Title : Bayesian tracking of neural activity in biomagnetic data
Magnetoencephalography (MEG) is a non-invasive brain imaging technique measuring the weak magnetic field due to neural activity. The estimation of the neural activity from this data involve the solution of a highly ill-posed inverse problem.
In this talk, I will present a probabilistic approach for the solution of this problem: a particle filter is implemented to realize a Bayesian tracking of the neural sources in a dipolar model framework. The effects of different cortical constraints on this method are also investigated.