Evoked Potential Estimation Using A Normalized Stochastic Gradient Lattice Joint Process Estimator
JAN NICOLAS RAINA, GERMAN GONZALEZ, MARIA CRISTI STEFANELLI, Trina Adrián de Pérez, F. D'Alvano · Journal of Clinical Engineering · 1992
When a sensory stimulus is applied to a subject, a small signal, the evoked potential response (EP), appears buried in the electroencephalogram (EEG). EPs are extensively used in many clinical applications. To extract the small evoked signal several techniques have been developed. The objective of this work was to improve the EP estimate through the use of an adaptive lattice filter noise canceller with the Normalized Stochastic Gradient adaptation algorithm. The reference signal is a filtered periodic extension of an average EP, obtained a priori from the primary channel signal. The required signals were generated using real evoked responses and EEGs, taken from normal subjects. Tests showed that this technique provides, with fewer stimuli, better individual estimates than the single one produced by the averaging technique over 100 EPs.