Simulation of Event-Related Potentials in a Wide Sense for Assessment of Detection∕Estimation Algorithms

Abstract

Event-related potentials in a wide sense (ERPWS) are bioelectric signals that appear associated to certain stimuli or event. The most common way to detect and estimate ERPWS is based on coherent averaging, but alternative techniques have been proposed to overcome certain limitations of the former. A flexible and realistic simulator is designed here to assess, in a controlled scenario, algorithms used to detect/estimate ERPWS. The system generates signals resembling real-world ERPWS, by adjusting waveforms, jitter, width and amplitude modulation, sampling/quantization effects, noise and interference. Having these simulated records, we can apply the algorithm under assessment, and easily estimate parameters as the variance and the bias of the remnant noise, as well as the signal-tonoise ratio. With those parameters as a reference, algorithms to detect/estimate ERPWS can be objectively compared.