Polynomial chaos and Bayesian inference in RPDE's: a biomedical application
Rob H. De Staelen, Karim Beddek, Tineke Goessens · Ghent University Academic Bibliography (Ghent University) · 2011
The electroencephalograph (EEG) is one of the most influential tools in the diagnosis of epilepsy and seizures. It measures electrical discharges of neurons in the human brain. The latter consists of many regions, all with a different electrical conductivity. Unfortunately one cannot measure this non invasively, e.g. preoperatively. In this paper, we investigate the uncertainty induced on the location of EEG current dipoles. A Bayesian framework is used, so as to include modeling error and noise, but combined with Polynomial Chaos expansions to represent random variables, speeding up computations. We evaluate this technique on a spherical head model with a standard clinical 27 sensor positioning.