Simulating of Magnetoencephalography Signals

Neural brain activities can be non-invasively characterized by analyzing both magneto encephalogram (MEG) and electroencephalogram (EEG) signals. Since the content of MEGs is more informative than EEGs, in this study, modeling of MEG signals is investigated to understand how MEGs are produced. To construct the right magnetic field, the location of magnetic dipoles inside the brain should be determined. In this way, a valid dataset of MEGs belong to different subjects was employed and by applying the inverse method to the MEGs, the location dipoles were detected and then by applying a forward model to the estimated dipoles and using a realistic head model, the simulated MEGs are produced. To see how the simulated signals are similar to the real ones, their power spectrum in different bands and their complexity (Lyapunov exponent) are compared. The comparison results demonstrate enough accuracy between the simulated and the real MEGs in all channels belong to different subjects.