Bayesian theory and artificial neural network approach in MEG inverse problem

Neuromagnetic source image reconstruction in a MEG inverse problem draws on a wide range of signal and image processing techniques. To address this ill-posed problem, we constitute a unifying and powerful theoretical stochastic regularization framework. Maximum posteriori (MP) Bayesian estimation enables the combination of different information to obtain the solution, and the main interest of the Markov random field (MRF) model lies in the consideration of local interactions at the local, energies. We propose a continuous prior model with binary line processes. Additionally, the MEG linear inverse problem being reformulated as an optimization, problem acquires a global minimum solution from the posterior energy function, many optimization methods (such as GA, SA, MFA etc.) are investigated in the literature. In order to obtain a physically plausible source image, we adopt a coupled gradient neural network approach. Finally, simulation results are presented.