Improving the diagnostic power of otoacoustic emissions. Arturo Moleti Physics Department University of Roma Tor Vergata

Transcription

1 Improving the diagnostic power of otoacoustic emissions Arturo Moleti Physics Department University of Roma Tor Vergata

2 The human ear Ear canal: resonant cavity Middle ear: impedance adapter and pressure amplifier Cochlea: high-sensitivity acoustical detector and spectrometer, with excellent frequency resolution, high gain, huge dynamic range

3 The human cochlea

4 The human cochlea Cochlear functions Frequency selectivity from BM tonotopicity Nonlinear active amplification by outer hair cell (OHC) electro-mechanical feedback Mechano-electrical transduction by inner ear cells (IHCs)

5 Cochlear Mechanics Analogue of an electrical nonlinear transmission line (P V and d /dt I)

6 Cochlear Mechanics We couple the equations of an incompressible fluid to that of a tonotopically resonant elastic membrane, getting a full cochlear model:

7 Cochlear Mechanics In the linear approximation (WKB) these equations lead to forward and backward propagation of slow (10m/s) transverse waves Each frequency component of the forward wave is strongly amplified approaching the resonant place, where perception (and power absorption) occurs.

8 Scale Invariance The physics is the same at all frequency scales, in the corresponding, longitudinally shifted, cochlear region Consequences: the TW phase lag from the base to the resonant place is independent of frequency the TW phase lag from the base to a fixed place is proportional to frequency the transmission time to the resonant place is inversely proportional to frequency

9 Nonlinear active models Schematization of the nonlinear active cochlear amplifier (anti-damping case) The OHC mechanism acts as an active filter that both amplifies the peak response and reduces the bandwidth. This effect is reduced with increasing stimulus levels

10 Nonlinear active models The BM response to a pure tone is peaked near the resonant place. The growth is nonlinear in the peak region The growth is linear far from the resonance

11 Nonlinearity and distortion BM nonlinearity also generates additional forward and backward waves Among them, intermodulation distortion products provide a useful diagnostic tool because they are generated only inside the cochlea The most important one is the intermodulation cubic distortion product 2f 1 -f 2, eventually producing additional acoustic perception (Tartini's third sound)

12 Backward waves and OAEs The cochlea is designed to transmit forward the acoustic energy towards the resonant place, but it partly reflects it towards the cochlear base due to random fluctuations of the mechanical impedance (roughness) The backward waves are transmitted back through the middle ear and eventually measured in the ear canal as otoacoustic emissions (OAEs) They are a by-product of the cochlear activity, sensitive to the level of the local BM displacement

13 OAE acquisition techniques TEOAE evoked by transient stimulus (click or TB) DPOAE evoked by two tones f 1 and f 2 at their main intermodulation distortion frequency 2f 1 -f 2 SFOAE evoked by a single tone at the same frequency

14 OAE-based diagnostics TEOAEs are routinely used as a newborn hearing screening test TEOAE, DPOAE and SFOAE levels show high correlation with audiometric threshold levels in cross-section studies on large populations Large inter-subject variability of the OAE levels prevents their use as a powerful diagnostic test on the single subject Part of this variability is related to the complexity of cochlear mechanics

15 OAE generation mechanisms Nonlinear distortion (wave-fixed) Scale Invariance implies flat OAE phase-frequency relation Reflection from cochlear roughness (place-fixed) (from different places) SI implies that OAE phase-gradient delay is proportional to the round trip transmission time to the reflection place (1/f, and longer for places closer to the resonance). BM nonlinearity implies that I/O functions are different for OAEs reflected from the resonant place (strong, compressive gain) and from more basal places (lower, close to linear gain).

16 OAE time-frequency analysis The wavelet transform provides a representation of the OAE response in the time-frequency domain, where cochlear (approximate) scaling symmetry manifests itself The t-curves show an isophysics coordinate, along which the physics of the OAE generation is the same across frequency. Along the arrows the physics changes

17 SFOAE time-frequency filtering Selecting the wavelet components between two t- curves, one picks up the component of the OAE response due to a specific physical mechanism, or generation place In the SFOAE and TEOAE cases, one can identify reflections from the peak region and from more basal regions, as well as double intracochlear reflections

18 DPOAE time-frequency filtering In the DPOAE case, t-f filtering separates the zerolatency component, associated with the backward DP wave, from the long-latency component, associated with reflection from the tonotopic region. This improves diagnostic accuracy, because: 1) interferences are cancelled, 2) SNR is improved (noise is uniformly distributed in the time-frequency plane).

19 Time-frequency filtering The growth-rates of the different OAE components separated by t-f filtering are different, in agreement with theory. This observation proves that filtering according to latency is capable of correctly disentangling the different generators

20 Conclusions Cochlear Mechanics tells us how acoustic waves are transmitted, amplified and backscattered in the cochlea Cochlear scale-invariance accurately predicts the phase-gradient delay of OAE components generated by different mechanisms (or by the same mechanism at different places) Time-frequency filtering allows us to discriminate different OAE components according to their different phase-gradient delays, eventually improving the OAE diagnostic capability