Improving the usability of the distortion product otoacoustic emisssions-sweep method: An alternative artifact rejection and noise-floor estimation

Abstract

The DPOAE-sweep method combined with a specific least-squares-fit (LSF) analysis provides a fast method to measure and analyze distortion product otoacoustic emissions (DPOAE) with a high frequency resolution. In studies using this technique the noise reduction, artifact rejection and noise estimation are typically realized in a “classical” way, i.e., as temporal averaging with preceding elimination of time epochs exceeding an artifact threshold level and a noise estimation based upon the analysis of the difference of two buffers of epoch averages. However, the choice of an artifact threshold is arbitrary to some extent and different choices can lead to differences in the estimation of DPOAE levels. The two-buffer technique is ambiguous as well since a different grouping of the epochs into the buffers leads to rather different noise estimates. Therefore, the present study proposes an alternative approach, which provides unique noise estimators for a given set of data, a robust artifact rejection without the need to select an arbitrary rejection threshold, as well as estimators including confidence intervals for the DPOAE levels and phases. The “classical” and suggested estimators for DPOAE levels, DPOAE phases, and noise levels are compared based on Monte Carlo simulations and real measured data sets.