An inversion method based on random sampling for real-time MEG neuroimaging
The MagnetoEncephaloGraphy (MEG) is a non-invasive neuroimaging technique with a high temporal resolution which can be successfully used in real-time applications, such as brain-computer interface training or neurofeedback rehabilitation.
The localization of the active area of the brain from MEG data results in a highly ill-posed and ill-conditioned inverse problem that requires fast and efficient inversion methods to be solved. In this paper we use an inversion method based on random spatial sampling to solve the MEG inverse problem. The method is fast, efficient and has a low computational load. The numerical tests show that the method can produce accurate map of the electric activity inside the brain even in case of deep neural sources.