Realistic synthetic background neuronal activity for the analysis of MEG probe configurations

Magnetoencephalography (MEG) sensors are capable of recording the tiny magnetic activity from the brain. They can be constituted of either magnetometers or gradiometers that respectively record the magnetic field or its gradient. In this paper, we present a framework for constructing realistic MEG signals. This framework can be used to test different probe configurations and source localization algorithms. The methodology of generation of synthetic signals is presented, and synthetic signals are compared to real signals. Paroxysmal activity generated with this model and originating from a deep cerebral source is determined with two different localization algorithms. Preliminary results show that gradiometers even with a short baseline perform close to magnetometer and that the use of hybrid systems should be further investigated.