Single-trial ERP estimation based on spatio-temporal filtering

This paper introduces a new approach to the problem of single-trial event-related potentials (ERP) estimation in EEG data with very low signal-to-noise (SNR) ratio. The method relies on modeling of the ERP component descriptors to be estimated and utilizes a spatial filter to constrain the output in the temporal domain. Results on cognitive EEG data with very low SNR show that our method successfully extracts the P300 component while other traditional methods fail. Some possible improvements on the present method are also discussed.