Estimation of the ischemic brain temperature with the particle filter method
In this work, a two-dimensional model was developed to analyze the transient temperature distribution in the head of a newborn, during local cooling promoted by the flow of cold water through a cap. The inverse problem dealt with the sequential estimation of the internal temperature of the head, by using non-invasive transient temperature measurements. A state estimation problem was solved with the Sampling Importance Resampling (SIR) algorithm of the Particle Filter method. Uncertainties in the evolution and observation models were assumed as additive, Gaussian, uncorrelated and with zero means. The uncertainties for the evolution model were obtained from Monte Carlo simulations, based on the uncertainties of the model parameters. The head temperature was accurately predicted with the Particle Filter method. Such a technique might be applied in the future to monitor the brain temperature of newborns and control the local cooling treatment of neonatal hypoxic-ischemic encephalopathy.