Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Jul 1.
Published in final edited form as: Int J Audiol. 2023 Jun 2;63(7):491–499. doi: 10.1080/14992027.2023.2215943

Rethinking the clinical utility of distortion-product otoacoustic emission (DPOAE) signal-to-noise ratio

Nicholas Portugal a, Gayla L Poling b, Laura Dreisbach a,*
PMCID: PMC10692308  NIHMSID: NIHMS1917899  PMID: 37267054

Abstract

Objective:

Distortion-product otoacoustic emission (DPOAE) levels are repeatable over time in normal-hearing individuals making DPOAE levels an ideal measurement for monitoring cochlear status in clinic and research applications. However, if DPOAE signal-to-noise ratio (SNR) values instead of levels are used for monitoring, the repeatability of this value needs to be established. This retrospective, cross-sectional study sought to determine DPOAE SNR repeatability in younger children, older children, young adults, and a patient population with normal hearing.

Design:

Each participant attended four sessions where DPOAE discrete frequency sweeps were collected at conventional (≤ 8 kHz) and/or extended-high frequencies (> 8 kHz). To examine the extent of variability to be expected for DPOAE SNR, average absolute SNR differences-between-trials were determined and compared to average absolute DPOAE level differences-between-trials.

Study Sample:

One hundred forty-five participants, incorporating four different groups from three different studies. Ages ranged from 3 to 55 years.

Results:

Average SNR differences-between-trials across all frequencies are greater than differences for average DPOAE levels. Improved calibration methods result in SNR differences-between-trials that are similar across all frequencies

Conclusions:

When monitoring cochlear health over an extended bandwidth, DPOAE levels are less variable across trials than SNR values, thus allowing earlier indicators of cochlear damage.

Keywords: Otoacoustic emissions, distortion product otoacoustic emissions, signal-to-noise ratio, normal hearing, monitoring, calibration

Introduction

Monitoring cochlear function over time is critical for those exposed to ototoxic agents or when evaluating the efficacy of pharmaceutical interventions. Thus, the tools used for monitoring must be repeatable across time and sensitive to cochlear change. One such widely used tool is distortion product otoacoustic emissions (DPOAEs), which are obtained in an objective, noninvasive manner. To interpret results, the level (and often the phase) of the DPOAE and the noise floor are provided. From these values a signal-to-noise ratio (SNR) is computed as the difference between the levels of the DPOAE and the noise floor. To use DPOAE results to their fullest capacity, both the level and the noise need to be understood given the critical relationship between the two for interpretation. Previous work indicates that the DPOAE level and phase are repeatable over time allowing a sensitive, early indicator of cochlear damage across the lifespan (Conrad & Dreisbach, 2011; Dreisbach et al., 2017; Dreisbach et al., 2019; Dreisbach et al., 2018; Dreisbach et al., 2006; Poling et al., 2014). Despite the notion that noise would be inherently more variable than a cochlear response, noise floor, and thus SNR, variability has not been widely reported. In some clinical applications there is a reliance on SNR values, not DPOAE levels or phase, for the interpretation and monitoring of responses. If SNR values are to be used to monitor cochlear function over time, the repeatability of this measurement needs to be established. In the current study the repeatability of DPOAE SNR values was determined and compared to DPOAE level repeatability in children and adults across an extended bandwidth.

DPOAEs are low level sounds generated by cochlear processes that involve outer hair cell (OHC) motility and are measured in the ear canal (e.g., Kemp et al., 1986). Specifically, DPOAEs occur in response to two simultaneous pure tones (f1 and f2, f1 < f2) presented to the cochlea that are close in frequency (Probst et al., 1991). Responses to the two tones overlap along the basilar membrane and due to the nonlinear properties of the cochlea, characteristic distortion-products are generated. The focus of clinical applications is the spectrum component 2f1-f2 because it is the most prominent in humans (Dorn et al., 1999). DPOAE levels depend on the stimulus frequencies (f1 and f2), levels (L1 and L2), and ratio (f2/f1) used to elicit the response (Dreisbach & Siegel, 2001; Gaskill & Brown, 1990). A discrete frequency sweep, or DP-gram, where only frequency (f1 and f2) is varied is traditionally used in clinical settings because it allows for a quick, noninvasive, and frequency-specific measurement of OHC function. Typically, moderate stimulus levels (e.g., L1/L2=65/55 dB SPL) and a frequency ratio of 1.22 are used because these parameters generate the largest DPOAE levels in most ears across clinically tested frequencies (≤ 8 kHz) (Gaskill & Brown, 1990).

DPOAE levels are further dependent on how the eliciting stimuli are calibrated. Traditional calibration methods, which use a microphone placed in the ear canal, allow for little control over the stimulus levels presented to the eardrum at frequencies influenced by standing waves. Standing waves occur at certain frequencies (e.g., 3 to 7 and ≥ 10 kHz for an adult human ear canal), dependent on how deeply inserted the microphone is placed, when closing off the ear canal with the DPOAE probe. The standing waves can cause a 15 to 20 dB change in the sound pressure level (SPL) delivered to the eardrum relative to what is measured at the microphone placed in the ear canal (e.g., Dreisbach & Siegel, 2001; Richmond et al., 2011; Siegel, 1994; Siegel & Hirohata, 1994). To help reduce the variability in stimulus levels at the eardrum, alternative calibration methods have been developed (Souza et al., 2014). Control over the intended stimulus level and a sufficiently low noise floor help improve the determination and interpretation of DPOAE responses (Poling et al., 2014). Similarly, a designated SNR value can be chosen to determine if a DPOAE response is considered interpretable or not beyond considering the DPOAE (and noise floor) levels alone (Gorga et al., 1999). Evidence supports that different SNR criteria have impacts for determining and monitoring DPOAE levels in the presence of varied testing conditions and populations (Lonsbury-Martin et al., 1990; Popelka et al., 1993; Smurzynski, 1994). Typically, the higher the SNR value (e.g., > 9 dB), the greater the likelihood that the recorded DPOAE level is of cochlear origin and not influenced by the noise floor (Whitehead et al., 1993). However, relying on only one variable of the recorded measurement (e.g., DPOAE level, noise floor, SNR) can lead to errors in interpretation.

Serial DPOAE measurements are used clinically for differential diagnosis and to assess cochlear status in the presence of ototoxic agents or pharmaceutical interventions (Konrad-Martin et al., 2016; Poling et al., 2019; Probst et al., 1991). DPOAE measurements in the conventional (≤ 8 kHz) and extended-high frequency (EHF, > 8 kHz) regions are used in individuals that cannot provide reliable behavioral responses and to identify the earliest signs of cochlear damage before primary speech frequencies are involved (Dreisbach et al., 2017; Dreisbach et al., 2018). In these applications, the DPOAE levels that meet defined criteria, for example, DPOAE levels greater than the individual’s noise floor by a specific amount (e.g., 6 dB) and above system distortion levels, are interpreted as a response. These DPOAE responses are classified clinically as a determinant of normally or close-to-normally functioning OHCs (e.g., Kemp et al., 1986).

To be considered a viable diagnostic tool for serial monitoring, the tool needs to be repeatable. DPOAE levels are repeatable over time in healthy, normal-hearing adults, children, newborns (Conrad & Dreisbach, 2011; Dreisbach et al., 2019; Dreisbach et al., 2006; Franklin et al., 1992; Newman & Dreisbach, 2012; Poling et al., 2014; Reavis et al., 2015; Roede et al., 1993) and a patient population (Dreisbach et al., 2018) making DPOAE levels an ideal measurement for monitoring cochlear status in clinical and research applications. If DPOAE SNR values, not levels, are used for monitoring purposes, the repeatability of this value needs to be established for clinical utility. This retrospective, cross-sectional study seeks to determine DPOAE SNR repeatability and make comparisons to previously reported DPOAE level repeatability from three different studies where younger children (YC), older children (OC), young adults (YA), and a patient population were examined (Conrad & Dreisbach, 2011; Dreisbach et al., 2018; Dreisbach et al., 2006; Newman & Dreisbach, 2012).

Materials and Methods

Participants

This retrospective study comprised 145 participants, incorporating four different groups from three different studies (refer to Table 1). Each participant had normal middle-ear function, defined as a Type A tympanogram using a 0.226 kHz probe tone and a present contralateral acoustic reflex at 90 dB HL at 1 kHz to screen the integrity of the acoustic reflex pathway. The first study included children that were divided into younger children (YC group) and older children (OC group). The second study consisted of young adults (YA group). The participants for these groups were included if they had present DPOAEs, defined as SNR values of minimally 6 dB and DPOAE levels > −20 dB SPL, through 16 kHz. Additionally, the OC and YA groups had behavioral hearing thresholds < 30 dB SPL for 1 through 8 kHz, which approximates hearing within normal limits. The third study was comprised of patients with cystic fibrosis (CF group) where only responses greater than 8 kHz were collected. Each participant had at least one clear ear canal, measurable behavioral hearing thresholds (minimally from 8 to 12.5 kHz) and at least two consecutive, present DPOAEs (8 to 16 kHz), defined as SNR values ≥ 6 dB and levels > −20 dB SPL. All studies from which data were included had approval from the Institutional Review Board at San Diego State University and the patient population (CF) study also had approval from the University of California, San Diego.

Table 1.

Participant demographics including age range, number of participants, sex, DPOAE stimulus frequencies (f2, kHz), ratio (f2/f1), and level (L1/L2, dB SPL) combination used, and calibration method (depth-compensated or traditional) used for the four participant groups, young children (YC), older children (OC), young adults (YA), and patients with cystic fibrosis (CF) examined in this study.

YC OC YA CF
Age Range (yrs) 3-6 10-12 18-29 19-55
# Participants 39 41 25 40
Females/ Males 19/20 17/24 14/11 17/23
f2 (kHz) 1-16 1-16 2-16 8-16
f2/f1 1.2 1.2 1.2 1.2
L1/L2 65/50 65/50 60/50 65/50
Calibration Depth Depth Traditional Depth
Open in a new tab

Equipment, Calibration, and Software

All studies used a non-commercially available emission probe system that reliably measured DPOAEs up to 16 kHz. To measure DPOAEs in the children (YC and OC) and CF group, the signals for both the f2 and the f1 channels were generated at the digital-to-analog (D/A) converter of a MOTU 96 kHz Audio Firewire Interface, which was connected to a computer, and presented through Sennheiser CX 300 ultrahigh frequency transducers. The transducers were connected to a low-noise DPOAE probe assembly positioned in the subject’s ear canal. The probe assembly also housed the emission probe microphone (Etymotic Research ER-10B+), which transduced the sound pressure in the ear canal and amplified (20 dB) the signal before being digitized by the analog-to-digital (A/D) converter of the MOTU Audio Firewire Interface. An alternative depth-compensated simulator SPL calibration method was used to alleviate previously described calibration errors related to standing waves. Details surrounding the depth-compensated calibration procedure used for the children and CF group have been published previously (Conrad & Dreisbach, 2011; Dreisbach et al., 2018; Newman & Dreisbach, 2012). In short, real-time simultaneous response computing software, SysRes (Neely & Stevenson, 2002), was used to record the half-wave resonance of the participant’s ear canal allowing the calibration of the stimulus levels. At each of the four trials, the participant’s half-wave resonance was measured, and the ear simulator microphone pressure response with the closest half-wave resonance frequency was selected as the estimate of the participant’s eardrum pressure, as described in Dreisbach et al. (2019; 2018). Matching probe placements for each trial was attempted for added control of individual variability. It should be noted that the cavity (Bruel & Kjær 4157 ear simulator) used to create the stimulus files for the depth-compensated simulator SPL calibration procedure is based on impedance characteristics of the adult ear canal. If higher impedances exist for children, it is possible that higher stimulus levels were delivered to the enclosed ear canal in comparison to adults. However, these differences in the stimulus levels would be approximately a few decibels and given the compressive growth of DPOAE levels at moderate stimulus levels, as used in this study, this alternative calibration method was chosen as a better alternative to traditional calibration methods.

In the YA study, a similar approach was used to measure DPOAEs using slightly different instrumentation. The D/A converter was a Card Deluxe digital signal processing board and Tucker Davis Technologies Electrostatic EDI frequency drivers generated the stimuli. The same emission probe system (ER-10B+) used with the children and CF group was used for the YA participants. Traditional calibration methods were used where both the stimulus levels and emissions levels were recorded in the ear canal.

System distortion (energy recorded in the test cavity at 2f1-f2) was maximally −20 dB SPL for all stimulus levels and frequencies used when measured in an acoustic cavity (Bruel & Kjær 4157). . Otoacoustic Emission Averager (EMAV, Neely & Liu, 1993) software was used to collect all DPOAE data, where the ear-canal sound pressure is sampled in 46-ms time windows. The samples are accumulated into 2 interleaved buffers (A and B) and averaged in the time domain. The grand average ([A + B]/2) is transformed to estimate the levels and phases of both stimuli and DPOAEs present in the ear canal while the two subaverages are subtracted (A – B) and Fast Fourier transformed to provide an estimate of the background noise floor at the distortion-product frequencies. Sampling occurred, for a minimum of 4 seconds or longer, until one of two stopping rules were met: the noise floor at the DPOAE was less than −20 dB SPL or until 4 seconds of artifact-free sampling had been averaged.

Procedure

For all studies, each participant attended four trials (T) separated in time by approximately one week. Some variability in the time difference between trials existed and is defined in detail in the corresponding methods sections of the previous studies (Conrad & Dreisbach, 2011; Dreisbach et al., 2018; Dreisbach et al., 2006; Newman & Dreisbach, 2012). Data were collected in a sound booth for the YC, OC, and YA groups while the CF group was tested in a quiet clinical examination room. Upon normal outer and middle ear status confirmation, DPOAEs were measured in one ear. Specifically, 2f1-f2 DPOAE data at discrete f2 frequencies from 1 to 16 kHz were elicited for the YC and OC groups, while DPOAEs for f2 frequencies from 2 to 16 kHz were collected for the YA group, and f2 frequencies between 8 to 16 kHz were used for the CF group (refer to Table 1). The levels of the primary tones (L1/L2) were fixed at 65/50 dB SPL for the YC, OC, and CF groups, while levels of 60/50 dB SPL were utilized for the YA group. A fixed frequency ratio (f2/f1) of 1.2 was used for all participants.

Data Analyses

SNRs were calculated by subtracting the level of the noise floor from the DPOAE level when the level of the DPOAE was greater than −20 dB SPL (Konrad-Martin et al., 2016). The DPOAE was considered present if it met this DPOAE level criterion and had an SNR of 6 dB or more. To determine the average SNR for each group, only present DPOAEs were included. The average SNR values across trials were calculated separately for conventional (2 to 8 kHz) and EHFs (> 8 kHz) to allow for comparisons between groups.

To examine the extent of variability to be expected for DPOAE SNR values, the absolute value of the SNR differences-between-trials (6 calculations; ∣T1-T2∣, ∣T1-T3∣, ∣T1-T4∣, ∣T2-T3∣, ∣T2-T4∣, ∣T3-T4∣) for each of the frequencies tested was determined from those participants with present DPOAEs. Average absolute SNR differences-between-trials were determined and compared to previously reported average absolute DPOAE level differences-between-trials for conventional and EHFs. In addition, standard deviations (SDs), correlations, and standard error (SE) of the measurements were calculated.

All data were analyzed in Excel spreadsheets for each group separately. A 2-factor analysis of variance with repeated measures (ANOVA-R), with frequency and trial as variables, was completed and compared to previously reported DPOAE level results. Significance for this study was defined as p < 0.05.

Results

DPOAE Level and Noise Floor

For each group, average DPOAE levels (Ldp) and noise floor (Ndp) data for each of the four trials as a function of frequency are shown in Figure 1 for participants with present emissions. The four different test groups are represented in separate panels. Average DPOAE levels were greater for the conventional frequencies compared to the EHFs. DPOAE levels were relatively flat across most frequencies for the YC, OC, and CF groups, followed by a slight decline at the highest frequencies tested. There was more variability in the DPOAE levels across frequency for the YA group compared to the other groups. A slight peak was noted in the DPOAE level from approximately 4 through 6 kHz and was accompanied by a valley in the average noise floor for the YC, OC, and YA groups. Group averages for DPOAE levels, collapsed across trial, for the conventional frequencies (2 to 8 kHz) and EHFs (> 8 kHz) for the YC, OC, and YA groups and EHFs for the CF group are represented by the horizontal dashed lines in Figure 1. The average DPOAE levels across trials for conventional frequencies ranged between −2.16, and 2.15 dB SPL and between −5.19 and −1.60 dB SPL for the EHFs. The average noise floor level across trials for conventional frequencies (2 to 8 kHz) ranged from −30.78 to −27.82 dB SPL and from −28.67 to −24.88 for the EHFs (> 8 kHz).

Figure 1.

Figure 1.

Open in a new tab

Average DPOAE levels (Ldp) and noise floor (Ndp) data for various f2 frequencies are shown for the four different trials. Each panel represents data for one of the participant groups with the children (YC and OC) in the top row and the adults (YA and CF) in the bottom row. The diamonds, circles, triangles, and squares correspond to trials 1, 2, 3, and 4, respectively, and error bars are standard error of the mean (SEM). The solid lines represent the Ldp while the dashed lines represent the Ndp. Group averages for DPOAE levels, collapsed across trials and frequencies (conventional frequencies on the left and EHFs on the right, except for CF where only EHFs were tested) are represented by the horizonal dotted lines in each panel.

DPOAE SNR

For each group, average DPOAE SNR as a function of frequency for the four trials are shown in separate panels in Figure 2 for participants with present emissions. Average DPOAE SNR values were greater for the conventional frequencies compared to the EHFs. SNR values across the frequencies tested in the YC, OC, and YA groups resulted in the greatest values occurring at approximately 4 through 6 kHz, corresponding to the peak in DPOAE levels and lower noise floors. Group averages for SNR, collapsed across trial, for the conventional and EHFs, are represented in Figure 2 by the horizontal dashed lines and are all greater than 20 dB and within approximately 3.5 dB of each other. Repeatability of DPOAE SNR across trial was examined using a 2-factor (frequency × trial) repeated-measures ANOVA. The main effect of trial and interaction of frequency by trial were not significant (p > 0.05). However, a significant main effect for frequency [F(20, 780) = 48.65; p < 0.01; F(20, 800) = 42.83; p < 0.01; F(13, 312) = 55.10; p < 0.01; F(15, 570) = 50.76; p < 0.01] for the YC, OC, YA, and CF groups, respectively, was identified, indicating that DPOAE SNR varied across frequency, which is seen in Figure 2.

Figure 2.

Figure 2.

Open in a new tab

Average DPOAE signal-to-noise ratio (SNR) for f2 frequencies are shown for the four different trials. Each panel represents the four different participant groups, with the children (YC and OC) in the top row and the adults (YA and CF) in the bottom row. The diamonds, circles, triangles, and squares correspond to trials 1, 2, 3, and 4, respectively, and error bars are SEM. Group averages for SNR, collapsed across trial, for the conventional frequency range (on the left) and EHFs (on the right, except for CF where only EHFs were tested) are represented by the horizontal dashed lines in each panel.

Difference Scores

To examine the extent of variability to be expected for DPOAE SNR, absolute value differences-between-trials (6 calculations) for each of the frequencies tested were determined for those with present DPOAEs in each group. Average absolute SNR differences-between-trials for conventional and EHFs are represented for each group in separate panels in Figure 3 (dashed lines). For comparison, the previously reported average absolute DPOAE level differences-between-trials, for conventional and EHFs (Conrad & Dreisbach, 2011; Dreisbach et al., 2018; Dreisbach et al., 2006; Newman & Dreisbach, 2012), also are represented in Figure 3 (dotted lines). Greater variability was found for SNR values compared to DPOAE levels (Ldp) for all groups. The average DPOAE SNR differences-between-trials for the higher (> 8 kHz) and lower (2 to 8 kHz) frequencies for the YC group were 7.33 (Ldp 4.55 dB) and 7.94 dB (Ldp 3.38 dB), respectively. OC group differences for high and low frequencies were 7.12 (Ldp 4.44 dB) and 7.43 dB (Ldp 2.97 dB), respectively. YA group differences for high and low frequencies were 6.52 (Ldp 5.48 dB) and 4.57 dB (Ldp 3.15 dB), respectively, while SNR differences at higher frequencies for the CF group were 6.09 dB (Ldp 1.96 dB). Average SNR differences-between-trials were slightly greater for lower frequencies than higher frequencies with depth-compensated calibrations (YC, OC groups), whereas the reverse was found for the YA group and traditional calibration methods but with larger differences between the two frequency regions.

Figure 3.

Figure 3.

Open in a new tab

Average absolute DPOAE SNR difference-between-trials for the frequencies tested are shown for the four different participant groups, with the children (YC and OC) in the top row and the adults (YA and CF) in the bottom row. The varying shades of black lines represent the different trial combinations. The average absolute SNR difference-between-trials collapsed across trial combinations for the frequencies tested, are represented by the thick line. Thicker horizontal dashed lines without symbols in each panel represent group averages for DPOAE levels, collapsed across trials and frequencies (conventional frequencies and EHFs for all participant groups, except for CF where only EHFs were tested). For comparison, the previously reported (in Figure 1) average absolute DPOAE level differences-between-trials collapsed across trial combinations for conventional and EHFs, are represented by the dotted lines.

The SNR absolute differences-between-trials across frequency ranged from 0.0 to 40.77, 41.73, 29.10, 50.62 dB for the YC, OC, YA, and CF groups, respectively. The DPOAE levels absolute differences-between-trials across frequency ranged from 0.0 to 28.53, 28.16, 24.03, 20.97 dB for the YC, OC, YA, and CF groups, respectively. The noise floor levels absolute differences-between-trials across frequency ranged from 0.0 to 51.16, 42.40, 35.93, 51.07 dB for the YC, OC, YA, and CF groups, respectively. The DPOAE levels varied the least, while noise floor levels varied the most, between trials for all groups.

The average absolute SNR difference-between-trials, SE of the measurement, and correlations for all trial comparisons (e.g., T1-T2) and for each frequency for all participant groups can be found in the supplemental materials (Supplemental Tables 1-3). All but one (11/12) of the maximum variabilities between trials occurred in the conventional frequencies for the YC and OC groups versus EHFs (6/6) for the YA group. The minimum variability between trials occurred most frequently at 16 kHz (7/24) followed by 2 and 6 kHz (3/24). The maximum variability for each frequency in each trial comparison (e.g., T1-T2) was summed and divided by the total number of frequencies (72). No trial comparison yielded the largest SNR differences more than 19% (14/72) of the time. Standard deviations ranged from 2.04 to 10.26, SE of the measurement ranged from 2.37 to 10.69, and correlations ranged from −0.42 to 0.89 for the average SNR difference-between-trials.

Discussion

This study explored the feasibility of using DPOAE SNR, rather than DPOAE level, as a means for monitoring cochlear function over time by comparing the repeatability of DPOAE SNR values with DPOAE levels. The clinical utility of tools used for monitoring cochlear status must be repeatable over time and able to detect significant change in cochlear status. DPOAE levels measured at both conventional (2 to 8 kHz) and EHFs (>8 kHz) are a measurement that meets these criteria (Conrad & Dreisbach, 2011; Dreisbach et al., 2017; Dreisbach et al., 2019; Dreisbach et al., 2018; Dreisbach et al., 2006; Konrad-Martin et al., 2016; Newman & Dreisbach, 2012; Reavis et al., 2015). However, in clinical applications SNR values are used most often to determine if DPOAE responses are valid and interpretable and are often used as a measure for monitoring cochlear status. Thus, to have confidence in DPOAE SNR as a measurement for use in monitoring programs, the repeatability of this measure must be ascertained and compared to those previously determined to be viable measurements, namely DPOAE levels.

For clinical monitoring purposes, DPOAEs elicited at conventional frequencies with standard parameters (i.e., L1/L2 =65/55 dB SPL, f2/f1 =1.22) are measured at baseline and are compared to subsequent outcomes. In general, differences in DPOAE levels from baseline, in either a positive or negative direction, are considered clinically significant when they exceed test-retest variability. Thus, the ability to detect the smallest changes possible making the tool used more sensitive relies on minimizing the variability across trials (Konrad-Martin et al., 2016). At conventional frequencies, changes in DPOAE level from baseline of ± 4 to 6 dB are generally considered a significant change (Beattie et al., 2003; Franklin et al., 1992; Reavis et al., 2015; Roede et al., 1993), while level changes of ± 5 to 6 dB are considered significant for EHFs (Dreisbach et al., 2018). However, it is important to note that in clinical interpretations the levels of the DPOAE and noise floor, as well as SNR values need to be evaluated (e.g., SNR can remain constant over trials while DPOAE levels can change significantly).

Tools identifying the earliest changes to cochlear function are ideal for ototoxicity monitoring programs and DPOAE serial measurements identify damage at the same time or earlier than behavioral threshold measures (Dreisbach et al., 2017; Katbamna et al., 1999; Knight et al., 2007; Mulheran & Degg, 1997; Ress et al., 1999; Stavroulaki et al., 2001). Ototoxic agents commonly impact the basal portion of the cochlea first, which is detected initially by monitoring EHF results (e.g., Sha et al., 2001). This highlights the importance of including EHF DPOAE measurements, if the goal is to detect the earliest signs of change in cochlear function (Dreisbach et al., 2017; Poling et al., 2019). Conventional and EHF DPOAE level repeatability were examined and previously reported for the four different groups utilized in this study, but noise floor and SNR variability were not explored.

DPOAE considerations across an extended bandwidth

The overall DPOAE levels were similar between the groups for both conventional and EHFs. The DPOAE level for the conventional frequencies tended to be relatively flat with a peak in the 4 to 6 kHz region. Then a gradual decline in DPOAE level as frequency increased was seen for the EHFs. The YA group had the most varied DPOAE levels across frequency. The noise floors also were similar between the groups with each having a valley in the 4 to 6 kHz region. In general, the SNR gradually increased as frequency increased until the 4 to 6 kHz region, then gradually decreased as frequency continued to increase. The peak in the DPOAE level and the subsequent valley in the noise floor coincide with increased SNR values in the 4 to 6 kHz region. For the studies that examined both conventional and EHFs, the average EHF DPOAE level was consistently less robust and had reduced average SNR values in comparison to conventional frequencies.

Effects of calibration on DPOAE variability

Individual differences in ear canal acoustics are not accounted for by traditional calibration techniques and can impact serial measurements. Traditional calibration techniques have been shown to alter DPOAE levels in adult human ear canals from 3 to 7 and ≥ 10 kHz because of the standing waves influence on the stimulus levels and response level measurement (Dreisbach & Siegel, 2001; Siegel, 1994). One possible way to reduce DPOAE variability across trials is to have better control over the stimulus levels being presented to the ear by using improved calibration techniques (Souza et al., 2014). One major difference between the studies examined was the calibration procedure. Traditional calibration techniques were used for the YA group, whereas depth-compensated calibration procedures were used for the other groups.

Across frequency (Figure 1), DPOAE levels appeared to be more similar for depth-compensated compared to traditional calibration methods, as has been reported previously (Dreisbach et al., 2018; Poling et al., 2014). As mentioned previously, each group exhibited an increase in DPOAE level, a decrease in noise, and subsequently an increase in SNR for the 4 to 6 kHz region. However, the peak in DPOAE levels from 4 to 6 kHz are more pronounced in the YA group compared to the YC and OC groups possibly suggesting less variability in levels across frequency due to improved calibration techniques. There also appears to be a less prominent effect for DPOAE levels ≥ 10 kHz when using depth-compensated calibration techniques. The YA group has a more defined increase in DPOAE level and SNR in the 10 to 12 kHz region compared to the YC, OC, and CF groups. A noticeable increase in average SNR difference-between-trials is prominent in that 10 to 12 kHz region for the YA group. Recall the cavity used for the depth-compensated simulator SPL method is based on impedance characteristics of adults and this could possibly influence the data collected in the OC and YC groups more so than the CF group. However, the errors that would occur without using the improved calibration method would be even greater.

Across trials (Figure 3), the average differences-between-trials for both DPOAE levels (dotted lines) and SNR values (dashed lines) are more similar for the low and high frequency regions when using the depth-compensated (YC and OC) compared to traditional (YA) methods for calibration. Additionally, the average differences-between-trials for the DPOAE levels in the 2 to 8 kHz region for YC (3.38 dB), OC (2.97 dB) and YA (3.15 dB) are more similar than for frequencies ≥ 8 kHz (YC=4.55 dB, OC=4.44 dB, YA=5.48 dB). This could reflect use of consistent probe placement reducing variability between trials when traditional calibration methods are used for frequencies ≤ 8 kHz. Depth-compensated calibration procedures were an early attempt to overcome these individual differences in ear canal acoustics and influence of standing waves by using an in-the-ear calibration technique (Souza et al., 2014). There are now other improved calibration techniques available that have further decreased the variability of measuring DPOAEs, but they still need further exploration (Scheperle et al., 2011; Souza et al., 2014). For example, the combination of forward pressure level (FPL) calibration and post hoc conversion to emitted pressure level (EPL) “largely eliminates standing-wave effects from DPOAE measurements, and thereby decreases spurious variability across repeated sessions” (Charaziak & Shera, 2017, p. 524).

Improving DPOAE test-retest variability in serial monitoring

Even though average DPOAE SNR data were repeatable with no significant differences between the four trials at all frequencies tested for the four different groups, SNR variability was greater than DPOAE level variability. At the conventional frequencies for YC and OC, the variability for the SNR values (7.94 and 7.43 for YC and OC, respectively) was more than two times the variability for the DPOAE levels (3.38 and 2.97 for YC and OC, respectively). This difference in variability between SNR values (4.57) and DPOAE levels (3.15) at the conventional frequencies was diminished in the YA group. This might have to do with the calibration methods employed where better control over stimulus levels were used with the children. At the EHFs the SNR variability (7.33 and 7.12 for YC and OC, respectively) is greater than one-and-one-half times that of the DPOAE level variability (4.55 and 4.44 for YC and OC, respectively). Again, this difference in variability between SNR values (6.52) and DPOAE levels (5.48) is less for the YA group at EHFs. The SNR variability (6.09) was three times greater than the variability for the DPOAE levels (1.96) at the EHFs for the CF group. To further illustrate this difference in variability between DPOAE levels and SNR values in the CF group, the 95% range of data for the absolute differences-between-trials for all frequencies and trial combinations was 6.26 dB for DPOAE levels compared to 15.63 dB for SNR values. When trying to identify the smallest changes in a measure to have confidence that a true physiological change has occurred, DPOAE levels are better suited for monitoring purposes.

The greater SNR variability observed in the OC and YC groups compared with the YA and CF groups could have been influenced by several factors. Because the DPOAE level repeatability amongst the groups is relatively similar, this suggests that the variability in the noise floor between sessions is driving the SNR variability differences. This increase in noise floor variability amongst children is predictable considering some of the implicit difficulties when testing children. Additionally, the CF group having the best SNR repeatability for the EHFs was anticipated as Dreisbach et al. (2018) explored various explanations for the improved repeatability of the DPOAE level compared to the YA group and determined that the improved calibration procedure, depth-compensated simulator SPL, was driving this improvement. They tested an additional group of young, normal-hearing adults with identical parameters used for the CF group, including the improved calibration methods, and reported average difference-between-trials (1.82 dB) similar to the CF group (1.96 dB).

As expected, DPOAE SNR had reduced repeatability when compared to the DPOAE level. The SNR is a calculation between the DPOAE level and the noise floor. Thus, it is influenced by both the repeatability of the DPOAE level and the variability of the noise floor, including the environmental and physiological noise occurring at the time of the recording. The combination of these factors makes the SNR inherently less repeatable than the DPOAE level. The noise floor is comprised of both intrinsic and extrinsic noise (e.g., Kimberley et al., 1997). By keeping the equipment, calibration, and software consistent between recordings, the intrinsic influences can be more constant and therefore minimized. However, certain extrinsic influences are harder to control. Obtaining measurements in a sound treated room can help reduce extrinsic noise variability, but this environment for testing may not be feasible. In summary, improved calibration appears to reduce DPOAE level variability and the ability to reduce background noise will aid SNR repeatability.

Conclusions and future directions

SNR is a valuable consideration and should be used to determine if DPOAE responses are interpretable. DPOAE SNR values across the full bandwidth of frequencies tested in the YC, OC, and YA groups resulted in the greatest values at approximately 4 to 6 kHz. Improved (depth-compensated) calibration methods result in SNR difference-between-trials that are similar across all frequencies. Clinically, it is critical to understand how the repeatability of various DPOAE measures impact the potential to assess and monitor cochlear function to promote earliest detection of dysfunction. Average SNR differences-between-trials across all frequencies are not statistically different but are greater than those for DPOAE levels. When monitoring cochlear health, DPOAE levels show less variability across trials than SNR values for the populations examined. Future studies and clinical applications should rely on improved calibration techniques for more accurate control over the stimulus levels being presented (e.g., FPL) and the recorded DPOAE response levels (e.g., EPL) as this has been shown previously to reduce DPOAE variability (Charaziak & Shera, 2017; Dreisbach et al., 2019; Dreisbach et al., 2018; Maxim et al., 2019; Poling et al., 2014). The current results imply that DPOAE levels, not SNR, should be used for monitoring in those exposed to ototoxicants or undergoing pharmaceutical interventions to characterize test-retest and to identify the earliest changes to cochlear health.

Supplementary Material

Supp Table 1
Supp Table 2
Supp Table 3

Acknowledgements

We are grateful to Amanda Conrad and Shelli Newman for their contributions to this project. Special thanks to past and present members of the Auditory Physiology and Psychoacoustics Laboratory, for data collection and entry. We also thank the Mayo Clinic Department of Otolaryngology-Head and Neck Surgery Research Committee for their support. Heartfelt thanks to the research participants and patients who participated. Portions of this work were presented at the 36th and 37th American Auditory Society Meeting.

Funding

Work supported by the National Institutes of Health [DC008195 PI-Dreisbach] and The San Diego Foundation Blasker-Rose-Miah Fund [PI-Dreisbach].

References

  1. Beattie RC, Kenworthy OT, & Luna CA (2003). Immediate and short-term reliability of distortion-product otoacoustic emissions. International Journal of Audiology, 42, 348–354. [DOI] [PubMed] [Google Scholar]
  2. Charaziak KK, & Shera CA (2017). Compensating for ear-canal acoustics when measuring otoacoustic emissions. Journal of the Acoustical Society of America, 141(1), 515. 10.1121/1.4973618 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Conrad A, & Dreisbach L (2011). Repeatability of high-frequency DPOAE measures in normal-hearing children. American Auditory Society Abstracts, 36, 42. [Google Scholar]
  4. Dorn PA, Piskorski P, Gorga MP, Neely ST, & Keefe DH (1999). Predicting audiometric status from distortion product otoacoustic emissions using multivariate analyses. Ear and Hearing, 20(2), 149–163. [DOI] [PubMed] [Google Scholar]
  5. Dreisbach L, Ho M, Reid E, & Siegel J (2017). Effects of Oxaliplatin, Carboplatin, and Cisplatin Across Treatment on High-Frequency Objective and Subjective Auditory Measures in Adults. Perspectives of the ASHA Special Interest Groups, 2(6), 17–36. [Google Scholar]
  6. Dreisbach L, Portugal N, Aguilar E, Hansen W, Kite D, McIntyre S, & Minasian C (2019). Acquisition and Repeatability of High-Frequency Distortion Product Otoacoustic Emissions Using Two Different Calibration Methods in Newborns. Applied Sciences, 9(15)(2947), 1–16. 10.3390/app9152947 [DOI] [Google Scholar]
  7. Dreisbach L, Zettner E, Chang Liu M, Meuel Fernhoff C, MacPhee I, & Boothroyd A (2018). High-Frequency Distortion-Product Otoacoustic Emission Repeatability in a Patient Population. Ear and Hearing, 39(1), 85–100. 10.1097/AUD.0000000000000465 [DOI] [PubMed] [Google Scholar]
  8. Dreisbach LE, Long KM, & Lees SE (2006). Repeatability of high-frequency distortion-product otoacoustic emissions in normal-hearing adults. Ear and Hearing (01960202), 27(5), 466–479. [DOI] [PubMed] [Google Scholar]
  9. Dreisbach LE, & Siegel JH (2001). Distortion-product otoacoustic emissions measured at high frequencies in humans. Journal of the Acoustical Society of America, 110(5), 2456–2469. [DOI] [PubMed] [Google Scholar]
  10. Franklin DJ, McCoy MJ, Martin GK, & Lonsbury-Martin BL (1992). Test/retest reliability of distortion-product and transiently evoked otoacoustic emissions. Ear and Hearing, 13(6), 417–429. [DOI] [PubMed] [Google Scholar]
  11. Gaskill SA, & Brown AM (1990). The behavior of the acoustic distortion product, 2f1-f2, from the human ear and its relation to auditory sensitivity. Journal of the Acoustical Society of America, 88(2), 821–839. [DOI] [PubMed] [Google Scholar]
  12. Gorga MP, Neely ST, & Dorn PA (1999). Distortion product otoacoustic emission test performance for a priori criteria and for multifrequency audiometric standards. Ear and Hearing, 20(4), 345–362. 10.1097/00003446-199908000-00007 [DOI] [PubMed] [Google Scholar]
  13. Katbamna B, Homnick DN, & Marks JH (1999). Effects of chronic tobramycin treatment on distortion product otoacoustic emissions. Ear and Hearing, 20(5), 393–402. [DOI] [PubMed] [Google Scholar]
  14. Kemp DT, Bray P, Alexander L, & Brown AM (1986). Acoustic emission cochleography--practical aspects. Scandinavian Audioliolgy Supplementum, 25, 71–95. [PubMed] [Google Scholar]
  15. Kimberley BP, Brown DK, & Allen JB (1997). Distortion Product Emissions and Sensorineural Hearing Loss. In Robinette MS & Glattke TJ (Eds.), Otoacoustic emissions: Clinical applications (pp. 181–204). Thieme. [Google Scholar]
  16. Knight KR, Kraemer DF, Winter C, & Neuwelt EA (2007). Early changes in auditory function as a result of platinum chemotherapy: use of extended high-frequency audiometry and evoked distortion product otoacoustic emissions. Journal of Clinical Oncology, 25(10), 1190–1195. 10.1200/JCO.2006.07.9723 [DOI] [PubMed] [Google Scholar]
  17. Konrad-Martin D, Poling GL, Dreisbach LE, Reavis KM, McMillan GP, Lapsley Miller JA, & Marshall L (2016). Serial Monitoring of Otoacoustic Emissions in Clinical Trials. Otology and Neurotology, 37(8), e286–294. 10.1097/MAO.0000000000001134 [DOI] [PubMed] [Google Scholar]
  18. Lonsbury-Martin BL, Harris FP, Stagner BB, Hawkins MD, & Martin GK (1990). Distortion product emissions in humans. I. Basic properties in normally hearing subjects. Annals of Otology, Rhinology, & Laryngology - Supplement, 147, 3–14. [PubMed] [Google Scholar]
  19. Maxim T, Shera CA, Charaziak KK, & Abdala C (2019). Effects of Forward- and Emitted-Pressure Calibrations on the Variability of Otoacoustic Emission Measurements Across Repeated Probe Fits. Ear and Hearing. 10.1097/AUD.0000000000000714 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Mulheran M, & Degg C (1997). Comparison of distortion product OAE generation between a patient group requiring frequent gentamicin therapy and control subjects. British Journal of Audiology, 31, 5–9. [DOI] [PubMed] [Google Scholar]
  21. Neely S, & Stevenson R (2002). SysRes. Tech Memo No. 1 (Boys Town National Research Hospital, Omaha), 1–10. [Google Scholar]
  22. Neely ST, & Liu Z (1993). EMAV: Otoacoustic Emission Averager. Tech Memo No. 17 (Boys Town National Research Hospital, Omaha), 1–26. [Google Scholar]
  23. Newman S, & Dreisbach L (2012). Repeatability of high-frequency behavioral and DPOAE measures in normal-hearing children. . American Auditory Society Abstracts, 37, 57. [Google Scholar]
  24. Poling G, Vlosich B, & Dreisbach L (2019). Emerging Distortion Product Otoacoustic Emission Techniques to Identify Preclinical Warning Signs of Basal Cochlear Dysfunction Due to Ototoxicity. Applied Sciences, 9 (15)(3132), 1–23. 10.3390/app9153132 [DOI] [Google Scholar]
  25. Poling GL, Siegel JH, Lee J, Lee J, & Dhar S (2014). Characteristics of the 2f(1)-f(2) distortion product otoacoustic emission in a normal hearing population. Journal of the Acoustical Society of America, 135(1), 287–299. 10.1121/1.4845415 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Popelka GR, Osterhammel PA, Nielsen LH, & Rasmussen AN (1993). Growth of distortion product otoacoustic emissions with primary-tone level in humans. Hearing Research, 71(1-2), 12–22. [DOI] [PubMed] [Google Scholar]
  27. Probst R, Lonsbury-Martin BL, & Martin GK (1991). A review of otoacoustic emissions. [Review]. Journal of the Acoustical Society of America, 89(5), 2027–2067. [DOI] [PubMed] [Google Scholar]
  28. Reavis KM, McMillan GP, Dille MF, & Konrad-Martin D (2015). Meta-Analysis of Distortion Product Otoacoustic Emission Retest Variability for Serial Monitoring of Cochlear Function in Adults. Ear and Hearing, 36(5), e251–260. 10.1097/AUD.0000000000000176 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Ress BD, Sridhar KS, Balkany TJ, Waxman GM, Stagner BB, & Lonsbury-Martin BL (1999). Effects of cis-platinum chemotherapy on otoacoustic emissions: The development of an objective screening protocol. Third place--Resident Clinical Science Award 1998. Otolaryngology - Head and Neck Surgery, 121(6), 693–701. [DOI] [PubMed] [Google Scholar]
  30. Richmond SA, Kopun JG, Neely ST, Tan H, & Gorga MP (2011). Distribution of standing-wave errors in real-ear sound-level measurements. Journal of the Acoustical Society of America, 129(5), 3134–3140. 10.1121/1.3569726 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Roede J, Harris FP, Probst R, & Xu L (1993). Repeatability of distortion product otoacoustic emissions in normally hearing humans. Audiology, 32(5), 273–281. [DOI] [PubMed] [Google Scholar]
  32. Scheperle RA, Goodman SS, & Neely ST (2011). Further assessment of forward pressure level for in situ calibration. Journal of the Acoustical Society of America,, 130(6), 3882–3892. 10.1121/1.3655878 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Sha SH, Taylor R, Forge A, & Schacht J (2001). Differential vulnerability of basal and apical hair cells is based on intrinsic susceptibility to free radicals. Hearing Research, 155(1-2), 1–8. [DOI] [PubMed] [Google Scholar]
  34. Siegel JH (1994). Ear-canal standing waves and high-frequency sound calibration using otoacoustic emission probes. Journal of the Acoustical Society of America, 95(5), 2589–2597. [Google Scholar]
  35. Siegel JH, & Hirohata ET (1994). Sound calibration and distortion product otoacoustic emissions at high frequencies. Hearing Research, 80(2), 146–152. [DOI] [PubMed] [Google Scholar]
  36. Smurzynski J. (1994). Longitudinal measurements of distortion-product and click-evoked otoacoustic emissions of preterm infants: preliminary results. Ear and Hearing, 15(3), 210–223. [DOI] [PubMed] [Google Scholar]
  37. Souza NN, Dhar S, Neely ST, & Siegel JH (2014). Comparison of nine methods to estimate ear-canal stimulus levels. Journal of the Acoustical Society of America,, 136(4), 1768. 10.1121/1.4894787 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Stavroulaki P, Apostolopoulos N, Segas J, Tsakanikos M, & Adamopoulos G (2001). Evoked otoacoustic emissions- An approach for monitoring cisplatin induced ototoxicity in children. International Journal of Pediatric Otorhinolaryngology, 59, 47–57. [DOI] [PubMed] [Google Scholar]
  39. Whitehead ML, Lonsbury-Martin BL, & Martin GK (1993). The influence of noise on the measured amplitudes of distortion-product otoacoustic emissions. Journal of Speech and Hearing Research, 36(5), 1097–1102. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp Table 1
NIHMS1917899-supplement-Supp_Table_1.xlsx (17.3KB, xlsx)
Supp Table 2
NIHMS1917899-supplement-Supp_Table_2.xlsx (17.3KB, xlsx)
Supp Table 3
NIHMS1917899-supplement-Supp_Table_3.xlsx (17.4KB, xlsx)

RESOURCES