Objective detection of transiently evoked otoacoustic emissions

Newborn hearing screening with transiently evoked otoacoustic emissions (TEOAE) is a well-established method. A screening device must be equipped with a test procedure for objective TEOAE detection. The statistical tests implemented in the commercially available screening devices are the correlation, an estimation of the signal-to-noise ratio and a binominal test. The aim of the present study is to compare the TEOAE detection performance of these tests with that of several other tests in the time and frequency domains (variance ratio F(SP) and its modification F(SP)*, Friedman test, modified q-sample uniform scores test). The comparison was based on a data sample of 420 TEOAE. The frequency range examined was 1.5-4.0 kHz. As a new feature, two frequency sub-ranges (1.5-2.5 kHz, 2.5-4.0 kHz) were tested separately. The modified variance ratio F(SP)* was the most powerful test, whereas the tests implemented in the known screening devices showed the lowest detection performance.