Joint Bayesian estimation of brain regional activity and network functional connectivity

Robust estimation of brain regional activity is vital for the studying of the neural mechanism and network functional connectivity of our brain. Most of the existing research are using Beamformer for the estimation of the brain regional activity, then reconstructing brain networks, which is unsuitable for the correlated sources estimation. This paper presents a novel Bayesian based algorithm for the joint estimation of brain regional activity and network functional connectivity from MEG data. The novel algorithm is not only able to accurately distinguish activities of different brain regions in scenarios with high correlation among active brain regions, but also to completely reconstruct all brain region source activities in scenarios with a high number of active brain regions. Simulations and real data set are used to evaluate the performance of the proposed new algorithm compared with the widely used conventional algorithms, results show superiority to standard benchmark algorithms, which improves the potential application of electromagnetic source imaging in real clinical world.