HOW BEAMFORMING CAN SOLVE THE BRAIN SOURCE IMAGING PROBLEM?

Localizing and reconstructing the time course of spatially distributed sources from electroencephalographic recordings is a challenging task in beamforming. Indeed, the location of the dipoles constituting a given spatially distributed source must be known in advance to create a filter that will not cancel them out. In this paper, we present a new scheme named SABLE which uses Sparsity And Beamforming for brain source Localisation and Estimation without requiring such a prior. SABLE shows how the two beamforming steps (i.e. filtering and localization) can be merged and extended to deal with spatially distributed sources by modeling the hyper synchronous nature of their activity. Numerical experiments show the good behaviour of the SABLE technique in the context of epilepsy.