allows to extract a number of neural networks form task-less studies. There is a paucity of reports on normal networks and their variants found in normal children. We describe in this study the characterization of neural networks found in a group of normal children.Methods: 40 datasets of normal children who underwent rs-fMRI were analyzed with ICA utilizing MELODICA from FSL library. For each subject, the independent components were classified between neural networks and non-neural networks based on a heuristic approach and performed by an expert on the field. Further characterization of the neural networks was accomplished based on localization of maxima, frequency of oscillation, and profile of oscillation. Frequency of network yield across subjects was assessed along frequency of the network oscillation and main variants.Results: 24 distinct neural networks were found. Oscillation frequency ranged from 0.0168 to 0.072 Hz. Main pattern variants consist of network-merging and iterations of the same network at different frequencies. The most frequently found networks across the subjects were the right executive network (97.5%), the precuneus (82%) and visual (77.5%). The networks less frequently found were the hippocampus (7.5%) and the amygdala (10%).