Auditory-filter based compression estimates for extended clinical diagnosis of hearing loss

Michal Fereczkowski

One of the most prominent properties of a healthy cochlea is its compressive behavior, which has two important functional consequences. First, it increases the dynamic range of levels the auditory system can process. The second is an increase in the frequency selectivity. The compression exponent, is a significant predictor of speech intelligibility in speech-shaped noise and the exponent’s value cannot be predicted from the audiogram. Hence a time-efficient measure of cochlear compression could be of value for clinical hearing diagnostics.

This project investigates the extent to which the residual compression can be predicted from behavioral estimates of auditory-filter characteristics (e.g., bandwidths, tip-to-tail gains) obtained using a clinically-feasible diagnostic test. A recently developed threshold-tracking method, the Grid method, which allows for estimation of an auditory filter shape in 5 minutes or less will be used for data collection from a large number of hearing impaired listeners at the Bispebjerg Hospital clinic. Additionally, compression estimates from an objective measure, based on distortion-product otoacoustic emissions (DPOAE), will be investigated in the same listeners. Recent developments in otoacoustic emissions (OAE) signal-processing (e.g., source-unmixing) will be incorporated, to improve the accuracy of individuals’ estimates. The clinical utility of both tests will be validated through cross-correlation and post hoc comparison of hearing-aid fitting and satisfaction.

An overall objective of the project is to provide clinically-feasible tools for characterization of individual compression loss, which then might lead to better modelling of individual hearing loss and, consequently, improved hearing-aid fitting.
This project is funded by the Oticon Foundation