Wavelet based denoising of otoacoustic emissions

Abstract

We propose a technique to improve the ability of transient evoked otoacoustic emission (TEOAEs) to separate normal hearing and hearing impaired subjects. The technique combines signal decomposition in three octave bands by wavelet transform, denoising, separate reconstruction of components representing octave bands and windowing in time. As pass/fail criterion we have used cross correlation coefficient between processed signal realizations. The performance of the proposed technique is evaluated on a database of 5214 records by means of receiver operating characteristics curves.