Delineating single subject oscillatory brain networks with Spatio-Spectral Eigenmodes

The spatial and spectral structure of oscillatory networks in the brain provide a readout of the underlying neuronal function. Within and between subject variability in these networks can be highly informative but also poses a considerable analytic challenge. Here, we describe a method that simultaneously estimate spectral and spatial network structure without assumptions about either feature distorting estimation of the other. This enables analyses exploring how variability in the frequency and spatial structure of oscillatory networks might vary both across the brain and across individuals. The method performs a modal decomposition of an autoregressive model to describe the oscillatory signals present within a time-series based on their peak frequency and damping time. Moreover, an alternate mathematical formulation for the system transfer function can be written in terms of these oscillatory modes; describing the spatial topography and network structure of each component. We define a set of Spatio-Spectral Eigenmodes (SSEs) from these parameters to provide a parsimonious description of oscillatory networks. Crucially, the SSEs preserve the rich between-subject variability and are constructed without pre-averaging within specified frequency bands or limiting analyses to single channels or regions. After validating the method on simulated data, we explore the structure of whole brain oscillatory networks in eyes-open resting state MEG data from the Human Connectome Project. We are able to show a wide variability in peak frequency and network structure of alpha oscillations and reveal a distinction between occipital high-frequency alpha and parietal low-frequency alpha. The frequency difference between occipital and parietal alpha components is present within individual participants but is partially masked by larger between subject variability; a 10Hz oscillation may represent the high-frequency occipital component in one participant and the low-frequency parietal component in another. This rich characterisation of individual neural phenotypes has the potential to enhance analyses into the relationship between neural dynamics and a person9s behavioural, cognitive or clinical state