Extraction of Functional Brain Networks from EEG Signals in the Context of Visual Perception

A wide range of studies treats the brain as a structural and functional network using complex algorithms and features to characterize the network. However, less effort is invested into devising new and more robust ways of extracting the networks in the first place. Because network extraction is the first step of network analysis, this has tremendous importance, determining all subsequent results. This article addresses a new method through which the resulting functional networks are less dense but still relevant, in the context of visual perception tasks and perhaps other ones. The proposed method uses interquartile range (IQR) statistics to filter out irrelevant connections between cerebral areas. In addition, it relies on the Scaled Correlation Function, which enables the estimation of correlations on particular timescales. Our results indicate that this technique can yield functional networks that are more informative, capturing relevant brain dynamics correlated to perceptual and cognitive processes.