Periodogram based tests for distortion product otoacoustic emissions

Distortion product otoacoustic emissions (DPOAEs) are an important nonbehavioral measure of cochlear function, which provides a close analogue of the behavioral pure-tone audiogram. DPOAEs are sinusoidal distortion products (DPs) produced by nonlinearities in the healthy cochlea. Detection of DPs is accomplished in the Fourier domain with a periodogram based test. The test compares the power in the DP periodogram bin to a noise estimate derived from a certain number of the surrounding bins. Statistical properties of this test to date have only been examined by constructing receiver operator characteristics curves derived from DPOAE measurements in normal and hearing impaired individuals. In this paper the null distribution of this order-statistic based test is explicitly derived, and via simulations intended to mimic the nonwhite features of real-ear noise measurements, the power of the test is demonstrated. These simulations demonstrate that a local F test is more powerful than this DPOAE test, with critical values that are easier to calculate. Although the power of both tests increase with an increasing number of bins, the improvement is negligible at around four bins. Since the power of both tests decrease at lower DP frequencies, it is not recommended to use a large number of bins.