A Novel Approach to Robust Weighted Averaging of Auditory Evoked Potentials

The paper describes the robust weighted averaging method applied to averaging of auditory brainstem responses. This type of signals is characterized with extremely low signal-to-noise ratio. Suppression of noise that contaminates this type of signals can be achieved with the use of the averaging technique. The auditory evoked potentials are timealigned and then the average template is determined. The weighted averaging operation can be regarded as special case of clustering. In this work the averaging process is formulated as the problem of certain criterion function minimization. The maximum likelihood estimator of location based on the generalized Cauchy distribution is used as the measure of dissimilarity function. The proposed methods performance is experimentally evaluated and compared to the reference methods in the presence of the artificial noise and in the case of real signals. The experiments show usefulness of the proposed method for robust weighted averaging of periodic signals, for instance the evoked potentials.