Ensemble kalman filter for state estimation of brain activity by considering a large scale nonlinear dynamical model

Abstract

Data assimilation techniques, like the Ensemble Kalman filter (EnKF), have been successfully used for weather forecasting or in general for state space estimation tasks that involve large scale nonlinear complex dynamical models. In this paper a novel application of the EnKF is presented for estimation of neural activity into the brain considering a large scale time-varying nonlinear model to describe its dynamics. A comparative analysis is performed in terms of brain activity reconstruction and time-course reconstruction. As a result, the EnKF outperforms methods like the Multiple Sparse Priors (MSP) and Low Resolution Tomography (LORETA).