C. Bertrand et al., A probabilistic solution to the MEG inverse problem via MCMC methods: The reversible jump and parallel tempering algorithms, IEEE BIOMED, 48(5), 2001, pp. 533-542
We investigated the usefulness of probabilistic Markov chain Monte Carlo (M
CMC) methods for solving the magnetoencephalography (MEG) inverse problem,
by using an algorithm composed of the combination of two MCMC samplers: Rev
ersible Jump (RJ) and Parallel Tempering (PT), The MEG inverse problem was
formulated in a probabilistic Bayesian approach, and we describe how the RJ
and PT algorithms are fitted to our application. This approach offers bett
er resolution of the MEG inverse problem even when the number of source dip
oles is unknown (RJ), and significant reduction of the probability of erron
eous convergence to local modes (PT), First estimates of the accuracy and r
esolution of our composite algorithm are given from results of simulation s
tudies obtained with an unknown number of sources, and with white and neuro
magnetic noise. In contrast to other approaches, MCMC methods do not just g
ive an estimation of a "single best" solution, but they provide confidence
interval for the source localization, probability distribution for the numb
er of fitted dipoles, and estimation of other almost equally likely solutio
ns.