Joint time-frequency analysis and filtering of single trial event-related potentials.

The ongoing electrical activity of the brain is known as the electroencephalograph (EEG). Event related potentials (ERPs) are voltage deviations in the EEG elicited in association with stimuli. Their elicitation require cognitive processes such as response to a recognised stimulus. ERPs therefore provide clinical information by allowing an insight into neurological processes. The amplitude of an event-related potential is typically several times less than the background EEG. The background EEG has the effect of obscuring the ERP and therefore appropriate signal processing is required for its recovery. Traditionally ERPs are estimated using the synchronised averaging of several single trials or sweeps. This inhibits investigation of any trial-to-trial variation, which can prove valuable in understanding cognitive processes. An aim of this study was to develop wavelet-based techniques for the recovery of single trial ERPs from background EEG. A novel wavelet-based adaptive digital filtering method for ERPs has been developed. The method provides the ability to effectively estimate or recover single ERPs. The effectiveness of the method has been quantitatively evaluated and compared with other methods of ERP estimation.The ability to recover single sweep ERPs allowed the investigation of characteristics that are not possible using the conventional averaged estimation. The development of features of a cognitive ERP known as the contingent negative variation over a number of trials was investigated. The trend in variation enabled the identification of schizophrenic subjects using artificial intelligence methods.A new technique to investigate the phase dynamics of ERPs was developed. This was successfully applied, along with other techniques, to the investigation of independent component analysis (ICA) component activations in a visual spatial attention task. Two components with scalp projections that suggested that they may be sources within the visual cortex were investigated. The study showed that the two components were visual field selective and that their activation was both amplitude and phase modulated.