Optimized robust averaging of event-related potentials

Averaging is one of the most commonly used procedures in Neuroscience and is the basic procedure for obtaining event-related potentials (ERP). Averaging or the estimation of the location of data can be done in different ways, although only the arithmetic mean is routinely used in the current practice of ERP research. In this paper, a number of location estimators (mean, median, trimmed mean, Winsorised mean and recently introduced trimmed L-mean) are reviewed. A new robust location estimator tanh, which allows the data-dependent optimization, is proposed for averaging of small number of trials. The performance of the location estimators, expressed by the signal-to-noise ratio (SNR) of averaged waveforms, is compared for different number of epochs randomly drawn from a set of real auditory evoked potential (EP) data. Procedure of multi-averaging is developed to optimize the choice of the location estimator with respect to SNR or other criterion of the performance. Experiments with real auditory EPs demonstrate significant achievable gain compared to the mean averaging. The procedure of multi-averaging can be applied to a wide range of experimental data, when an objective criterion of the quality of the averaged data can be calculated