Joint variational bayesian extended Kalman filter for the estimation of the metabolic/hemodynamic model

摘要

This paper deals with a state space approach based on metabolic hemodynamic model (MHM) for Blood Oxygenation Level Dependent (BOLD) signal measured using functional Magnetic Resonance Imaging (fMRI). The proposed approach is based on the extended Kalman filter to infer the hidden states and parameters. Furthermore The extended kalman filter is enhanced by variational Bayesian approach for the estimation of the measurement noise.