J Am Acad Audiol 25:746-759 (2014)

Modeling DPOAE Inpnt/Output Function Compression: Comparisons with Hearing Thresholds DOI: 10.3766/jaaa.25.8.5 Shaum P. Bhagat*

Abstract Background: Basilar membrane input/ojtput (I/O) functions in mammalian animal models are character­ ized by linear and compressed segments when measured nearthe location corresponding to the character­ istic frequency. A method of studying basilar membrane compression indirectly in humans involves measuring distortion-product otoacoustic emission (DPOAE) I/O functions. Previous research has linked compression estimates from behavioral growth-of-masking functions to hearing thresholds.

Purpose: The aim of this study was to compare compression estimates from DPOAE I/O functions and hearing thresholds at 1 and 2 kHz.

Research Design: A prospective correlational research design was performed. The relationship between DPOAE I/O function compression estimates and hearing thresholds was evaluated with Pearson productmoment correlations.

Study Sample: Normal-hearing adults (n = 16) aged 2 2 -4 2 yr were recruited. Data Collection and Analysis: DPOAE I/O functions (L2 = 4 5 -7 0 dB SPL) and two-interval forcedchoice hearing thresholds were measured in normal-hearing adults. A three-segment linear regression model applied to DPOAE I/O functions supplied estim ates of com pression thresholds, defined as breakpoints between linear and compressed segments and the slopes of the compressed segments. Pearson productmoment correlations between DPOAE compression estimates and hearing thresholds were evaluated. Results: A high correlation between DPOAE compression thresholds and hearing thresholds was observed at 2 kHz, but not at 1 kHz. Compression slopes also correlated highly with hearing thresholds only at 2 kHz. Conclusions: The derivation of cochlear compression estimates from DPOAE I/O functions provides a means to characterize basilar membrane mechanics in humans and elucidates the role of compression in tone detection in the 1 -2 kHz frequency range.

Key Words: Cochlea, hearing, otoacoustic emissions, sensory thresholds Abbreviations: BM = basilar m em brane; DPO AE = distortion-product otoacoustic em ission; GOM = grow th-of-m asking; I/O = input/output; OHC = outer hair cell; SNR = signal-to-noise ratio; TM C = tem poral m asking curve

INTRODUCTION tudies examining basilar membrane (BM) input/ output (I/O) functions near the base of the healthy cochlea in mammals have revealed that for tones one-half octave below to one-quarter octave above the characteristic frequency for a given location, the functions

S

can be characterized by two or three segments that are separated by transitions in response growth (Cooper and Rhode, 1992; Nuttall and Dolan, 1996; Ruggero et al, 1997; Rhode and Recio, 2000). The first segment, reflect­ ing response growth at low stimulus levels, is linear with a slope near 1 dB/dB. The second segment, reflecting re­ sponse growth at moderate stimulus levels, is compressed

* Hearing Science Laboratory, School of Communication Sciences and Disorders, University of Memphis, Memphis, TIM Shaum P. Bhagat, Ph.D., Hearing Science Laboratory, School of Communication Sciences and Disorders, University of Memphis, 807 Jefferson Ave, Memphis, TIM; E-mail: [email protected] Portions of this study were presented in “ Estimates of human cochlear compression from distortion-product otoacoustic emission input-output functions and tone detection,” the 6th Meeting of the MidSouth Chapter of the Acoustical Society of America, Conway, AR, March 12-13, 2010.

74B

M odeling DPOAE C om pression/B hagat

and nonlinear, with a slope of less than 1 dB/dB. In certain cases, a third segment with a slope returning to 1 dB/dB is observed reflecting linear growth of the response for the highest stimulus levels. BM I/O functions with linear and compressed segments have also been measured at more apical locations, although the compression is less pronounced and more widely distributed than in basal locations (Cooper and Rhode, 1997; Rhode and Recio, 2000). Theoretical and experimental evidence have im pli­ cated outer h a ir cells (OHCs) as contributing to the compressive growth functions seen on the BM. Cochlear mechanical models have indicated th a t compression derives from nonlinear transduction properties of OHCs (Preyer and Gummer, 1996; de Boer and Nuttall, 2000; Stasiunas et al, 2003). Quinine, a drug known to alter OHC motility (Karlsson and Flock, 1990), causes the loss of sensitivity and linearization of BM responses after it is delivered intravenously (Ruggero et al, 1996). Adminis­ tration of furosemide abolishes the endocochlear poten­ tial, reduces OHC receptor potentials, and linearizes BM growth functions (Ruggero and Rich, 1991; Robles and Ruggero, 2001). Direct observation of BM mechanics in hum ans is not currently possible. However, considerable indirect psy­ chophysical evidence indicates th at the hum an periph­ eral auditory mechanism exhibits compression under certain conditions, and th at compression is linked to normal hearing (Oxenham and Bacon, 2003). A behav­ ioral method used to infer BM compression in humans initially reported by Oxenham and Plack (1997) has provided support for this view. This method compares growth-of-masking (GOM) functions for on-frequency narrow-band maskers th at are near to the probe fre­ quency to GOM functions for off-frequency maskers th at are below the probe frequency. More recently, tem­ poral masking curves (TMCs) obtained with forward masking have also been used to infer BM compression in hum an listeners (Nelson et al, 2001; Johannesen and Lopez-Poveda, 2008; 2010; Lopez-Poveda and Johannesen, 2012). A TMC depicts the level of a narrow-band masker required to mask the probe as a function of the interval between the masker and probe. In both GOM and TMC procedures, masker and probe frequencies are configured to reveal linear and compressive BM responses. Several studies using these behavioral measures have demonstra­ ted that some listeners with sensorineural hearing loss exhibit more linearized masking functions than listeners with normal hearing for the masker-probe conditions expected to evoke a compressive BM response (Oxenham and Plack, 1997; Rosengard et al, 2005; Stainsby and Moore, 2006). In other listeners with sensorineural hear­ ing loss, behavioral estimates of BM compression are sim­ ilar to estimates of BM compression in normal-hearing listeners (Plack et al, 2004; Lopez-Poveda et al, 2005). The individual differences in behavioral estimates of BM compression seen in listeners with sensorineural

hearing loss may be attributed to differential amounts of OHC loss in these listeners, with individuals with lin­ earized BM growth functions possibly exhibiting greater OHC damage (Lopez-Poveda et al, 2005; Lopez-Poveda and Johannesen, 2012). Another technique developed to indirectly estimate BM compression in hum ans involves m easurement of distortion-product otoacoustic emission (DPOAE) I/O functions (Neely et al, 2003; Gorga et al, 2007). Believed to partially originate from the nonlinear transduction properties of OHCs and partially to originate from linear reflection mechanisms (Shera and Guinan, 1999; Shera, 2004), DPOAEs are evoked by the simultaneous presen­ tation of two primary tones (/) & f 2) f o > f \ ) to an ear and are manifested as low-level sounds in the ear canal. They occur a t distortion-product frequencies predicted by mathematical combinations of the two primary tones. The DPOAE I/O function is usually constituted from a plot of the 2 / 1 - /2 DPOAE amplitude (in dB SPL) as a func­ tion of the level (in dB SPL) of one of the primary tones. Direct comparisons of BM I/O functions and 2 f , -f2 DPOAE I/O functions measured in chinchillas have indi­ cated similarities in growth patterns and comparable stimulus intensities where notches in the functions oc­ curred (Rhode, 2007). Group DPOAE I/O functions in normal-hearing humans exhibit linear and compressed segments that are reminiscent of BM I/O functions in lab­ oratory mammals (Dorn et al, 2001; Boege and Janssen, 2002). These findings suggest th a t m easurem ent of DPOAEs over a range of stim ulus levels provides a reasonable estimate of the growth of BM responses in humans. Several studies have com pared behavioral and DPOAE estimates of BM compression from the same human cochleae. Johannesen and Lopez-Poveda (2008) found that there was good correspondence between TMC-based and DPOAE-based estimates of compression at 4 kHz but not at 1 or 0.5 kHz. They attributed these findings to a high number of notches and plateaus in DPOAE I/O functions at the lower frequencies. They reported th at the notches and plateaus were present in the DPOAE I/O functions but were absent from the TMC functions, and this possibly compromised the ability of their model to provide accurate compression estimates. In a follow-up study, Johannesen and Lopez-Poveda (2010) found that optimizing the primary-tone levels used to measure DPOAE I/O functions on an individual basis did not reduce the number of notches and plateaus at 1 and 0.5 kHz. They confirmed the results of their earlier study, indicating that DPOAE and TMC-based estimates of compression were most similar at 4 kHz. In contrast to previous studies, Rodriguez et al (2011) compared the slopes of forward masked psychometric functions with DPOAE I/O function parameters at 1 and 6 kHz and found that these estimates of BM compression correlated only weakly at these frequencies. Investigation of the

747

J o u r n a l o f th e A m erica n A cad em y o f A u d io lo g y /Volume 25, Number 8, 2014

relationship between behavioral estim ates of BM com­ pression and hearing thresholds for tones has been pro­ vided by two studies conducted in norm al-hearing adults (Dubno et al, 2007; Horwitz et al, 2007). These investi­ gators m easured off-frequency GOM functions to esti­ m ate BM compression and applied a three-segm ent linear regression model (Yasin and Plack, 2003; Plack et al, 2004) to determ ine breakpoints between the linear and compressed segm ents of their functions and the slopes of the compressed segments. The lower break­ points separating the low-level linear segm ent from the compressed middle segm ent are conceptually equiv­ alent to compression thresholds. Overall, compression thresholds correlated significantly w ith hearing th resh ­ olds a t 2 and 4 kHz. The slopes of the compressed middle segm ents did not correlate significantly w ith the hearing thresholds. The results of these studies indicate th a t in individuals w ith lower hearing thresholds, BM compres­ sion is initiated a t lower input levels th an in participants with higher hearing thresholds, findings consistent with predictions from models of compression (Yates, 1990; Plack et al, 2004). The preponderance of evidence from previous research indicates th a t behavioral and DPOAE estim ates of BM compression are more sim ilar a t frequencies > 1 kHz, and are less sim ilar a t frequencies s i kHz. There are unresolved questions concerning the relationship between DPOAE-based estim ates of BM compression and hearing sensitivity. The purpose of the present study was to exam­ ine if DPOAE compression threshold estim ates from the same three-segment linear regression model used in pre­ vious studies using behavioral estim ates of BM compres­ sion (Dubno et al, 2007; Horwitz et al, 2007) would correlate with hearing thresholds at 1 and 2 kHz. A sec­ ondary purpose was to examine if DPOAE compression slopes from the model would correlate with hearing thresholds a t these frequencies. As the questions motivat­ ing th is stu d y w ere p rim arily clinical in n a tu re , pro­ cedures w ere chosen th a t h ad th e h ig h e st clinical relevance. As such, the hearing threshold m icrostructure was not m easured because this technique is not often used clinically. DPOAE fine structure was not m easured, as m any clinical DPOAE instrum ents do not offer suffi­ cient frequency resolution to m easure DPOAE fine struc­ tu re an d th ese m easu rem en ts have not com pletely tra n sitio n e d into clinical use.

METHODS Participants The participants were 16 hum an adults (14 females, 2 males). Participant age ranged from 22-42 yr (mean age = 27.6 yr, SD = 5.9 yr). Participant hearing was screened w ith a calib rated au diom eter (ANSI S3.6; A m erican N atio n al S ta n d ard s In s titu te [ANSI], 2004) before

748

participant enrollment. All participants adm itted into the study had hearing thresholds a t or better th a n 20 dB HL in both ears for the standard audiometric test fre­ quencies m easured a t interoctave intervals from 0.25-8 kHz. A calibrated m iddle-ear analyzer (ANSI S3.39; ANSI, 1987) was used to evaluate m iddle-ear function. Tympanometric evaluation, measured with a single probetone frequency (226 Hz), indicated norm al m iddle-ear function in both ears of each participant. The m ean static acoustic adm ittance across participants was 0.8 mL (SD = 0.4) in the rig h t ears and 0.7 mL (SD = 0.3) in the left ears. The m ean tym panom etric peak pres­ sure in decaPascals (daPa) was 3.8 daPa (SD = 12.9) in the right ears and 1.6 daPa (SD = 15.7) in the left ears. Middleear muscle reflex thresholds, obtained with a pulsed broad­ band noise activator presented to the ear contralateral to the test ear, were at or greater than 60 dB HL in both ears of every participant. Mean middle-ear muscle reflex thresh­ olds were 77.2 dB HL (SD = 9.6) in the right ears and 81.9 dB HL (SD = 14.4) in the left ears. Participants signed a consent form approved by the Institutional Review Board at the University of Memphis before participating in the study.

Stim uli and Apparatus H earing thresholds and DPOAE I/O functions were m easured in each participant a t two te st frequencies. All m easurem ents occurred in the sam e double-walled, sound-treated enclosure.

H earing Thresholds The te st frequencies selected for hearing threshold testing were 1 and 2 kHz. Thresholds were m easured w ith a Dell Optiflex GX 280 com puter interfaced w ith Tucker-Davis Technologies (TDT) System 3 hardw are and software. Tonal stim uli were digitally generated (TDT, R P2.1) a t a nom inal rate of 50 kHz. The electrical signals were sent to a headphone buffer (TDT, HB7) before being transduced by one of a pair of Sony MDR-V500 head­ phones. The duration of the tone signals was 300 msec, including 10 msec cosine-gated rise-and-fall times. Cali­ bration of stimulus levels was accomplished by placing the headphone on an acoustic coupler th a t was connected to a sound-level m eter (Larson Davis 800B). Frequency accuracy of the tonal signals was verified by the frequency counter of the sound-level meter. The frequency response of the headphones indicated th at the output level at 1 kHz was within 1 dB of the output level at 2 kHz.

DP I/O Functions The p rim ary -to n e stim u li w ere g e n e ra te d by an otoacoustic emissions analyzer (Otodynamics ILO 296) th a t was interfaced w ith th e sam e com puter used for

M odeling DPOAE C om pression/B hagat

threshold testing. Primary tones were routed to the ILO probe assembly. DPOAE I/O functions were acquired at f2frequencies of 1 and 2 kHz in each participant. The fjf\ ratio was fixed at 1.22. Primary-tone levels at the higherfrequency primary (L2) were incremented in 5 dB steps and were presented at targeted levels from 45-70 dB SPL. These primary levels fall within the compressive region of DPOAE PO functions and correspond well with primary levels chosen (40-65 dB SPL) in a previous study comparing behavioral and DPOAE estimates of BM compression (Johannesen and Lopez-Poveda, 2010). Primary-tone levels at the lower-frequency primary (Li) were calculated using the formula {L\ = 0.4L2+ 39) devel­ oped by Kummer et al (1998). Selection of these interpri­ mary level differences was based on findings indicating that these parameters produce both high-level DPOAEs across a wide f2 frequency range and DPOAE I/O functions th a t have good correspondence with BM responses in anim al models (Kummer et al, 2000; Boege and Janssen, 2002). Johannesen and LopezPoveda (2010) also found th a t the highest correspond­ ence between behavioral and DPOAE compression indices occurred for prim ary-tone levels set using the Kum m er et al (1998) formula.

was 5 dB, and the step size was decreased to 2 dB follow­ ing the first three reversals. Participants voted by using a mouse to click on the selected interval icon on a com­ puter monitor. Feedback was provided to the participant after each response indicating the interval that contained the signal tone. Each threshold run consisted of 50 trials, and three threshold runs were obtained for each measure­ ment condition. Threshold was defined as the average of the estimates obtained from the three runs. DPOAE I/O F u n c tio n s

The order of measurement conditions was counter­ balanced. Hearing thresholds were measured in oddnum bered participants first, followed by DPOAE measurements. In even-numbered participants, DPOAE measurements were completed first, followed by mea­ surement of hearing thresholds. The experimental ses­ sion at a given test frequency typically was completed in approximately 2 hr. In order to avoid the effects of fatigue, participants returned on another day to complete testing a t the rem aining test frequency. The interval between testing days was typically 1-2 days. Prelim inary tests, such as hearing screening and tympanometry, were completed on the second test­ ing day and replicated findings seen on the first testing day. Participants were cautioned to avoid exposure to loud sounds between testing days.

Before data collection, a checkfit procedure was con­ ducted in each participant. During the checkfit proce­ dure, the ILO probe was placed in the ear canal of the right ear and a repeated click stimulus was intro­ duced into the ear canal in order to provide an estimate of ear canal size. After this estimate was obtained, sus­ tained tones or chirps were delivered through the probe in order to measure the frequency response of the indi­ vidual ear canals. On the basis of these measurements, the levels of the primary tones in the ear canal were set to match the targeted primary-tone levels. It is known th at setting of the primary-tone level in the ear canal can be influenced by standing waves, but the effects of standing waves at the frequencies used in this study were expected to be minimal (Siegel, 1994). After com­ pletion of the checkfit procedure, two consecutive mea­ surements of DPOAE I/O functions were made at each f2 frequency. These measurements were separated in time by approximately 1 min intervals. The measurements began at the highest primary-tone levels, and descended in level until the lowest targeted primary-tone level was reached. Each measurement was terminated after three complete primary-tone level sweeps. The probe remained in the ear canal during the two measurements, and care was taken to avoid displacing the probe by advising the participants to remain still and quiet during the DPOAE procedure. The levels of the 2f\-f2DPOAE measured dur­ ing the two m easurem ents were recorded in each p ar­ ticipant. Estim ates of the noise floor were made by determining the mean level (in dB SPL) of 5 frequency bins above and below the 2f\-f2frequency bin.

H ea rin g T h resh o ld s

S ystem D isto rtio n

In order to reduce ear effects, only right ears were tested. Estimation of threshold was provided by a twointerval, two-alternative forced-choice procedure with a two-down one-up adaptive rule that tracked the 71% correct performance level in each participant. The forced-choice procedure was selected because it produces less response bias than conventional clinical procedures (Marshall and Jesteadt, 1986). The observation intervals were 300 msec in duration, and the time between inter­ vals was 350 msec. The initial step size of the procedure

For assessment of intermodulation distortion gener­ ated by the otoacoustic emissions analyzer, the ILO probe was placed in a 1 cc cavity th at was provided by the m anufacturer and a syringe with a volume of 1.5 cc before data collection. Assessment of system intermodulation distortion for primary-tone levels from 45-80 dB SPL at the f2frequencies of 1 and 2 kHz was carried out. Measurements in the cavity and syringe were repeated three times. The recorded levels at the 2/1-/2 frequency were tabulated for both the cavity

P ro ced u re

743

J o u rn a l o f th e A m erica n A cad em y o f Audiology./Volume 25, Number 8, 2014

and syringe, and these data were then averaged to pro­ vide an index of system distortion. Table 1 displays the averaged levels of system distortion for primary-tone levels from 45-80 dB SPL. C o m p ression E stim a tes

Compression estimates were derived offline without prior knowledge of participant hearing sensitivity from DPOAE I/O functions fit with the three-segment linear regression model (Yasin and Plack, 2003; Plack et al, 2004). In order to be considered for the fitting proce­ dure, the DPOAE I/O functions were required to have DPOAE signal-to-noise ratios (SNRs) of 3 dB or higher at a minimum of 3 consecutive points on the functions. In addition, DPOAE levels at all points on the I/O func­ tions were required to exceed the average levels of system distortion by at least 3 dB. These criteria are sim­ ilar to those reported by previous studies (Johannesen and Lopez-Poveda, 2008; Neely et al, 2009). DPOAE I/O functions at f 2 =1 kHz from participants S04, S08, and S10 and at f2 = 2 kHz from participants S02, S06, and S08 did not met these criteria and were omitted from further analysis. The three-segment linear regression model (Yasin and Plack, 2003; Plack et al, 2004) contains linear-compressedlinear segments that are representative of BM PO func­ tions. The low-level linear segment and compressed midlevel segm ent were joined by a lower breakpoint (BPx), and the compressed midlevel segment and high-level linear segment were joined by an upper breakpoint (BP2) Lout = Lin + G (Lin —B Pi),

(1)

Lout = c Lin + /-!+ G (BP1BP2),

(3)

where G is the gain (dB), c is the slope of the compressed segment (dB/dB), k x = BPx (1-c), k2 = BP2 (c-1) + k 1} Lin (level in dB SPL) and Lout (level of BM response, dB). The free param eters in the model, G, c, BP1; and BP2, were adjusted using the fminsearch function in MATLAB to provide the best fit (lowest rms error value) to the data. On the basis of the fitted functions, model Table 1. Average Levels of System Distortion for Each L2 Level

estimates of lower and upper breakpoints and the slope of the compressed segments were provided. For assessment of the variability of estimated breakpoints and slopes pro­ vided by the model, model fits were applied independently to the two DPOAE I/O functions measured at each f2 fre­ quency in each of the remaining participants. Pairedsamples /-tests were conducted on model estimates of the lower breakpoints, upper breakpoints, and slopes of the compressed segments for the first and second measure­ ments. No significant differences (p > 0.05) between the first and second measurements were detected in model esti­ mates of the lower breakpoints, upper breakpoints, and slopes of the compressed segments at 1 kHz or at 2 kHz. The levels of the 2f x-f2 DPOAE measured during the two measurements were then averaged in each partic­ ipant. Estimates of the noise floor were made by deter­ mining the mean level (in dB SPL) of 5 frequency bins above and below the 2/j-/’2 frequency bin. Noise floor estimates from the two measurements were also aver­ aged together in each participant. The model was then applied to the averaged DP PO functions in each partic­ ipant. In the case of one DPOAE I/O function at 1 kHz and one DPOAE I/O function at 2 kHz, the model pro­ vided an estimate of the lower breakpoint th at was below the range of input levels examined. Model esti­ mates from the fitted functions from these participants (S16 at 1 kHz and S07 at 2 kHz) were excluded from further statistical analysis. The average rms error value of the remaining 24 fitted functions was 0.56 dB, and there was one function with rms error values from 1.02.0 dB and one function with rms error values >2.0 dB. Although the model provided estimates of the upper break­ point of the functions, only estimates of the lower b rea k ­ points (L2 level in dB SPL a t lower breakpoint = compression threshold) and the slopes of the compressed segments were included in the statistical analysis, as these were the variables of interest and were intended to be compared with hearing thresholds at 1 and 2 kHz in this study. In order to examine if extending the low-level linear segment of averaged DPOAE I/O functions below the lowest measured L2 level of 45 dB SPL affected the comparison with hearing thresholds, a linear interpolation procedure was used involving pairs of data (x t, y t) where i = 0 ,1 ... n - 1 and y (DPOAE level) is interpolated from x (L2 level) for

L2 Level (dB SPL)

1 kHz

2 kHz

80

-1 9 .7

-2 0 .0

75

- 1 7 .2

-2 1 .7

x > x n- 1y = y n_1+ yn- 1 ~ yn~2 X(x - s„-i)

70

- 2 2 .6

-2 4 .1

%n—1

65

-1 9 .3

-2 3 .1

60

-1 8 .3

-2 5 .9

55

-2 4 .1

-2 2 .0

50

-2 0 .1

-2 3 .0

45

-2 2 .6

-2 2 .3

Note: Each L2 level is measured at f2 = 1 kHz and fz = 2 kHz.

750

X < X 0, y = y 0+ —— — X (x-xo) X]_ Xo

(4)

(5)

%n—2

(y. , , — y.,

X i < x< X i + i y = y i + f 1------- - X ( x - x i ) V*'i+1 %i)

(6)

Interpolated DPOAE levels for L2 levels at 40, 35, 30, and 25 dB SPL were used to extend the low-level lin ­ ear segm ents of the DPOAE I/O functions in each

M o d elin g DPOAE C om pression/B hagat

participant. The model was applied to the interpo­ lated DPOAE I/O functions, and estim ates of DPOAE compression thresholds and DPOAE compression slopes were obtained. Data A nalysis The variables evaluated included hearing thresholds measured in the quiet and compression estimates from fitted DPOAE I/O functions. Separate evaluations of thresholds and DPOAE compression estimates at the 1 and 2 kHz test frequencies were planned. Pearson product-moment correlations were calculated, which examined the association between (1) hearing thresh­ olds and DPOAE compression thresholds and (2) hear­ ing thresholds and DPOAE compression slopes.

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

RESULTS earing thresholds and DPOAE I/O functions were measured in all 16 participants. However, DPOAE I/O functions from 12 participants (aged 23—42 yr) met the study criteria at the f2 frequency of 1 kHz, and DPOAE functions from 12 participants (aged 22-42 yr) met the study criteria at the f2 frequency of 2 kHz. There were 9 participants meeting the study criteria for DPOAEs at both / 2 frequencies. Correlations between thresholds and DPOAE compression estimates were only evaluated for the 12 participants meeting the study requirements at each test frequency.

H

H earing Thresholds Hearing thresholds ranged from 7.4-15.5 dB SPL (mean = 11.0 dB SPL, SD = 2.6 dB SPL) at 1 kHz and from 5.0-19.4 dB SPL (mean = 11.2 dB SPL, SD = 4.6 dB SPL) at 2 kHz. DPOAE I/O Functions Individual DPOAE I/O functions at f2 = 1 kHz and f2 = 2 kHz are depicted in Figures 1-2 and Figures 3-4, respectively. In these figures, squares represent DPOAE levels and triangles represent noise levels. Each data point represents the average of two consecutive measure­ ments. These figures also depict model fits to the individ­ ual DPOAE PO functions (solid lines) and compression threshold estimates from the model (open circles). DPOAE levels were highly variable across individual participants. Several of the DPOAE PO functions at f2 = 1 kHz exhibited notches, and this was less commonly observed a tf2 = 2 kHz. The overall number of DPOAE PO functions exhibiting notches is listed in Table 2. Mean DPOAE lev­ els ranged from 2.5-5.1 dB SPL at 1 kHz and were from 2.3—7.4 dB SPL at 2 kHz. Mean DPOAE levels and mean noise levels at both frequencies are plotted in Figure 5.

F ig u r e 1. Individual DPOAE PO functions a t f2 = 1 kHz. S q u a re s re p re s e n t DPOAE levels, a n d tria n g le s re p re s e n t noise levels. M odel fits to th e in d iv id u a l DPOAE I/O fun ctions (solid lin es) a n d com pression th re sh o ld e stim a te s from th e m odel (open circles) a re also depicted. The rm s e rro r v alu es from th e m odel a re show n in th e u p p e r rig h t co rn e r o f each fig­ u re . P a rtic ip a n t n u m b e r is show n in th e low er rig h t co rn er of each figure.

The linear interpolation procedure successfully inter­ polated decreasing DPOAE levels for L2 levels of40, 35, 30 and 25 dB SPL in 10 of 12 participants at 1 kHz and 10 of 12 participants at 2 kHz. The linear interpolation pro­ cedure was not successful in interpolating decreasing

751

Journal of the American Academy of Audiology/Volume 25, Number 8, 2014

Com pression Estim ates Estimates of the lower breakpoints (compression thresholds) and slopes of the compressed segment from the DPOAE I/O functions were provided by the threesegment linear regression model. The compression thresh­ olds for the / 2 frequency at 1 kHz ranged from 45.0-58.1

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB S P L)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

Figure 2. Individual DPOAE I/O functions at f2 = 1 kHz. Squares represent DPOAE levels, and triangles represent noise levels. Model fits to the individual DPOAE I/O functions (solid lines) and compression threshold estimates from the model (open circles) are also depicted. The rms error values from the model are shown in the upper right corner of each figure. Participant number is shown in the lower right corner of each figure.

DPOAE levels in two participants at 1 kHz and two par­ ticipants at 2 kHz because of nonmonotonic DPOAE I/O functions in these participants. Examples of interpolated DPOAE I/O functions a t/ 2 = 1 kHz and 2 kHz are shown in Figures 6-7, respectively.

75S

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

Figure 3. Individual DPOAE I/O functions at f 2 = 2 kHz. Squares represent DPOAE levels, and triangles represent noise levels. Model fits to the individual DPOAE I/O functions (solid lines) and compression threshold estimates from the model (open circles) are also depicted. The rms error values from the model are shown in the upper right corner of each figure. Participant number is shown in the lower right comer of each figure.

Modeling DPOAE Compression/Bhagat

0.4 dB/dB (mean = 0.2 dB SPL, SD = 0.2 dB SPL) at fa = 2 kHz. Model fits applied to the interpolated DPOAE I/O functions revealed compression thresholds for the f 2 fre­ quency at 1 kHz th at ranged from 25.0-53.7 dB SPL (mean = 35.2 dB SPL, SD = 9.3 dB SPL) and ranged from 25.0-60.7 dB SPL (mean = 33.1 dB SPL, SD = 12.7 dB SPL) at f 2 = 2 kHz. The slope estimates for the interpolated DPOAE I/O functions ranged from -1.1 to 0.34 dB/dB (mean = -0.01 dB SPL, SD = 0.4 dB SPL) at f 2 = 1 kHz and from -0.01 to 0.48 (mean = 0.19 dB SPL, SD = 0.14 dB SPL) a t f 2 = 2 kHz. L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

L2 Level (dB SPL)

Figure 4. Individual DPOAE I/O functions at f2 = 2 kHz. Squares represent DPOAE levels, and triangles represent noise levels. Model fits to the individual DPOAE I/O functions (solid lines) and compression threshold estimates from the model (open circles) are also depicted. The rms error values from the model are shown in the upper right corner of each figure. Participant number is shown in the lower right corner of each figure.

dB SPL (mean = 48.5 dB SPL, SD = 3.7 dB SPL), and were from 45.0-55.0 dB SPL (mean = 48.1 dB SPL, SD = 3.4 dB SPL) at the f 2 frequency of 2 kHz. Slope estimates ranged from -1.7 to 0.2 dB/dB (mean = -0.2 dB SPL, SD = 0.5 dB SPL) a t /2 = 1 kHz and were from -0.1 to

C orrelations betw een H earing Thresholds and DPOAE Com pression Estim ates We calculated Pearson-product moment correlations to examine the correlation between DPOAE compres­ sion estimates and hearing thresholds. At 1 kHz, the correlation between DPOAE compression thresholds and hearing thresholds was not statistically significant (r = -0.52, p = 0.08). The correlation between DPOAE compression slope and hearing thresholds was also not statistically significant (r = 0.26, p = 0.42) at 1 kHz. However, statistically significant correlations between DPOAE compression thresholds and hearing thresh­ olds (r = 0.72,p < 0.01) and DPOAE compression slopes and hearing thresholds (r = 0.71, p < 0.01) at 2 kHz were detected. The correlations between DPOAE com­ pression estimates and hearing thresholds at 1 and 2 kHz are depicted in Figure 8. Correlations between hearing thresholds and compres­ sion estimates obtained from the interpolated DPOAE I/O functions were not higher than those seen for the com­ pression estim ates obtained from the DPOAE I/O functions w ith L 2 levels from 45-70 dB SPL. The correlation between interpolated DPOAE compression thresholds at f 2= 1 kHz and hearing thresholds at 1 kHz (r = -0.32, p = 0.38) and interpolated DPOAE com­ pression slopes and hearing thresholds (r = 0.14, p = 0.69) was not statistically significant. At f2 = 2 kHz, the correlation between interpolated DPOAE compres­ sion thresholds and hearing thresholds at 2 kHz was not statistically significant (r = 0.48, p = 0.16), and the correlation between interpolated DPOAE compres­ sion slopes and hearing thresholds (r = 0.62, p = 0.06) also was not statistically significant.

Table 2. Number of DPOAE I/O Functions Exhibiting Notches or Negative Slopes f2 Frequency (kHz) 1 2

Notches

Negative Slopes

7 5

2

7

753

Journal of the American Academy of AudioIogyA'olume 25, Number 8, 2014

but not at 1 kHz. The results of this investigation will be discussed in relationship to previous studies. C orrelations betw een H earing Thresholds and DPOAE I/O Function Com pression Thresholds

L2 Level (dB SPL)

L2 Level (dB SPL) Figure 5. Mean DPOAE levels (squares) and noise levels (trian­ gles) at f 2 = 1 kHz (top panel) and f 2 = 2 kHz (bottom panel).

DISCUSSION he aim of this study was to examine correlations between hearing thresholds and compression esti­ mates from DPOAE I/O functions. The findings from this study indicated th at statistically significant corre­ lations were observed between hearing thresholds and DPOAE compression estimates from DPOAE I/O func­ tions obtained for L2 levels from 45-70 dB SPL at 2 kHz,

T

■75a

The ranges of lower breakpoint or estimated com­ pression thresholds at both 1 and 2 kHz found by the three-segment linear regression model in this study (45.0-58.1 dB SPL) were consistent with previous stud­ ies th at fitted individual DPOAE I/O functions with dif­ ferent functions. Johannesen and Lopez-Poveda (2008) fit DPOAE I/O functions with third-order polynomials and defined the compression threshold as the input level at which the slope of the fitted polynomial was reduced from a value close to 1 at low input levels to 0.4 dB/dB at higher input levels. They reported DPOAE compression thresholds in the range of40-50 dB SPL at f2 frequencies of 1 and 2 kHz. Neely et al (2009) used a two-slope function to fit individual DPOAE I/O func­ tions and reported breakpoints between the low-level linear segment and the compressed segment that ranged between 40-50 dB SPL at most frequencies in partici­ pants with normal hearing, and the breakpoints in­ creased in level above 60 dB SPL for participants with sensorineural hearing loss. Trends in the data at 2 kHz (see Figure 8) indicated th at individuals with lower hearing thresholds had lower DPOAE compression thresholds and individuals with higher hearing thresholds had higher DPOAE com­ pression thresholds. This trend was verified statistically, as a significant correlation between hearing thresholds and DPOAE compression thresholds at 2 kHz was detected in the present study. This finding was consis­ tent with previous investigations linking compression thresholds derived from GOM functions to hearing thresholds (Dubno et al, 2007; Horwitz et al, 2007). These investigators measured the GOM functions with narrowband maskers presented from 45-80 dB SPL and used the same three-segment linear regression model used in the present study. They found a correlation between hearing thresholds and compression thresholds at 2 kHz, with individuals wdth lower hearing thresholds exhibit­ ing lower compression thresholds compared with indi­ viduals w ith higher hearing thresholds. The m ean DPOAE compression threshold at 2 kHz in the present study was 48.1 dB SPL, whereas compression thresholds ranged from 44.4—55.5 dB SPL at 2 kHz in the GOM studies (Dubno et al, 2007; Horwitz et al, 2007). Similar­ ities in compression threshold estimates derived from DPOAE I/O functions and GOM functions are not entirely unexpected, as it has been shown that compres­ sion threshold estimates from the two methods can be correlated at certain test frequencies (Johannesen and Lopez-Poveda, 2008). That the association between hear­ ing thresholds and DPOAE compression thresholds at

Modeling DPOAE Compression/Bhagat

Figure 6. DPOAE I/O functions at f2 = 1 kHz with interpolated DPOAE levels for L2 levels at 40, 35, 30, and 25 dB SPL in selected participants. Squares represent actual DPOAE levels elicited by L2 levels from 45-70 dB SPL. Stars represent interpolated DPOAE levels. Compression threshold estimates (open circles) from the original model fits are shown. Compression threshold estimates (open triangles) from the model fits to the interpolated DPOAE I/O functions are also depicted.

2 kHz observed in the present study replicated previous findings using psychoacoustic methodology' suggests that the mechanism linking hearing sensitivity to BM compression is robust in this frequency region. Unlike findings at 2 kHz, DPOAE compression thresholds and quiet thresholds at 1 kHz were not significantly associ­ ated. The majority of the participants exhibited DPOAE PO functions with plateaus or notches at 1 kHz. This trend may have influenced the fitting of the three-segment lin­ ear regression model to the DPOAE PO functions. The average rms error values were 0.74 dB at 1 kHz and 0.38 dB at 2 kHz. This suggested that the model provided better fits to the data at 2 kHz than at 1 kHz, perhaps because of the greater number of nonmonotonic functions at the lower test frequency. Therefore, estimates of com­ pression thresholds derived from the fitted functions may have been less accurate at 1 kHz than at 2 kHz. In order to determine if extending the low-level l i n e a r portion of the DPOAE I/O function affected the associ­ ation between DPOAE compression thresholds and hearing thresholds, DPOAE levels were interpolated from the existing functions for L2 levels at 40, 35, 30, and 25 dB SPL. Boege and Janssen (2002) and Gorga et al (2003) measured DPOAEs using L2 levels from 20-65 dB SPL and fit their DPOAE PO data with linear functions designed to extrapolate the DPOAE thresh­ old. Although the DPOAE compression thresholds in this study estimated by the model from interpolated DPOAE I/O functions tended to be lower than compres­

sion thresholds from noninterpolated DPOAE PO func­ tions, the associations between hearing thresholds and DPOAE compression thresholds were not strengthened at either test frequency. Boege and Janssen (2002) measured 4236 DPOAE PO functions in 30 normal­ hearing ears and 119 ears with sensorineural hearing loss, and found a close correspondence (r = 0.65) between extrapolated DPOAE thresholds and hearing thresholds. Gorga et al (2003) measured DPOAE PO functions from as many as 278 ears obtained from 97 normal-hearing participants and 130 participants with sensorineural hearing loss. They found th a t correlations between DPOAE thresholds and hearing thresholds varied from as low as r = 0.49 at 0.75 kHz to as high as r = 0.85 at 4 kHz. Differences in results between this study and the aforementioned studies may be related to differences in psychophysical procedures. Hearing thresholds in the Boege and Janssen (2002) study were measured using a method of adjustment with 1 dB precision, whereas the study by Gorga et al (2003) measured hearing thresholds with a clinical procedure with 5 dB precision. The present study used a forced-choice procedure with 2 dB precision. Like the Gorga et al (2003) study, DPOAEs and hearing thresholds were measured using different transducers in the present study, whereas DPOAEs and hearing thresholds were measured using the same transducer in the Boege and Janssen (2002) study. It is logical to assume that a closer correspondence between hearing thresholds and low-level DPOAE thresholds would occur when both

755

Journal of the American Academy of Audiology/Volume 25, Number 8, 2014

Figure 7. DPOAE I/O functions at f 2 = 2 kHz with interpolated DPOAE levels for L 2 levels at 40, 35, 30, and 25 dB SPL in selected participants. Squares represent actual DPOAE levels elicited by levels from 45—70 dB SPL. Stars represent interpolated DPOAE levels. Compression threshold estimates (open circles) from the original model fits are shown. Compression threshold estimates (open triangles) from the model fits to the interpolated DPOAE I/O functions are also depicted.

are measured by the same earphone. The value of the correlations between DPOAE compression estimates obtained from DPOAE I/O functions at higher L2 levels and hearing thresholds at 2 kHz observed in the present study were highly comparable to those found in the Boege and Janssen (2002) study. As Gorga et al (2003) have argued, it may be the case th at comparisons with behavioral thresholds will profit from DPOAE suprathreshold data, such as the DPOAE compression thresh­ old estimates provided by the model evaluated in the present study. Indices based on higher-level DPOAEs are presumably less susceptible to the influence of noise levels than are low-level DPOAEs, and correlations be­ tween low-level DPOAEs and behavioral hearing thresh­ olds may be compromised by noise levels interfering with the detection of low-level DPOAEs. Correlations betw een H earing Thresholds and DPOAE I/O F unction Com pression Slopes The range of DPOAE I/O function slopes found in this study at 1 kHz (-1.7 to 0.2 dB/dB) and at 2 kHz (-0.1 to 0.4 dB/dB) were in general agreement with the findings of previous reports. Dorn et al (2001) measured DPOAEs in normal-hearing participants and found slopes between 0.1-0.3 dB/dB corresponding to the compressive portion of DPOAE I/O functions. Williams and Bacon (2005) set primary-tone levels using the Kummer et al (1998) for­ mula (Lx = 0.4L2 + 39) to measure DPOAE I/O functions

756

in four participants and found that DPOAE I/O function slopes measured for input levels from approximately 4065 dB SPL varied from 0.18-0.35 dB/dB at 1 kHz and from 0.02—0.18 dB/dB at 2 kHz. There was a tendency for individuals in the present study with higher hearing thresholds to exhibit greater DPOAE slope estimates than individuals with lower hearing thresholds at 2 kHz. A statistically significant correlation between hearing thresholds and DPOAE slope estimates was detected at 2 kHz. This finding con­ trasted with the lack of association between hearing thresholds and compression slopes seen in the GOM studies. However, it was in agreement with previous research investigating DPOAEs with the Kummer et al (1998) formula. Kummer et al (1998) calculated slopes of DPOAE I/O functions for L 2 levels between 40-60 dB SPL in normal-hearing participants and par­ ticipants with sensorineural hearing loss and showed th at the slopes increased with increases in hearing threshold levels. The tendencies observed at 2 kHz in this study were not repeated at 1 kHz, as no significant correlation between hearing thresholds and DPOAE slope estimates was observed at the lower test fre­ quency. Many of the slope estimates at this frequency were less than 0 dB/dB; this may be partly attributed to the presence of plateaus or notches in DPOAE I/O functions. These negative slope estimates may have com­ promised the ability to detect significant associations. Johannesen and Lopez-Poveda (2008) also reported

Modeling DPOAE Compression/Bhagat

Figure 8. DPOAE compression thresholds are plotted against quiet thresholds (top panels), and DPOAE compression slopes are plotted against quiet thresholds (bottom panels) at 1 and 2 kHz. Pearson correlations and lines of best fit are shown in each figure.

negative slope estimates and a large number of notches in DPOAE I/O functions at 1 kHz. They found th at com­ pression estim ates based on DPOAE m easurem ents were lower than compression estimates based on psycho­ acoustic measurements at some frequencies, and they attributed these discrepancies to the negative slopes associated w ith plateaus or notches in DPOAE I/O functions. One explanation for nonmonotonic DPOAE I/O func­ tions seen in this study involves the DPOAE fine struc­ ture. The fine structure, or a rippled pattern of maximal and minimal DPOAE levels, is revealed when DPOAEs are measured with high resolution of the primary tones. According to current theory, the origin of the 2\f\-f2 DPOAE fine structure can be explained by a two-source interference model involving nonlinear generation near the place of maximal overlap of the primary tones (f2 place) on the basilar membrane as well as by linear reflections between the distortion-product place on the basilar membrane and the oval window (Talmadge et al, 1998; Talmadge et al, 1999; Dhar et al, 2002; Shera, 2004). Therefore, the ear-canal recorded DPOAE contains acoustic energy derived from a complex mix­ ture of sources, including the f 2 and 2f \ - f 2 locations on the basilar membrane. The fine structure is believed to arise from constructive and destructive interference between the spatially distributed generator and reflec­ tion sources. He and Schmiedt (1993) attributed the

presence of notches in DPOAE I/O functions to varia­ tions th at occurred in the DPOAE fine structure with increasing primary-tone levels. They found th at owing to transitions in the fine structure, DPOAE levels at a given frequency could be converted from maximal to minimal levels as primary-tone levels increased. In the present study, the potential influence of the fine structure on individual DPOAE I/O functions was not accounted for. Methods for reducing the DPOAE fine structure have been reported elsewhere (Kalluri and Shera, 2001; M auermann and Kollmeier, 2004). How­ ever, some studies have shown th at notches in DPOAE I/O functions at 0.5 and 1 kHz were still prominent even after attempts at reducing the fine structure were applied (Johannesen and Lopez-Poveda, 2008; Johannesen and Lopez-Poveda, 2010). These findings suggest that the techniques used in these studies were either insufficient in reducing the DPOAE fine structure, or that other fac­ tors besides the DPOAE fine structure contribute to the formation of notches in DPOAE I/O functions. Another explanation for the presence of notches in DPOAE I/O functions in the present study involves the DPOAE SNR criteria. In this study, DPOAE I/O functions were required to have DPOAE SNRs of 3 dB or higher at a minimum of 3 consecutive points on the functions. Previous studies th at have compared DPOAE thresholds with behavioral thresholds (Boege and Janssen, 2002; Gorga et al, 2003) have used DPOAE

757

J o u r n a l of th e A m erican A cadem y of Audiology/Volume 25, Number 8, 2014

SNR criteria of 6 dB. It may be possible th a t spikes in the noise floor levels could contribute to the formation of notches in DPOAE I/O functions, particularly those th a t include contributions from DPOAEs closer to the noise floor.

CONCLUSIONS revious research has linked hum an hearing sensi­ tivity for tones to compression threshold estimates derived from a three-segment model applied to GOM functions at selected test frequencies. The results of this investigation, using a different approach to indirectly study BM responses in hum ans with measurements of DPOAE I/O functions, were consistent with the psy­ chophysical data at 2 kHz insofar as DPOAE compres­ sion thresholds from the three-segment model correlated with hearing thresholds. However, unlike the findings from GOM studies, DPOAE compression slopes were also linked to hearing sensitivity at 2 kHz. The slope esti­ mates derived from studies using masking approaches for investigation of BM response growth may be compli­ cated by factors that bias listener performance, whereas these factors are not believed to affect measurement of DPOAEs. Future research should compare compression slope estimates derived from both behavioral and DPOAE methods across a wider frequency range in order to determine which procedure is suited best for compar­ ison with threshold data. From a clinical perspective, suprathreshold data from DPOAE I/O functions within the compressive range of these functions may prove to be profitable measures to compare with hearing thresh­ olds at certain test frequencies.

P

Acknowledgments. The author th an k s Dr. Ju d y Dubno and Dr. Amy Horwitz for thoughtful discussions concerning th is study; Dr. E nrique Lopez-Poveda for his assistance w ith th e three-segm ent linear regression model; and two anony­ mous review ers for th e ir helpful comments. P aul C arter and Janice Tanedo contributed to th is project.

REFERENCES American National Standards Institute (ANSI). (1987) Specifica­ tions for Instruments to Measure Aural Acoustic Impedance and Admittance. ANSI S3.39-1987. New York, NY: ANSI. American National Standards Institute (ANSI). (2004) American National Standard Specification for Audiometers. ANSI S3.62004. New York, NY: ANSI. Boege P, Janssen T. (2002) Pure-tone threshold estimation from extrapolated distortion product otoacoustic emission I/O-functions in normal and cochlear hearing loss ears. J Acoust SocAm 111(4): 1810-1818. Cooper NP, Rhode WS. (1992) Basilar membrane mechanics in the hook region of cat and guinea-pig cochleae: sharp tuning and non­ linearity in the absence of baseline position shifts. Hear Res 63(1-2): 163-190.

758

Cooper NP, Rhode WS. (1997) Mechanical responses to two-tone distortion products in the apical and basal turns of the mammalian cochlea. J Neurophysiol 78(l):261-270. de Boer E, N uttall AL. (2000) The mechanical waveform of the basilar membrane. III. Intensity effects. J Acoust Soc A m 107 (3): 1497-1507. Dhar S, Talmadge CL, Long GR, Tubis A. (2002) Multiple internal reflections in the cochlea and their effect on DPOAE fine structure. J Acoust Soc A m 112(6):2882-2897. Dorn PA, Konrad-Martin D, Neely ST, Keefe DH, Cyr E, Gorga MP. (2001) Distortion product otoacoustic emission input/ output functions in normal-hearing and hearing-impaired human ears. J Acoust Soc Am 110(6):3119-3131. Dubno JR, Horwitz AR, Ahlstrom JB. (2007) Estimates of basilarmembrane nonlinearity effects on masking of tones and speech. Ear Hear 28(1):2-17. Gorga MP, Neely ST, Dorn PA, Hoover BM. (2003) Further efforts to predict pure-tone thresholds from distortion product otoacoustic emission input/output functions. J Acoust SocAm 113(6):3275-3284. Gorga MP, Neely ST, Dierking DM, et al. (2007) Low-frequency and high-frequency cochlear nonlinearity in humans. J Acoust Soc Am 122(3):1671-1680. He NJ, Schmiedt RA. (1993) Fine structure of the 2fl-f2 acoustic distortion product: changes with primary level. J Acoust Soc A m 94(51:2659-2669. Horwitz AR, Ahlstrom JB, Dubno JR. (2007) Speech recognition in noise: estimating effects of compressive nonlinearities in the basilarmembrane response. Ear Hear 28(51:682-693. Johannesen PT, Lopez-Poveda EA (2008) Cochlear nonlinearity in normal-hearing subjects as inferred psychophysically and from distortion-product otoacoustic emissions. J Acoust Soc Am 124(4): 2149-2163. Johannesen PT, Lopez-Poveda EA. (2010) Correspondence between behavioral and individually “optimized” otoacoustic emis­ sion estimates of human cochlear input/output curves. J Acoust Soc Am 127(61:3602-3613. Kafluri R, Shera CA. (2001) Distortion-product source unmixing: a test of the two-mechanism model for DPOAE generation. J Acoust SocA m 109(2):622-637. Karlsson KK, Flock A. (1990) Quinine causes isolated outer hair cells to change length. Neurosci Lett 116(1-2):101-105. Kummer P, Janssen T, Arnold W. (1998) The level and growth behavior of the 2 fl-f2 distortion product otoacoustic emission and its relationship to auditory sensitivity in normal hearing and cochlear hearing loss. J Acoust Soc Am 103(61:3431-3444. Kummer P, Janssen T, Hulin P, Arnold W. (2000) Optimal L(l)-L (2) primary tone level separation remains independent of test fre­ quency in humans. Hear Res 146(l-2):47-56. Lopez-Poveda EA, Johannesen PT. (2012) Behavioral estimates of the contribution of inner and outer hair cell dysfunction to individ­ ualized audiometric loss. J Assoc Res Otolaryngol 13(4):485-504. Lopez-Poveda EA, Plack CJ, Meddis R, Blanco JL. (2005) Cochlear compression in listeners with moderate sensorineural hearing loss. Hear Res 205(1-21:172-183.

M o d e lin g D PO A E C o m p re ssio n /B h a g at

M arshall L, Je ste a d t W. (1986) Comparison of pure-tone audibility thresholds obtained w ith audiological and tw o-interval forcedchoice procedures. J Speech H ear R es 29(1):82-91. M auerm ann M, Kollmeier B. (2004) D istortion product otoacoustic emission (DPOAE) input/ou tp u t functions and th e influence of the second DPOAE source. J Acoust Soc A m 116(4 P t 1):2199-2212. Neely ST, Gorga MP, Dorn PA. (2003) Cochlear compression esti­ m ates from m easurem ents of distortion-product otoacoustic em is­ sions. J Acoust Soc A m 114(3):1499—1507. Neely ST, Johnson TA, Kopun J, D ierking DM, Gorga MP. (2009) D istortion-product otoacoustic emission in put/output characteris­ tics in norm al-hearing and hearing-im paired hum an ears. J Acoust Soc A m 126(2):728-738. Nelson DA, Schroder AC, W ojtczak M. (2001) A new procedure for m easuring peripheral compression in norm al-hearing and hearing-im paired listeners. J Acoust Soc A m 110(4):2045-2064. N u ttall AL, Dolan DF. (1996) S teady-state sinusoidal velocity responses of the basilar m em brane in guinea pig. J Acoust Soc A m 99(3):1556-1565. O xenham AJ, Bacon SP. (2003) Cochlear compression: perceptual m easures and im plications for norm al and im paired hearing. E ar H ear 24(5):352-366. O xenham AJ, Plack CJ. (1997) A behavioral m easure of basilarm em brane nonlinearity in listeners w ith norm al and im paired hearing. J Acoust Soc A m 101(6):3666—3675. Plack CJ, Drga V, Lopez-Poveda E A (2004) Inferred basilar-membrane response functions for listeners with mild to moderate sensorineural hearing loss. J Acoust Soc A m 115(4):1684—1695. Preyer S, Gummer AW. (1996) Nonlinearity of mechanoelectrical trans­ duction of outer hair cells as the source of nonlinear basilar-membrane motion and loudness recruitment. Audiol Neurootol 1(1):3-11. Rhode WS. (2007) D istortion product otoacoustic emissions and basilar m em brane vibration in th e 6-9 kHz region of sensitive chinchilla cochleae. J Acoust Soc A m 122(5):2725-2737. Rhode WS, Recio A. (2000) Study of mechanical motions in the basal region of the chinchilla cochlea. J Acoust Soc A m 107(6):3317-3332. Robles L, Ruggero MA. (2001) Mechanics of th e m am m alian coch­ lea. Physiol Rev 81(3):1305-1352. Rodriguez J, Neely ST, Je ste a d t W, T an H, Gorga MP. (2011) Com­ parison of distortion-product otoacoustic emission grow th rates and slopes of forward-m asked psychom etric functions. J Acoust Soc A m 129(2):864-875. R osengard PS, O xenham AJ, B raid a LD. (2005) C om paring different estim ates of cochlear com pression in listen ers w ith

no rm al an d im p aired h earin g . J A co u st Soc A m 3028-3041.

117(5):

Ruggero MA, Rich NC. (1991) Furosem ide alters organ of Corti mechanics: evidence for feedback of outer h a ir cells upon th e basi­ la r m em brane. J Neurosci 11(4):1057-1067. R uggero MA, R ich N C, R obles L, Recio A. (1996) T he ef­ fe c ts of aco u stic tr a u m a , o th e r c o ch le ar in ju ry , a n d d e a th on b a sila r-m e m b ra n e re sp o n se s to sou n d . In: A xellson A, B o rc h g re v in k H , H a m e rn ik R P, e t al, eds. S cientific Bases o f N o ise-Induced H earing Loss. New York, NY: Thiem e M edical P re ss, 23-35. Ruggero MA, Rich NC, Recio A, Narayan SS, Robles L. (1997) Basilarmembrane responses to tones a t the base of the chinchilla cochlea. J Acoust Soc A m 101(4):2151-2163. S hera CA. (2004) M echanism s of m am m alian otoacoustic emission and th e ir im plications for th e clinical u tility of otoacoustic em is­ sions. E ar Hear 25(2):86-97. S h e ra CA, G u in a n J J , J r . (1999) E voked otoacoustic em issions a rise by tw o fu n d a m e n ta lly d iffe re n t m ech an ism s: a ta x o n ­ omy for m a m m a lia n OAEs. J A co u st Soc A m 105(2 P t 1): 782-798. Siegel JH . (1994) E ar-canal standing waves and high-frequency sound calibration using otoacoustic emission probes. J Acoust Soc A m 95:2589-2597. Stainsby TH, Moore BC. (2006) Tem poral m asking curves for hearing-im paired listeners. H ear Res 2 1 8 (l-2 ):9 8 -lll. S tasiu n as A, V erikas A, Kemesis P, e t al. (2003) A non-linear cir­ cuit for sim ulating OHC of th e cochlea. M ed E ng Phys 25(7): 591-601. Talm adge CL, Long GR, Tubis A, D h ar S. (1999) E xperim ental confirmation of th e two-source interference model for the fine stru ctu re of distortion product otoacoustic emissions. J Acoust Soc A m 105(l):275-292. Talm adge CL, Tubis A, Long GR, Piskorski P. (1998) Modeling otoacoustic emission and hearing threshold fine structures. J Acoust Soc A m 104(3 P t 11:1517-1543. W illiams E J, Bacon SP. (2005) Compression estim ates using behavioral and otoacoustic emission m easures. H ear Res 201 (l-2):44-54. Y asin I, Plack CJ. (2003) The effects of a high-frequency suppres­ sor on tu n in g curves and derived basilar-m em brane response functions. J Acoust Soc A m 114(l):322-332. Y ates GK. (1990) B asilar m em brane nonlinearity and its influence on auditory nerve rate-in ten sity functions. H ear Res 50(1-2): 145-162.

753

Copyright of Journal of the American Academy of Audiology is the property of American Academy of Audiology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.