Acquisition of Otoacoustic Emissions Using Swept-Tone Techniques

Project Enquiry:

Fields with * are mandatory

ABSTRACT

Otoacoustic emissions  (OAEs) have been under investigation since their discovery 30 years ago (Kemp,  1978). Otoacoustic emissions are quiet sounds generated within the cochlea that can be detected with a  sensitive microphone placed within the ear canal. They are used clinically as a hearing screening tool but have the potential for diagnostic and monitoring purposes.

For this dissertation, high-resolution instrumentation was developed for improving the acquisition of OAEs. It was shown that a high bit-depth device is required in order to simultaneously characterize the ear  canal and the cochlear responses. This led to a reduction in the stimulus artifact that revealed early  latency, high-frequency otoacoustic emissions.

Next, a swept-tone technique originally developed for use in acoustical systems was formally developed for  use in the human ear. The swept-tone technique allows for the simultaneous acquisition of a system’s  impulse response and its distortion components. The swept-tone was first used in this study to characterize the ear canal transfer properties. From that transfer function, a compensation routine was developed  which equalized the magnitude and phase distortions of the ear canal.

As a result, an improved acoustical click could be presented to the ear, which allowed for further reduction of the stimulus  artifact,  revealing  early latency emissions. Spectral flatness and effective duration measurements of the compensated click showed an improvement over traditional click stimuli. Furthermore, wavelet analysis and time-frequency latency computations showed that higher frequency otoacoustic emissions were recoverable when using a compensated click stimulus.

The swept-tone technique was then utilized for the direct acquisition of otoacoustic emissions. The swept-tone response was compressed to an impulse response and  compared to a standard click response. It  was  found  that  several  similarities  exist  between the two response types.  The divergences, primarily in the low-frequencies, have implications in the generation mechanisms involved in a click-evoked otoacoustic  emission.

The swept-tone response provided some clinical benefits, namely in an improved signal-to-noise ratio, and in the removal of obstructive synchronized spontaneous OAEs when compared to a standard click response. Current methods are restricted by noise contamination, and the use of a swept-tone technique can reduce the acquisition time by up to a factor of four, compared to standard click methods. These implications and future potential studies are discussed.

METHODS AND MATERIALS

Instrumentation:

A custom biomedical device was built at the Neurosensory Lab for high-fidelity acquisition of OAEs. The  device interfaces with an Etymōtic Research ER-10D OAE probe and to MATLAB via UART (universal asynchronous receiver/transmitter) communication and a custom software program. The device is capable of acquiring a  large dynamic range at a high sampling rate using an Analog Devices ADSP-21369 EZ-Kit Lite.

A High-resolution (24 Bit, 48 Khz) Oae Acquisition Device Was Built. This Board Interfaces With Custom Software, Allowing for Adaptable Functionality and Real-time Communications With a PC.

A High-resolution (24 Bit, 48 Khz) Oae Acquisition Device Was Built. This Board Interfaces With Custom Software, Allowing for Adaptable Functionality and Real-time Communications With a PC.

Swept-tone Method:

A novel impulse response measurement technique was developed for the characterization of weakly onlinear,  approximately time-invariant systems (Farina, 2007). The method uses an exponentially-swept sinusoid, and was originally developed for the measurement of audio systems, such as loudspeakers and room acoustical environments. It was developed as an alternative to traditional impulse response, MLS, and time-stretched  pulse (TSP) methods, which require strict linearity and time-invariance.

Time-frequency Analysis:

The wavelet transform allows for a relatively short signal to be decomposed into its elementary frequency  constituents with high frequency and time resolution. The selection of a mother wavelet is important in the subsequent analysis. For these experiments, a Morlet mother wavelet was chosen. The mother wavelet  is scaled and shifted to the proper analysis frequency and latency, which in this study ranged from 0.8 to 8.0 kHz in 0.05 kHz steps from 0 to 20 ms post stimulus onset.

This Figure Shows a Wavelet Decomposition of a Synthesized Teoae Signal. Note That High-frequency Components Occur in the Early Latencies, and Low-frequency Components Occur in the Late Latencies. From (Wit Et Al., 1994).

This Figure Shows a Wavelet Decomposition of a Synthesized Teoae Signal. Note That High-frequency Components Occur in the Early Latencies, and Low-frequency Components Occur in the Late Latencies. From (Wit Et Al., 1994).

COMPENSATION FOR THE MEATUS

In acquiring TEOAEs, as mentioned previously, electrical rectangular pulses (henceforth referred to as  clicks) are fed to an acoustical receiver, or speaker. This speaker resides within a probe assembly, which  is also coupled with a microphone for recovery of the response from within the outer auditory canal. However, the transfer function of the speaker, the probe assembly, the ear canal, and the microphone all conspire to distort the click, and what results acoustically within the ear canal is actually far from a click. One such distortion is the acoustical ringing of the  meatus. If thought of as a single open-ended Helmholtz resonator 27 mm in length, the meatus will have a half-wave resonance frequency around 3 kHz. The acoustic ringing primarily occurs at this resonant frequency.

Theory and Method:

The objective is to design an equalization filter whose magnitude and phase responses will compensate that  of any meatal response. Given a recorded ear response h(t), the corresponding frequency-domain transfer  function will be H(f).

After Obtaining the Impulse Response of the Ear via the Swept-tone Technique, Time Windowing Is Performed to Obtain the Mr Signal, H[n].

After Obtaining the Impulse Response of the Ear via the Swept-tone Technique, Time Windowing Is Performed to Obtain the Mr Signal, H[n].

Results:

Electrical and acoustical characteristics of the standard and compensated clicks are characterized as  recorded from one subject (SUB01R). The rectangular click waveform displayed in the top row shows the   expected slowly decreasing magnitude response with flat phase. The corresponding acoustic waveform plots  in the second row display the acoustic ringing (the meatal response) resulting from the ear canal characteristics.

Time and Frequency Characteristics of the Signals of the Oae System: (Left Column) Time-domain Signals, (Middle Column) Corresponding Magnitude Response, and (Right Column) Phase Response.

Time and Frequency Characteristics of the Signals of the Oae System: (Left Column) Time-domain Signals, (Middle Column) Corresponding Magnitude Response, and (Right Column) Phase Response.

SWEPT-TONE TEOAE

In this experiment, an interpretation of SFOAEs is explored in which TEOAE-like responses are derived from a swept-tone single frequency stimulus. The response can be considered as an SFOAE because, although the stimulus frequency is constantly rising, at any instantaneous moment there is only a single pure tone being presented to the subject.  However, the swept-tone response can be presented as a response to an impulse through some  post-processing  techniques. This derived impulse response closely resembles a click-evoked  OAE in phase and time-frequency characteristics. Through this sort of analysis, direct comparisons between SFOAE and TEOAE responses can be made.

The Stimulus, S[n] Is Created According to Eq. (2.26). the Inverse Sweep, S-1[n], Is Generated by a Time-reversal and a -10 Db Per Decade Envelope.

The Stimulus, S[n] Is Created According to Eq. (2.26). the Inverse Sweep, S-1[n], Is Generated by a Time-reversal and a -10 Db Per Decade Envelope.

FUTURE IMPROVEMENTS

While this dissertation discusses a novel method for acquiring OAEs, it perhaps raises more questions than it answers. Some of these questions include: What would extended nonlinear responses (i.e.,longer  than 5-10 ms) tell us, especially with regard to the late latency, low frequency regions? Are spontaneous  emissions or intermodulation distortions the primary source of low-frequency energy in click responses,  and can a modified swept-tone study help determine this? Can the swept-tone method be extended to  operate under DPOAE modality? Can the swept-tone be compensated for flat envelope inside the auditory canal? Can  the swept-tone method be extended to operate under bone conduction modality?.

Source: University of Miami
Author: Christopher Lee Bennett

Download Project

Project Enquiry:

Fields with * are mandatory

Leave a Comment

Your email address will not be published. Required fields are marked *