Intrasubject variability in power reflectance

Preliminary results of classifying otosclerosis and disarticulation using a convolutional neural network trained with simulated wideband acoustic immittance data

Current noninvasive methods of clinical practice often do not identify the causes of conductive hearing loss due to pathologic changes in the middle ear with sufficient certainty. Wideband acoustic immittance (WAI) measurement is noninvasive, inexpensive and objective. It is very sensitive to pathologic changes in the middle ear and therefore promising for diagnosis. However, evaluation of the data is difficult because of large interindividual variations. Machine learning methods like Convolutional neural networks (CNN) which might be able to deal with this overlaying pattern require a large amount of labeled measurement data for training and validation. This is difficult to provide given the low prevalence of many middle-ear pathologies. Therefore, this study proposes an approach in which the WAI training data of the CNN are simulated with a finite-element ear model and the Monte-Carlo method. With this approach, virtual populations of normal, otosclerotic, and disarticulated ears were generated, consistent with the averaged data of measured populations and well representing the qualitative characteristics of individuals. The CNN trained with the virtual data achieved for otosclerosis an AUC of 91.1 %, a sensitivity of 85.7 %, and a specificity of 85.2 %. For disarticulation, an AUC of 99.5 %, sensitivity of 100 %, and specificity of 93.1 % was achieved. Furthermore, it was estimated that specificity could potentially be increased to about 99 % in both pathological cases if stapes reflex threshold measurements were used to confirm the diagnosis. Thus, the procedures' performance is comparable to classifiers from other studies trained with real measurement data, and therefore the procedure offers great potential for the diagnosis of rare pathologies or early-stages pathologies. The clinical potential of these preliminary results remains to be evaluated on more measurement data and additional pathologies.

Absorbance Measurements From Normal-hearing Ears in the National Health and Nutrition Examination Survey, 2015-2016 and 2017-2020

OBJECTIVE

To summarize absorbance and impedance angles from normal-hearing ears within the 2015-2016 and 2017-2020 US National Health and Nutrition Examination Surveys (NHANES).

DESIGN

Two publicly available NHANES datasets were analyzed. Ears meeting criteria for normal hearing and valid absorbance and impedance angle measurements were identified. Measurements were summarized via descriptive statistics within categories of age cohort, race/ethnicity cohort, sex (male, female), and ear (left, right).

RESULTS

A total of 7029 ears from 4150 subjects, ages 6 to 80 years, met inclusion criteria. Differences between subgroups within all categories (age, race/ethnicity, sex, and ear) were fractions of the sample SDs. The largest differences occurred between age cohorts younger than 20 years.

CONCLUSIONS

The NHANES absorbance and impedance angle measurements are consistent with published literature. These results demonstrate that trained professionals, using the Titan instrument in a community setting inclusive of all demographics, produce comparable measurements to those in laboratory settings.

Effects of earlens lens placement on sound field thresholds, tympanometric measurements and wideband acoustic immittance

OBJECTIVE

The Earlens is a direct-drive hearing device consisting of a lens which physically displaces the umbo to achieve appropriate gain. The objective is to determine the clinical acceptability of clinical immittance measurements in Earlens wearers.

DESIGN

Controlled before-after within-subjects repeated measures study.

STUDY SAMPLE

Data is reported for measurements obtained on 15 subjects (average age of 72.2 years) with data from 30 ears.

RESULTS

There was a small effect of lens placement on sound field thresholds in most subjects. The largest damping effect of 4 dB was observed at 1000 Hz. An average reduction of 0.17 mL was identified in compliance following lens placement (p < 0.05). An effect of the lens on power absorbance obtained at ambient and peak pressure was found. The lens resulted in an increase in power absorbance at low frequencies (below 500 Hz) and a decrease in the mid to high-frequency range of approximately 500-3500 Hz (p < 0.05).

CONCLUSIONS

Lens wear had a small effect on audiometric thresholds and tympanometry for most patients. Clinicians who use compliance and power absorbance should take into consideration lens effects on these measurements. Additional work is required to develop clinical normative ranges of these measures for wearers of the Earlens.

Preserving Wideband Tympanometry Information With Artifact Mitigation

OBJECTIVE

Absorbance measured using wideband tympanometry (WBT) has been shown to be sensitive to changes in middle and inner ear mechanics, with potential to diagnose various mechanical ear pathologies. However, artifacts in absorbance due to measurement noise can obscure information related to pathologies and increase intermeasurement variability. Published reports frequently present absorbance that has undergone smoothing to minimize artifact; however, smoothing changes the true absorbance and can destroy important narrow-band characteristics such as peaks and notches at different frequencies. Because these characteristics can be unique to specific pathologies, preserving them is important for diagnostic purposes. Here, we identify the cause of artifacts in absorbance and develop a technique to mitigate artifacts while preserving the underlying WBT information.

DESIGN

A newly developed Research Platform for the Interacoustics Titan device allowed us to study raw microphone recordings and corresponding absorbances obtained by WBT measurements. We investigated WBT measurements from normal hearing ears and ears with middle and inner ear pathologies for the presence of artifact and noise. Furthermore, it was used to develop an artifact mitigation procedure and to evaluate its effectiveness in mitigating artifacts without distorting the true WBT information.

RESULTS

We observed various types of noise that can plague WBT measurements and that contribute to artifacts in computed absorbances, particularly intermittent low-frequency noise. We developed an artifact mitigation procedure that incorporates a high-pass filter and a Tukey window. This artifact mitigation resolved the artifacts from low-frequency noise while preserving characteristics in absorbance in both normal hearing ears and ears with pathology. Furthermore, the artifact mitigation reduced intermeasurement variability.

CONCLUSIONS

Unlike smoothing algorithms used in the past, our artifact mitigation specifically removes artifacts caused by noise. It does not change frequency response characteristics, such as narrow-band peaks and notches in absorbance at different frequencies that can be important for diagnosis. Also, by reducing intermeasurement variability, the artifact mitigation can improve the test-retest reliability of these measurements.

Wideband tympanometry in ears with superior canal dehiscence before and after surgical correction

OBJECTIVES

Wideband tympanometry (WBT) has been shown to be sensitive to mechanical changes in the ear. This study investigated the effect of surgical correction of superior canal dehiscence (SCD) on WBT (i.e. absorbance and middle ear resonance frequency) compared to those on common surgical outcomes such as symptom resolution, vestibular evoked myogenic potentials (VEMP), and hearing thresholds.

STUDY SAMPLE AND STUDY DESIGN

Seven patients (eight ears with SCD) who underwent surgical correction of SCD underwent WBT in addition to pure-tone audiometry and VEMP assessment.

RESULTS

Postoperatively, all ears showed normalised/decreased absorbance at low frequencies and slightly enhanced absorbance in the middle frequency range (7/8 ears). The middle ear resonance frequency, which was initially lower than normal in most patients, increased in 6/8 operated ears, and decreased in two ears with no/partial symptom relief. In comparison, complete symptom control was observed in 6/8 operated ears, VEMP amplitudes reduced or normalised in all ears, and hearing thresholds remained stable or improved in 6/8 ears and worsened in two ears.

CONCLUSIONS

Surgery seems to change the response to WBT in patients with SCD. The results of WBT may represent mechanical changes induced by SCD, and should be considered when evaluating surgical outcomes.

Wideband Acoustic Immittance in Cochlear Implant Recipients: Reflectance and Stapedial Reflexes

OBJECTIVES

to characterize differences in wideband power reflectance for ears with and without cochlear implants (CIs), to describe electrically evoked stapedial reflex (eSR)-induced changes in reflectance, and to evaluate the benefit of a broadband probe for reflex threshold determination for CI recipients. It was hypothesized that reflectance patterns in ears with CIs would be consistent with increased middle ear stiffness and that reflex thresholds measured with a broadband probe would be lower compared with thresholds obtained with a single-frequency probe.

DESIGN

Eleven CI recipients participated in both wideband reflectance and eSR testing. Ipsilateral reflexes were measured with three probes: a broadband chirp (swept from 200 to 8000 Hz), a 226 Hz tone, and a 678 Hz tone. Wideband reflectance measures acquired from 28 adults without CIs and with normal middle ear function served as a normative data set for comparison.

RESULTS

Considering the group data, average reflectance was significantly greater for ears with CIs across 250 to 891 Hz and 4238 to 4490 Hz compared with the normative data set, although individual reflectance curves were variable. Some CI recipients also had low 226 Hz admittance, which contributed to the group finding, considering the control group had clinically normal 226 Hz admittance by design. Electrically evoked stapedial reflexes were measurable in nine of 14 ears (64.3%) and in 24 of 46 electrodes (52.5%) tested. Reflex-induced changes in reflectance patterns were unique to the participant/ear, but similar across activators (electrodes) within a given ear. In addition, reflectance values at or above 1000 Hz were affected most by activating the stapedial reflex, even in ears with clinically normal 226 Hz admittance. This is a higher-frequency range than has been reported for acoustically evoked reflex-induced reflectance changes and is consistent with increased middle ear stiffness at rest. Electrically evoked reflexes could be measured more often with the 678 Hz or the broadband probe compared with the 226 Hz probe tone. Although reflex thresholds were lower with the broadband probe compared with the 678 Hz probe in 16 of 24 conditions, this was not a statistically significant finding (Wilcoxon signed-rank test; p = 0.072).

CONCLUSIONS

The applications of wideband acoustic immittance measurements (reflectance and reflexes) should also be considered for ears with CIs. Further work is needed to describe changes across time in ears with CIs to more fully understand the reflectance pattern indicating increased middle ear stiffness and to optimize measuring eSRs with a broadband probe.

Measurements of ear-canal cross-sectional areas from live human ears with implications for wideband acoustic immittance measurements

Wideband acoustic immittance (WAI) measures are noninvasive diagnostic measurements that require an estimate of the ear canal's area at the measurement location. Yet, physical measurements of the area at WAI probe locations are lacking. Methods to measure ear-canal areas from silicone molds were developed and applied to 169 subjects, ages 18-75 years. The average areas at the canal's first bend and at 12 mm insertion depth, which are likely WAI probe locations, were 63.4 ± 13.5 and 61.6 ± 13.5 mm², respectively. These areas are substantially larger than those assumed by current FDA-approved WAI measurement devices as well as areas estimated with acoustical methods or measured on cadaver ears. Left and right ears from the same subject had similar areas. Sex, height, and weight were not significant factors in predicting area. Age cohort was a significant predictor of area, with area increasing with decade of life. A subset of areas from the youngest female subjects did not show an effect of race on area (White or Chinese). Areas were also measured as a function of insertion depth of 4.8-13.2 mm from the canal entrance; area was largest closest to the canal entrance and systematically decreased with insertion depth.

Refining Measurements of Power Absorbance in Newborns: Probe Fit and Intrasubject Variability

OBJECTIVES

Because unresolved debris in the ear canal or middle ear of newborns may produce high false positive rates on hearing screening tests, it has been suggested that an outer/middle ear measure can be included at the time of hearing screening. A potential measure is power absorbance (absorbance), which indicates the proportion of power in a broadband acoustic stimulus that is absorbed through the outer/middle ear. Although absorbance is sensitive to outer/middle dysfunction at birth, there is large variability that limits its accuracy. Acoustic leaks caused by poor probe fitting further exacerbate this issue. The objectives of this work were to: (1) develop criteria to indicate whether a change in absorbance occurs in association with probe fit; (2) describe the variability in absorbance due to poor fitting; and (3) evaluate test-retest variability with probe reinsertions, excluding poor fits.

DESIGN

An observational cross-sectional design was used to evaluate changes in absorbance due to probe fit and probe reinsertion. Repeated measurements were recorded in 50 newborns (98 ears) who passed TEOAE screenings and were <48 hours of age. One absorbance measurement was chosen as the baseline that served as a best-fit reference in each ear. Changes in absorbance, called absorbance probe-fit Δ, were calculated relative to the baseline in each ear. Correlations were assessed between the absorbance probe-fit Δ and low-frequency absorbance, impedance magnitude, impedance phase, and equivalent volume, to determine which measures predicted poor fits. Criteria were derived from the strongest of these correlations and their performance was analyzed. Next, measurements with poor/leaky fits were identified, and the changes in absorbance that they introduced were analyzed. Excluding the poor fits, test-retest differences in absorbance, called reinsertion Δ, were determined. Variability was assessed using the SDs associated with absorbance, absorbance probe-fit Δ, and reinsertion Δ.

RESULTS

Based on the analysis of 12 moderate-strong correlations, the following criteria were adopted to identify measurements with poor fits: (1) impedance phase-based criterion (500 to 1000 Hz) > -0.11 cycles and (2) absorbance-based criterion (250 to 1000 Hz) > 0.58. Poor-fit measurements introduced statistically significant increases in absorbance up to 0.1 for 1000 to 6000 Hz, and up to 0.4 for frequencies <1000 Hz. Reinsertion Δ were ≤0.02, and were significant for 500 to 5000 Hz. The SDs of absorbance probe-fit Δ were greatest and similar to overall absorbance SD in the low frequencies. Separately, the SDs of reinsertion Δ were also greatest and similar to low-frequency absorbance SD.

CONCLUSIONS

Poor probe fits introduced the greatest inflation in absorbance for frequencies < 500 Hz, and a smaller but significant inflation for higher frequencies, consistent with controlled experiments on acoustic leaks in adults. Importantly, inflation of absorbance in diagnostically sensitive 1000 to 2000 Hz may impact its clinical performance. Test-retest with probe reinsertion contributed significantly to absorbance variability, especially in the low frequencies, consistent with reports in adults, even though changes were smaller than those associated with poor probe fit. The results indicate that variability in absorbance was reduced by minimizing acoustic leaks. Pending further validation, the probe-fit criteria developed in this work can be recommended to ensure proper probe fit.

Measurement of Wideband Absorbance as a Test for Otosclerosis

The purpose of this study was to investigate the effectiveness of wideband energy absorbance in diagnosing otosclerosis by comparing the differences in acoustic absorbance between otosclerotic and normal ears. Exactly 90 surgically confirmed otosclerotic ears were included in the test group. The control group consisted of 126 matched normal-hearing subjects. The Titan hearing test platform (Interacoustics) was used for absorbance and acoustic immittance tests. Energy absorbance, measured at tympanometric peak pressure, was analyzed in the range 226–8000 Hz. Differences between normal and otosclerotic ears were analyzed in quarter-octave bands. Wideband absorbance, i.e., absorbance averaged over the 226–2000 Hz band, and resonance frequency were calculated and compared between normal and otosclerotic ears. Significant differences between the absorbance of normal and otosclerotic ears were found, especially at low and middle frequencies. No significant effect of ear side or gender was observed. For average wideband absorbance and resonance frequency, less pronounced (although significant) differences were found between normal and otosclerotic ears. Measurement of peak-pressure energy absorbance, averaged over a frequency band around 650 Hz, provides a valid criterion in testing for otosclerosis. The test is highly effective, with a sensitivity and specificity of over 85% and area under receiver operating characteristic curve above 0.9. Average wideband absorbance can also be used, but its effectiveness is lower. Other immittance-related measures are considerably less effective.

Causality-constrained measurements of aural acoustic reflectance and reflection functions

Causality-constrained procedures are described to measure acoustic pressure reflectance and reflection function (RF) in the ear canal or unknown waveguide, in which reflectance is the Fourier transform of the RF. Reflectance calibration is reformulated to generate causal outputs, with results described for a calibration based on a reflectance waveguide equation to calculate incident pressure and source reflectance in the frequency domain or source RF in the time domain. The viscothermal model RF of each tube is band-limited to the stimulus bandwidth. Results are described in which incident pressure is either known from long-tube measurements or calculated as a calibration output. Calibrations based on constrained nonlinear optimizations are simpler and more accurate when incident pressure is known. Outputs measured by causality-constrained procedures differ at higher frequencies from those using standard procedures with non-causal outputs. Evanescent-mode effects formulated in the time domain and incorporated into frequency-domain calibrations are negligible for long-tube calibrations. Causal reflectance and RFs are evaluated in an adult ear canal and time- and frequency-domain results are contrasted using forward and inverse Fourier transforms. These results contribute to the long-term goals of improving applications to calibrate sound stimuli in the ear canal at high frequencies and diagnose conductive hearing impairments.

Wideband Absorbance and 226-Hz Tympanometry in the Prediction of Optimal Distortion Product Otoacoustic Emission Primary Tone Levels

PURPOSE

Distortion product otoacoustic emission (DPOAE) amplitude is sensitive to the primary tone level separation effective within the cochlea. Despite potential for middle ear sound transmission characteristics to affect this separation, no primary tone level optimization formula accounts for its influence. This study was conducted to determine if inclusion of ear- and frequency-specific immittance features improves primary tone level optimization formula performance beyond that achieved using a univariate, L2-based formula.

METHOD

For 30 adults with normal hearing, DPOAE, wideband absorbance, and 226-Hz tympanometry measures were completed. A mixed linear modeling technique, incorporating both primary tone and acoustic immittance features, was used to generate a multivariable formula for the middle ear-specific recommendation of primary tone level separations for f2 = 1-6 kHz. The accuracy with which L1OPT, or the L1 observed to maximize DPOAE level for each given L2, could be predicted using the multivariable formula was then compared with that of a traditional, L2-based univariate formula for each individual ear.

RESULTS

Use of the multivariable formula L1 = 0.47L2 + 2.40A + f2param + 38 [dB SPL] resulted in significantly more accurate L1OPT predictions than did the univariate formula L1 = 0.49L2 + 41 [dB SPL]. Although average improvement was small, meaningful improvements were identified within individual ears, especially for f2 = 1 and 6 kHz.

CONCLUSION

Incorporation of a wideband absorbance measure into a primary tone level optimization formula resulted in a minor average improvement in L1OPT prediction accuracy when compared with a traditional univariate optimization formula. Further research is needed to identify characteristics of ears that might disproportionately benefit from the additional measure.

Normative Wideband Reflectance, Equivalent Admittance at the Tympanic Membrane, and Acoustic Stapedius Reflex Threshold in Adults

OBJECTIVES

Wideband acoustic immittance (WAI) measures such as pressure reflectance, parameterized by absorbance and group delay, equivalent admittance at the tympanic membrane (TM), and acoustic stapedius reflex threshold (ASRT) describe middle ear function across a wide frequency range, compared with traditional tests employing a single frequency. The objective of this study was to obtain normative data using these tests for a group of normal-hearing adults and investigate test-retest reliability using a longitudinal design.

DESIGN

A longitudinal prospective design was used to obtain normative test and retest data on clinical and WAI measures. Subjects were 13 males and 20 females (mean age = 26 years). Inclusion criteria included normal audiometry and clinical immittance. Subjects were tested on two separate visits approximately 1 month apart. Reflectance and equivalent admittance at the TM were measured from 0.25 to 8.0 kHz under three conditions: at ambient pressure in the ear canal and with pressure sweeps from positive to negative pressure (downswept) and negative to positive pressure (upswept). Equivalent admittance at the TM was calculated using admittance measurements at the probe tip that were adjusted using a model of sound transmission in the ear canal and acoustic estimates of ear-canal area and length. Wideband ASRTs were measured at tympanometric peak pressure (TPP) derived from the average TPP of downswept and upswept tympanograms. Descriptive statistics were obtained for all WAI responses, and wideband and clinical ASRTs were compared.

RESULTS

Mean absorbance at ambient pressure and TPP demonstrated a broad band-pass pattern typical of previous studies. Test-retest differences were lower for absorbance at TPP for the downswept method compared with ambient pressure at frequencies between 1.0 and 1.26 kHz. Mean tympanometric peak-to-tail differences for absorbance were greatest around 1.0 to 2.0 kHz and similar for positive and negative tails. Mean group delay at ambient pressure and at TPP were greatest between 0.32 and 0.6 kHz at 200 to 300 μsec, reduced at frequencies between 0.8 and 1.5 kHz, and increased above 1.5 kHz to around 150 μsec. Mean equivalent admittance at the TM had a lower level for the ambient method than at TPP for both sweep directions below 1.2 kHz, but the difference between methods was only statistically significant for the comparison between the ambient method and TPP for the upswept tympanogram. Mean equivalent admittance phase was positive at all frequencies. Test-retest reliability of the equivalent admittance level ranged from 1 to 3 dB at frequencies below 1.0 kHz, but increased to 8 to 9 dB at higher frequencies. The mean wideband ASRT for an ipsilateral broadband noise activator was 12 dB lower than the clinical ASRT, but had poorer reliability.

CONCLUSIONS

Normative data for the WAI test battery revealed minor differences for results at ambient pressure compared with tympanometric methods at TPP for reflectance, group delay, and equivalent admittance level at the TM for subjects with middle ear pressure within ±100 daPa. Test-retest reliability was better for absorbance at TPP for the downswept tympanogram compared with ambient pressure at frequencies around 1.0 kHz. Large peak-to-tail differences in absorbance combined with good reliability at frequencies between about 0.7 and 3.0 kHz suggest that this may be a sensitive frequency range for interpreting absorbance at TPP. The mean wideband ipsilateral ASRT was lower than the clinical ASRT, consistent with previous studies. Results are promising for the use of a wideband test battery to evaluate middle ear function.

The area discontinuity between probe and ear canal as a source of power-reflectance measurement-location variability

This study examined the effect of the area discontinuity between the measurement-probe sound source and ear canal on the plane-wave approximation of power reflectance. The area discontinuity was hypothesized to introduce measurement-location sensitivity to the power reflectance, especially above 5 kHz. Measurements were made in human and artificial ear canals (tubes coupled to an IEC711 ear simulator). In both cases, the power reflectance exhibited a high-frequency notch that decreased in frequency as the residual canal length increased. The area discontinuity between probe and canal was modeled as an inductance in series with the canal's acoustic impedance. To compensate for the effects of the discontinuity, the discontinuity's impedance was subtracted from the measured load impedance of the canal. In the artificial ears, compensation for the estimated area discontinuity removed the high-frequency notch and reduced the position dependence of the power reflectance. Subtracting the estimated discontinuity impedance from the load impedance in the human ears had a minimal effect on the power-reflectance measurement-location variability and magnitude of the high-frequency notch. The area-discontinuity between probe and ear canal is not supported as the primary source of measurement-variability in the plane-wave approximation of the power reflectance in human ears.

Reflectance measurement validation using acoustic horns

Variability in wideband acoustic reflectance (and absorbance) measurements adversely affects the clinical utility of reflectance for diagnosis of middle-ear disorders. A reflectance standard would encourage consistency across different measurement systems and help identify calibration related issues. Theoretical equations exist for the reflectance of finite-length exponential, conical, and parabolic acoustic horns. Reflectance measurements were repeatedly made in each of these three horn shapes and the results were compared to the corresponding theoretical reflectance. A method is described of adjusting acoustic impedance measurements to compensate for spreading of the wave front that propagates from the small diameter sound port of the probe to the larger diameter of the acoustic cavity. Agreement between measured and theoretical reflectance was less than 1 dB at most frequencies in the range from 0.2 to 10 kHz. Pearson correlation coefficients were greater than 0.95 between measured and theoretical time-domain reflectance within the flare region of the horns. The agreement suggests that the distributed reflectance of acoustic horns may be useful for validating reflectance measurements made in human ear canals; however, refinements to reflectance measurement methods may still be needed.