Comparison of alternative methods in neonatal hearing screening: tone-burst otoacoustic emissions and time-frequency filtering

Abstract

INTRODUCTION: The click-evoked otoacoustic emission (CEOAE) technique has a drawback of high referral rate in first time neonatal hearing screening and an associated high false positive rate. Among several factors, noise has been highlighted as the major contributor to false positive results in OAE screening programmes, particularly affecting lower frequency CEOAE recordings. Two alternative strategies¡ªa combined CEOAE plus 1 kHz tone burst evoked OAE (TBOAE) response and a time-frequency filtering technique¡ªhave been developed to improve CEOAE detection rates in the lower frequency ranges, and to concurrently reduce the referral rates in initial CEOAE neonatal hearing screening. OBJECTIVES: The aim of the present study was to compare the above two methods in a large group of neonates and explore their feasibility in a universal hearing screening program. METHODOLOGY: A total of 1079 neonates (2158 ears) from well-baby nurseries were enrolled. Both CEOAEs and TBOAEs were recorded in the nonlinear response mode. CEOAE measurement used the default ¡°QuickScreen¡± acquisition mode of Echoport ILO 292 USB system (Otodynamics Ltd., UK). A tone burst stimulus with a 1 kHz centre frequency was used to elicit a TBOAE response for each neonate. A SNR ¡Ý 3 dB was used to define a clear response at each frequency band. All raw CEOAE data obtained by ILO V6 software were offline transferred into MATLAB programmes which were developed in-house for the further time-frequency filtering. The principal steps for signal processing were to (1) map the original CEOAE recordings on the t-f domain by continuous wavelet transform (CWT), (2) localize the specific area of lower frequency noise on the t-f domain and reduce noise using a 2D t-f filter, then (3) reconstruct the processed signals using an inverse continuous wavelet transform (ICWT). Parents were informed of the results based on the CEOAE-only screening outcomes and referred babies were rescreened within one month of initial screening. RESULTS: Comparing the two alternative protocols, results showed an advantage in using the t-f filtering approach over the 1 kHz TBOAE/CEOAE approach for generating better SNRs, greater pass rates, and lower noise levels at 1 and 1.4 kHz frequency bands (p < .05). In terms of the signal levels at the above two frequency bands, TBOAE responses were significantly greater than in comparable frequency regions for the denoised CEOAE response with the t-f denoising method (p < .05). CONCLUSION: This study compared the effectiveness of using a combined 1 kHz TBOAE/CEOAE protocol as well as a newly developed signal processing algorithm to improve OAE detection at lower frequency bands, and hence to reduce the number of refer cases in initial OAE screening. Both approaches achieved the research aims, with the t-f denoising approach being better than the combined TBOAE/CEOAE method for improving overall pass rate and OAE detection at low frequency bands.