Recursive Bayesian estimation of single trial evoked potentials

A method for the sequential estimation of single trial evoked potentials (EP) is presented. The method is based on recursive Bayesian Mean Squares estimation. The evoked potentials are estimated sequentially using the old estimates as the prior information. The estimated EPs are constrained to a principal subspace of the ensemble of measurements. The method is shown to be capable of tracking slow trends in parameters of the EP. The performance of the method is evaluated with realistic simulated evoked potential measurements.