Fig. 3

Simplified illustration of data generation and analyses of simulated MEG and human MEG. A computational brain model comprised of 78 neural masses coupled according to human DTI-derived binary structural connectivity matrix [31] was used to simulate whole-brain MEG-like oscillations. The simulated oscillations are derived from fluctuations in average membrane membrane potential of the excitatory neurons of each neural mass. We investigated the effect of different AD-driven neuronal function changes (see Fig. 2) on network oscillations and spike density of the pyramidal neuron populations. Fast Fourier Transformation was applied to simulated and human MEG data for spectral analysis. Human resting-state eyes-closed MEG was available for 18 prodromal AD patients (amyloid positive patients diagnosed with mild cognitive impairment) as well as 18 age- and gender-matched elderly control subjects