Distortion-Product Otoacoustic Non-Emissions :

Transcription

1 1 Distortion-Product Otoacoustic Non-Emissions : A Discourse on Distortion Products, Acoustic Distortion Products, Electronic Distortion Products, Electroacoustic Distortion Products, Otoacoustic Distortion Products, and Otoacoustic Distortion-Product Emissions by Andrew J. Hotaling, M.D. 2 Candace R. Blank, M.A. 3 Albert H. Park, M.D. 2 Gregory J. Matz, M.D. 2 William A. Yost, Ph.D. 4 and Michael J.M. Raffin, Ph.D. 2, Paper presented at the 20th. Annual Meeting of the American Auditory Society, Halifax, Nova Scotia, Canada, 3 July Department of Otolaryngology, Loyola University Medical Center, Maywood, IL Department of Audiology, Loyola University Medical Center, Maywood, IL Parmly Hearing Institute, Loyola University, Chicago, IL

2 ABSTRACT Difficulties in some clinical conditions for obtaining recognizable distortion-product otoacoustic emissions (DPAOE) at frequencies less than 1.5 khz due to interference from ambient noise are documented. Difficulties were also encountered for obtaining valid and reliable emissions at frequencies greater than 6 khz. Moreover, in the frequency region between 1.5 and 4.0 khz, the existence of very reliable distortion products recorded from hard-walled cavities, syringes, and post-autopsy cadavers (as high as 11-dB SPL and 24 db above the noise floor) is demonstrated. These distortion products exhibit characteristics that render them indistinguishable from emissions in terms of amplitude, fast-fourrier transforms and input-output functions. The behavior of these distortion products in terms of stimulus parameters commonly used in the clinical applications of DPOAE is explained. Some simple procedures for the evaluation of instrumentation that can help the clinician to identify instrument limitations and criteria to differentiate emissions from artifacts are illustrated. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page ii

3 TABLE OF CONTENTS Page ABSTRACT... ii INTRODUCTION... 1 METHOD... 6 DPOAE... 6 Hard-Walled Cavities... 7 Soft-Walled Cavities... 8 RESULTS AND DISCUSSION Clinical Entities: A Patient Clinical Entities: Fixed Hard-Walled Cavities Clinical Entities: Variable-Volume, Hard-Walled Cavities Clinical Entities: Rubber Ear Clinical Entities: Posthumous Human Ear Input-Output Functions Distortion Products as a Function of L 1 and L DPOAE Details and/or Fast-Fourrier Transformations Electronic Distortion Products: Basic Noise Electromagnetic Distortion Products: Crosstalk Electroacoustic Distortion Products: Dummy Mike Electroacoustic Distortion Products: Disconnected Earphones CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page iii

4 INTRODUCTION The discovery of otoacoustic emissions (OAEs) is credited to Kemp (Glattke and 1 Kujawa, 1991 ; Gorga, Neely, Bergman, Beauchaine, Kaminski, Peters, Shulte and 2 3 Jestead, 1993 ). These may be defined, in Kemp's words : "A new auditory phenomenon has been identified in the acoustic impulse response of the human ear. Using a signal averaging technique, a study has been made of the response of the closed external acoustic meatus to acoustic impulses near to the threshold of audibility. Particular attention has been paid to the waveform of the response at post excitation times in excess of 5 ms. No previous worker appears to have extended observations into this region." (p. 1386) 4 OAEs can be evoked by acoustic stimuli (EOAEs) (Kemp, Ryan and Bray, 1990 ). EOAEs produced by clicks are known as transient-evoked OAEs (TEOAEs). EOAEs evoked by two simultaneous pure tones are known as distortion-product OAEs (DPOAEs). Under certain conditions, the cochlea responds to two simultaneous tones at frequencies f 1 and f 2 (known as primary frequencies and where f 2 > f 1), by generating distortion products. The distortion products present at frequencies equal to 2f 1- f 2and to 2f 2- f 1are known as cubic distortion products. These distortion products appear to originate in the Organ of Corti in the frequency region corresponding to the primary frequencies and they are mechanically propagated through the motion of the cochlear partition (Siegel, Kim and 5 Molnar, 1982 ). EOAEs are thought to be measurable in almost all normal-hearing adults (Martin, 6 Probst and Lonsbury-Martin, 1990 ) as well as in newborns (Stevens, Webb, Hutchinson, 7 Connell, Smith and Buffin, 1990 ). Although the precise incidence is not known, there appears to be general agreement that it is on the order of 95% to 100% in these 8 populations (Bonfils, Dumont, Marie, François, and Narcy, 1990 ; Bonfils, François and D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 1

5 9 10 Trotoux, 1990 ; Robinette, 1992 ), as long as the middle ear is not compromised (Glattke 1 & Kujawa, 1991 ). Moreover, DPOAEs disappear as hearing loss increases (Nielsen, 11 Popelka, Rasmussen and Osterhammel, 1993 ) and are absent in sensorineural hearing 9 loss for which thresholds exceed 15- to 65-dB HL (Bonfils, François and Trotoux, 1990 ; Glattke and Kujawa, 1991 ; Kemp, et al., 1990 ; Lonsbury-Martin and Martin, 1990 ; Probst and Hauser, 1990 ; Robinette, 1992 ; Spektor, Leonard, Kim, Jung and Smurzynski, 1991 ). Some data (Kimberley, Hernadi, Lee and Brown, 1994 ) indicate that DPOAE data, when used with appropriate statistical analyses and other relevant variables, may be used to predict pure-tone thresholds. However, at least two cases of EOAEs have been documented in patients with severe hearing loss (Katona, Büki, Farkas, Pytel, Simon-Nagy and Hirschberg, 1993 ; Prieve, Gorga and Neely, 1991 ). The combination of these observations (that EOAEs are present in virtually all normal hearing individuals and absent in hearing-impaired populations) has led some researchers to suggest the use of EOAE tests for the screening for hearing loss (Bonfils, et al., 1990 ; Bonfils, et al., 1990 ; Glattke and Kujawa, 1991 ; Kemp, et al., 1990 ; 18 Lafrénière, Jung, Smurzynski, Leonard, Kim and Sasek, 1991 ; Martin, Ohlms, Franklin, Harris and Lonsbury-Martin, 1990 ; Norton and Widen, 1990 ; Robinette, 1992 ; Smurzynski, Jung, Lafrénière, Kim, Vasudeva Kamath, Rowe, Homan and Leonard, ; Stevens, et al., 1990 ; White, K.R., Vohr, B.R., Maxon, A.B., Behrens, T.R., 22 McPherson, M.G. and Mauk, G.W., 1994 ). The enthusiastic endorsement of this measurement as a screening tool by some professionals has led the National Institutes of Health to adopt a consensus statement in which the use of OAEs appears to be the 23 preferred screening measurement (National Institutes of Health, 1993 ). While EOAEs may be easily recorded from normal-hearing persons, the emissions appear to exhibit a high degree of variance between subjects (Harris, Probst and Wenger, ; Kemp, et al., 1990 ; Smurzynski, et al., 1993 ). Within subjects, however, the 1 24 emissions are quite stable over time (Glattke and Kujawa, 1991 ; Harris, et al., 1991 ; D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 2

6 19 11 Martin, et al., 1990 ; Nielsen, et al., 1993 ). This stability of the emissions over time within a given subject indicates that EOAEs may be useful for monitoring changes in cochlear function over time for such applications as effects of noise exposure on auditory 25 function, therapy effects, and surgical procedures (Brown, McDonell and Forge, 1989 ; Glattke and Kujawa, 1991 ; Harris, et al., 1991 ; Kemp, et al., 1990 ; Lonsbury-Martin, Harris, Stagner, Hawkins and Martin, 1990 ; Lonsbury-Martin and Martin, 1990 ; Martin, et al., 1990 ; Martin, et al., 1990 ; Norton and Widen, 1990 ). The presence of inner and outer hair cells both are thought to be necessary for the generation of DPOAEs (Canlon, Marklund and Borg, 1993 ; Schrott, Puel and Rebillard, 1991 ) While DPOAEs are not inherently more frequency-specific than TEOAEs, they allow for the selection of specific frequencies at which measurements may be made (Gorga, ; Gorga, et al., 1993 ; Lafrénière, et al., 1991 ; Lonsbury-Martin and Martin, 1990 ; Lonsbury-Martin, et al., 1990 ; Martin, et al., 1990 ; Martin, et al., 1990 ; Nielsen, et al., ; Probst and Hauser, 1990 ; Spektor, et al., 1990 ). In addition, a mathematical 30 model proposed by Zurek (1992 ) provides the mathematical bases that suggest that DPOAEs may be advantageous over TEOAEs in terms of test administration time for a given error rate, or in terms of signal-to-noise ratio for a given test administration time. Thus, the preponderance of the published research seems to indicate some clinical applications for EOAEs both TEOAEs and DPOAEs. However, the application of these technologies to patient populations should not be accomplished without considerable caution. For example, while DPOAEs may be found at frequencies greater than 6 khz, they do not appear to be well correlated with auditory thresholds at those high frequencies 13 (Probst and Hauser, 1990 ). While some investigators believe that the DPOAE amplitude 26 actually increases at these high frequencies (Lonsbury-Martin, et al., 1990 ), others have 31 indicated that DPOAEs at 6 khz or higher frequencies may be artifactual (Arjmand, 1994 ; 29 Gorga, 1994 ). Moreover, ambient noise in combination with physiologic noise may make 31 the identification of DPOAEs at frequencies less than 1.5 khz difficult (Arjmand, 1994 ; D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 3

7 Bonfils, Avan, François, Trotoux and Narcy, 1992 ; Gorga, 1994 ; Gorga, et al., 1993 ; Probst and Hauser, 1990 ; Smurzynski, et al., 1993 ). Finally, a seal must be carefully 33 established in the ear to avoid additional artifacts (Baer and Hall, 1992 ; Martin, et al., ). While Bonfils, et al. (1992 ) did not find instrument artifacts in a 2-cm cavity with 13 stimulus intensities of 84-dB SPL, Probst and Hauser (1990 ) have cautioned that instrument-related distortion products (DPs) must be differentiated from biological emissions. Certain characteristics of DPOAEs are noted. DPOAE amplitude in neonates is 29 larger than that found in adults (Gorga, 1994 ) and both groups typically saturate at near dB SPL (Spektor, et al., 1991 ). Because of noise problems, some authors have suggested that a DP not be identified as an emission unless its amplitude exceeded the noise floor. The amount by which this response amplitude exceeds the noise floor is not standardized and in some cases is established through visual inspection of the hard copy or screen display. Amplitudes as small as 3 db greater than the noise floor have been 26 suggested as criteria for the identification of DPOAEs (Lonsbury-Martin, et al., 1991 ; Smurzynski, et al., 1993 ), as well as for TEOAEs (White, et al., 1994 ). 34 Recently, Siegel (1994 ) published an elegant discourse on standing-wave phenomena in the ear canal that could seriously affect the effective calibration of and, by extension, the reliability of emissions measured at, acoustic signals beyond 2 khz. His mathematical model accurately predicts an underestimation of eardrum sound-pressure levels at high frequencies by as much as 20 db. In that case, a nominal setting of 75-dB SPL may well generate 95-dB SPL at the eardrum. In addition, individual differences in ear-canal dimensions will produce different standing-wave patterns that will, in turn, introduce substantial measurement error. Under these conditions, the emissions seen at the plane of measurement (the probe tip) may appear different from those measured in proximity to the eardrum. Moreover, one might expect that the depth of insertion of the D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 4

8 probe tip also would affect the appearance of emissions and/or the magnitude of the measurement-plane sound-pressure levels. If substantial measurement errors are likely to be routine, then these considerations should produce a healthy skepticism among clinicians regarding the immediate clinical advisability of the use of otoacoustic emissions as a clinical diagnostic tool. It seems warranted that clinicians should critically evaluate the need to apply EOAE measurements to patient populations. The purpose of this presentation is to present some data that may be used as a basis for the formulation of criteria that may define a DPOAE. These data also may provide some "clinical" evidence that substantiates some of the implications of 34 Siegel's (1994 ) mathematical model. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 5

9 METHOD All measurements were obtained with a clinical instrument (Virtual Corporation, Model 330) and associated software (Virtual, Version 1.00) approved by the United States Food and Drug Administration (FDA) for clinical use. Prior to any measurement, the instrument was calibrated in accordance with the manufacturer's specifications. Tubing was cut at 0.2-inch beyond the rubber cuff tip immediately prior to the SPL-measurement procedure. This procedure incorporated the use of a Type 0 (American National Standard 35 Institute [ANSI] Standard S ) sound-level meter (Brüel & Kjaer, Type 2209) with an octave-band analyzer (Brüel & Kjaer, Type 1613) conforming to ANSI Standards (ANSI 36 Standard S ) due, in part, to the need for accurate measurements in the frequency range exceeding 3 khz. A half-inch pressure microphone (Brüel & Kjaer, Type 4133) was connected to the sound-level meter with an extension cable (Brüel & Kjaer, Type UA0196). The test cavity, provided by the manufacturer, also serves as an adaptor to join the sound-pressure measurement system to the instrument's probe assembly. These measurements were undertaken in a double-walled, sound-insulated test room (Industrial Acoustics Corporation, Order # ). All measurements obtained were corrected for that microphone's (SN: ) open-circuit pressure response, as determined from its frequency-response curve for each measurement frequency of the instrument (200 through Hz). Corrections for the effects of the microphone grid also were incorporated. DPOAE: Unless otherwise indicated, for the generation of DPOAE-grams (DPOAEs as a function of specified stimulus frequency), the two stimulus frequencies (f 1 and f 2) were presented at equal sound-pressure levels (L 1 = L 2 = 75-dB SPL). Two spectral averages were obtained at each test frequency, and test frequencies were defined as 1/6th.-octave intervals from 500 Hz through 8 khz on the basis of the geometric mean frequency (f e) of ½ f 1 and f 2 ([{f 1}{f 2}] ). The f 2/f 1 frequency ratio was 1.21 and the results were plotted as a D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 6

10 function of f e. Prior to the insertion of the probe into any cavity, the earphone tubing was temporarily disconnected in order to avoid a pressure buildup in the cavity related to probe insertion and to the achievement of a seal. The instrument allows for the measurement of input-output (I/O) functions. Various investigators have shown patterns of I/O functions in clinical populations (see for example Arjmand, 1994 ; Gorga, 1994 ; Lasky, Snodgrass and Hecox, 1994 ). This measurement technique has not been approved for clinical use by the FDA and is therefore considered experimental in nature. For the present application, no official research protocol was filed with the FDA (via the manufacturer) since no live subjects were involved. I/O functions were obtained with the input sound pressure being increased systematically in 2-dB increments from 35- to 75-dB SPL. Hard-Walled Cavities: Hard-walled cavities originally designed to check the calibration of acousticimmittance measurement instruments (Amplaid, Model 720; GSI, ) were used for practice purposes because these are readily available in the clinic. For that part of the study involving syringes, some explanations are in order. Syringes with different total volume capacity were obtained. The syringes are disposable, made of plastic, and are identified by nominal total volume capacity. The tip of each syringe was cut off using a hobby-knife/razor-blade tool and then the edges were buffed smooth, so as to provide a straight cylinder. Three syringes were used and adjusted to a volume of 0.9 cc as follows: 1. a 1-cc syringe (Beckton-Dickinson, Stock W-12430) (length 5.83 cm; radius 0.23 cm); 2. a 3-cc syringe (Beckton-Dickinson, Stock W-12454) (length 1.75 cm; radius cm); 3. a 5-cc syringe (Beckton-Dickinson, Stock W ) (length 0.88 cm; radius 0.60 cm). All dimensions were measured visually with a ruler placed adjacent to the syringe. Length was measured from the tip of the rubber cuff placed on the instrument probe tip to the rubber ring of the syringe plunger, when the probe D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 7

11 tip was secured in the syringe. The probe-tip rubber cuffs of nominal diameters of 4 mm, 9 mm, and 12 mm were used with each of the three syringes (1-cc, 3-cc, and 5-cc syringe respectively). The seal provided by this arrangement was estimated by first connecting said rubber cuffs to an aural acoustic-immittance measurement system (Virtual, Model 310). A positive pressure of 500 dapa was introduced into the syringe, and the pressure was measured at 25 minutes after the onset of the positive pressure. The worst leakage resulted from the use of the 12-mm rubber cuff installed with the 5-cc syringe. This condition produced a pressure of 380 dapa 25 minutes after the introduction of the 500- dapa into the 0.9-cc volume. Thus, a hermetic seal was not obtained, although the seal was watertight (as demonstrated by no noticeable loss of colored water after 25 minutes). When a larger rubber cuff (13 mm) was used, a seal became impossible to maintain because the sides of the cuff became crimped as it was inserted into the syringe. Syringes were deemed advantageous over hard-walled cavities because the volume of air trapped at the tip of the probe assembly could be varied systematically, thus providing for a systematic change in the acoustic impedance load applied to the instrument's transducers. Syringe volume adjustments were implemented with the probe-assembly's tubing disconnected from the earphones in order to avoid pressure buildup in the cavity. Soft-Walled Cavities: Data also were obtained by securing the instrument probe into a rubber ear (Hal Hen, Catalog #2317), such as that frequently used to show patients how a hearing-aid might look or be made to fit on an ear, or used to train individuals in earmold-impression making techniques. The rubber was cut off at the eardrum location, and the hole covered by thin adhesive tape (3-M, Scotch Cat. 122). The tape covered the hole with a slight bulge induced by a pendil eraser protruding slightly beyond the edges of the hole (rather than being stretched tightly across the hole). The adhesive tape and its edges were secured firmly against the artificial ear with glue to guarantee a hermetic seal (no loss of pressure from 500 dapa after 25 minutes) while forming a non-rigid membrane. The purpose of this procedure was to determine whether providing the probe with a very small D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 8

12 acoustic volume, made of quasi sound-absorbent material and terminated with a non-rigid membrane, would reduce the magnitude of the distortion products observed in hard-walled cavities such as acoustic-immittance calibration cavities or syringes. Cadavers were obtained following autopsy at the morgue of a major medical center. Cadaver ears were used to provide the best estimate of the absorption, vibration and impedance characteristics of the human ear. The cadaver ear canal was cleaned of all debris, and otoscopically examined to document the absence of fluid prior to any measurement. Measurements were begun more than twelve hours following death and several hours after the completion of a formal autopsy without brain removal. Under these conditions, in which blood is replaced by toxic preservative chemicals and viscera are removed for comprehensive dissection, it is unlikely that any hair cells could be functional. Experience with intraoperative monitoring of electrical activity originating in the cochlea indicates that electrical potentials disappear within seconds of interruption of cochlear blood flow and that this loss becomes irreversible and is accompanied by profound loss 38 of hearing when the interruption is maintained more than 15 minutes (Møller, 1994 ; 39 Nedzelski, 1994 ). Measurements were obtained on these patients in the same manner as they are routinely obtained in live patients. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 9

13 Clinical Entities: A Patient RESULTS AND DISCUSSION Figure 1 Audiogram obtained on an adult, male, healthy volunteer. Figure 1 shows the audiogram obtained on a volunteer subject. Thresholds for frequencies less than 3 khz are borderline normal, and beyond 3 khz, they indicate a moderate to severe sensorineural hearing loss. Figure 2 shows a DPOAE-gram obtained from the subject whose audiogram is shown in Figure 2 DPOAE-gram obtained on an adult male (f 2/f 1ratio = 1.21; L 1 = 75-dB SPL; L 2 = 69-dB SPL; 2 Spectral averages, 32 time averages). Trial 1 (with noise floor labelled Noise 1) was obtained in an office, Trial 2 (with noise floor labelled Noise 2) was obtained 1 month later in a double-walled sound-treated test room. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 10

14 Figure 1. Two trials were obtained. Trial 1 was administered in an office while Trial 2, obtained one month later, was administered in a double-walled sound-treated room. The differences in the noise floors (Noise 1 and Noise 2) are probably related to the acoustic environment of the room in which the test is administered. Thus, the administration of the test in a quiet environment resulted in a 10-dB reduction in the level of the noise floor for the low frequencies. Even under controlled environmental conditions, the noise floor (Noise 2 in Figure 2) is relatively elevated (approximates 0-dB SPL) for frequencies less than 1.6 khz. The standard deviations obtained for the DPOAEs (not plotted here) were about 1.5 db at all frequencies for which the DPOAE amplitude nominally exceeded the magnitude of the noise for both trials. Moreover, a repeat of this test without removing the probe, resulted in superimposed tracings for both the DPOAE and the noise curves for each trial. The disappearance of apparent emissions beyond 3 khz (for which frequency this patient's threshold is 35-dB HL) between 40- and 60-dB HL is not inconsistent with 1 29 previously published information (Glattke and Kujawa, 1991 ; Martin, et al., 1990 ). Finally, the absolute amplitudes of these apparent DPOAEs (in the range of 1.8 to 3.0 khz) were on the order of 10-dB SPL, with a minor peak of 0-dB SPL at 5 khz. This peak, however, was less than 10 db greater than the noise floor and was therefore not considered an emission a priori. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 11

15 The fast-fourrier transformations of the realtime data illustrated in Figure 2 show the amplitude characteristics of the stimuli and the response as a function of frequency in greater detail than the DPOAE-gram (see Figure 3 and Figure 4). In these figures, the cubic distortion products (2f 1 - f 2 labelled f dp) are clearly identifiable at 14.6 db and 21.4 db greater than the noise floor (f n) respectively. Figure 3 Detail of DPOAE obtained on an adult male (f 2/f1 ratio = 1.21; L 1 = 75-dB SPL; L 2 = 69-dB SPL; 32 time averages; 2 spectral averages). Filled circle ( ) indicates cubic distortion product. Figure 4 Detail of DPOAE obtained on an adult male (f 2/f1 ratio = 1.21; L 1 = 75-dB SPL; L 2 = 69-dB SPL; 32 time averages; 2 spectral averages). Filled circle ( ) indicates cubic distortion product. Clinical Entities: Fixed Hard-Walled Cavities Upon the acquisition of the instrument in question, staff members of the clinic were encouraged to practice using the instrument with the probe assembly connected to various calibration cavities available in the clinic, in order to become proficient in the use of the instrument. During this phase of training, interesting results were obtained. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 12

16 Some members of the staff practiced on an acousticimmittance system calibration cavity (Grason-Stadler, Inc, Model ). The results of this measurement are shown in Figure 5. Results of two independent measurements are shown with one noise floor since the noise floors for both trials were about the same. These may be called acoustic DPs since they are measured Figure 5 Distortion products measured in an acoustic-immittance calibration cavity as a function of the geometric mean frequency of f 1 and f 2 (f 2/f 1 ratio = 1.21; L 1 = L 2 = 75-dB SPL; 32 time averages; 4 spectral averages). acoustically. These measurements show a fair degree of replicability. Sharp response peaks are seen near 1.5 khz with amplitudes that can exceed 10-dB SPL, while lesser peaks with amplitudes less than 0-dB SPL also are noted. The results aroused some concern because at some frequencies, the distortion products measured cannot be otoacoustic, nor can they be emissions, but they exceeded the noise floor by 23.7 db (at 1.6 khz in the curve labelled Trial 1 in Figure 5). The second measurement showed a peak of 16.7 db (at 1.4 khz in the curve labelled Trial 2 in Figure 5) greater than the noise floor, with a slight shift in the frequency at which the peak is noted. In neither trial is this peak response sustained for more than a-rd. octave. A second highly reliable peak response is noted at 3.2 khz with an amplitude exceeding the noise floor by 9.9 db for both trials. Finally, a large peak response (18.4 db and 19.2 db greater than the noise floor, for Trial 1 and Trial 2 respectively) is noted at the geometric mean frequency of 8 khz. These responses are internally consistent as shown by a standard deviation of the measurement of about 2 db in all instances, except for the response observed in Trial 1 at 8 khz (which was on the order of 7 db). The differences between Trial 1 and Trial 2 may be the result of different insertion depths of the probe into the cavity. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 13

17 A detail of the DPOAE-gram shown in Figure 5 is illustrated in Figure 6. These data indicate that artifacts observed for very low frequencies (in this case for f e = 707 Hz) may be related to the proximity of the cubic distortion- product frequency (f dp = 507 Hz) to the slope of the f1 stimulus. It is possible that this artifact may be related to the rise/fall times of the stimuli as well as to instrument circuitry and transducer characteristics. Figure 6 Detail of DPOAE-gram obtained in an acousticimmittance calibration cavity (GSI ). The filled circle ( ) indicates the cubic distortion product. To explore the possibility that these findings might be related to a unique construction geometry of the cavity, measurements also were obtained in another cavity used for the calibration of an acoustic-immittance system (Amplaid, Model 720). These results are shown in Figure 7. Except at frequencies less than 600 Hz, no peak exceeds 0-dB Figure 7 Distortion products measured in an acousticimmittance calibration cavity as a function of the geometric mean frequency of f 1 and f 2 (f 2/f 1 ratio = 1.21; L 1 = L 2 = 75-dB SPL; 32 time averages; 4 spectral averages). D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 14

18 SPL. On the first measurement, in the frequency range extending from 1.8 khz through 2.5 khz, the response amplitude ranges from 10.1 db to 16.2 db greater than the noise floor, and the maximum peak is seen at 2 khz. On the second trial, the response does not affect as many frequencies (extending from 2.0 khz to 2.5 khz) and is of a lesser amplitude (ranging from 10.9 db to 13 db greater than the noise floor). However, in the low-frequency region (600 through 800 Hz), reliable response peaks of 6 db to 8 db greater than the noise floor are seen. Moreover, at 8 khz, both trials indicate a huge response amplitude of 17.2 db to 19.4 db greater than the noise floor. It is not inconceivable that similar distortion products exist in the range of 2.6 to 6 khz, however the relatively high noise levels inherent in this specific instrument may preclude an accurate assessment of that possibility. While the volumes of the two cavities (Amplaid and GSI) are nominally the same, the DPs measured show peaks at different frequencies. These divergences may be the result of different dimensions that characterize these cavities (in terms of length and diameter). Moreover, the measurement of DPs in one cavity (Amplaid) required the use of a rubber cuff on the probe tip, whereas the probe tip fitted directly into the opening of the other cavity (GSI) without the need for any rubber cuff. Cavities that have been designed for the calibration of acoustic-immittance measurement devices may have some unusual geometric patterns which could produce unusual amounts of acoustic distortion. Specifically, they may not have a constant inside diameter, but rather may be characterized by a narrow neck opening out into a wider cavity to provide the acoustic load appropriate 40 for those instruments (Birck, 1994 ). Thus, hard-walled cavities such as these may well exhibit strange standing-wave patterns when point sound sources are introduced in them. These may contribute to enhancement and cancellation effects that result in the observed acoustic DPs throughout these measurements. Such constrictions in the diameter of a cylinder followed by a widening chamber may not be unlike that produced by an ear canal endowed with either debris or partial ceruminous obstruction. The use of a straight-walled cylinder would avoid any effects of constrictions in the cavity. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 15

19 Clinical Entities: Variable-Volume, Hard-Walled Cavities Syringes may be modeled as variable-volume hard-walled cavities with fixed 34 diameter. One of the implications of Siegel's (1994 ) model would indicate that perhaps different impedances and standing-wave patterns might be produced in cavities having different dimensions, even when the net volume is held constant. Syringes were used, in part, to determine whether such changes would result in frequency shifts in the location of apparent DPs. Thus, the measurement of DPs in fixed volumes with differing length to diameter aspect ratios was undertaken. absorption due to the presence of a rubber tip on the plunger. Syringes also may provide partial sound Figure 8 illustrates the DPOAE-gram obtained in the 1- cc capacity syringe. This syringe was adjusted to provide an acoustic load to the transducers equivalent to a 0.9- cc volume of air. In this arrangement, four regions of peak responses are observed: at geometric-mean frequencies less than 800 Hz, between 1.6 and 2.5 khz, between 3.7 and 5 khz and finally at 8 khz. Figure 8 Distortion products measured in a 1-cc capacity syringe (adjusted for a volume of 0.9 cc) as a function of the geometricmean frequency of f 1 and f 2 (f 2/f 1 ratio = 1.21; L 1 = L 2 = 75-dB SPL; 128 time averages; 2 spectral averages). Except at very low frequencies, and also in the region of 4 khz, the peaks do not appear to reliably exceed 0-dB SPL in amplitude. While these peak responses tend to be less than 10 db greater than the noise floor for the very low frequencies, they are between 18 to 21 db greater than the noise floor in the 2-kHz frequency region, 10 to 12 db greater than the noise floor in the 4-kHz region, and finally they are conspicuous by being 17 to 38 db greater than the noise floor at 8 khz. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 16

20 Figure 9 shows the results obtained in a 3-cc modified syringe, adjusted to provide a 0.9-cc acoustic load to the transducers. These results are similar to those observed in Figure 8. Two major peaks are noted and they are both less than 0-dB SPL in amplitude. The first peak is at 2 khz and is 23.2 db greater than the noise floor. The second peak is at 8 khz and is 20.1 db greater than the noise floor. Figure 9 Distortion products measured in a 3-cc capacity syringe (adjusted for a volume of 0.9 cc) as a function of the geometricmean frequency of f 1 and f 2 (f 2/f 1 ratio = 1.21; L 1 = L 2 = 75-dB SPL; 128 time averages; 2 spectral averages). Figure 10 illustrates the results of measurements obtained in the 5-cc capacity syringe adjusted to a volume of 0.9 cc. The pattern demonstrated by these data differs visually from that demonstrated by the data illustrated in Figure 8 and in Figure 9. The entire frequency range from 2 khz through 4 khz is nominally greater than the noise floor, though not by more Figure 10 Distortion products measured in a 5-cc capacity syringe (adjusted for a volume of 0.9 cc) as a function of the geometricmean frequency of f 1 and f 2 (f 2/f 1 ratio = 1.21; L 1 = L 2 = 75-dB SPL; 128 time averages; 2 spectral averages). D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 17

21 than 10 db. At 8 khz, the DP amplitude is 17.1 db greater than the noise floor. Not one of these peaks is greater than 0-dB SPL. Figure 11 compiles the data presented in Figure 8, Figure 9 and Figure 10. Generally, it would appear that the response peak measured near 2 khz can be shifted to a slightly greater frequency through a reduction in the diameter of the enclosed cavity, by the concomitant increase in the length of the cavity or by the interactive effects of these two Figure 11 Distortion products as a function of the geometric-mean frequency of f 1 and f 2 for a constant closed volume of 0.9 cc for each of three different cavity dimensions. dimensions on the acoustic resonances produced by changing acoustic load. Figure 12 converts the data illustrated in Figure 11 to DPs in db relative to the level of the noise. For a constant volume of 0.9 cc, there appear to exist two major (more than 10 db greater than the noise amplitude) response peaks: In the region of 2 khz and at 8 khz. Figure 12 Distortion products as a function of the geometric-mean frequency of f 1 and f 2 for a constant closed volume of 0.9 cc for each of three different cavity dimensions. Response amplitude is expressed relative to the noise amplitude. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 18

22 Thus, it would appear that the frequency at which given DPs are observed may be altered by changes in the internal diameter as well as in the length of the cavity into which the probe assembly is sealed. The responses observed under these conditions may be acoustic DPs. The worst-case implication of the above findings is that the peaks observed in these illustrative examples may be assumed as artifacts, and that therefore no response of this magnitude observed in patients can reasonably be differentiated from artifact. Responses of this magnitude or less, therefore, must be identified as artifacts until proven otherwise. Clinical Entities: Rubber Ear The length of the ear-canal portion of the rubber ear is very short. Therefore, in that sense, it is not representative of the average adult ear canal. With the probe in place, the tips of the tubes (extending 0.2-inch beyond the tip of the rubber cuff) were within 2 mm of the tape terminating the canal portion of the rubber ear. Moreover, the diameter of the canal opening is much larger than that typically encountered in pediatric populations. The contribution of this entity is to provide a semi absorbent surface with partial vibration characteristics as it is terminated in a thin adhesive tape with an admittance of about 0.15 mmho. Figure 13 shows the DPs (perhaps these could be called artificial otoacoustic distortion products since they are measured in an artificial ear) obtained in a rubber ear. Of interest, some large peaks are noted at frequencies less than 800 Hz, with some minor peak responses in the 2.2 khz Figure 13 Distortion products measured in a rubber ear as a function of the geometric-mean frequency of f 1 and f 2 (f 2/f 1 ratio = 1.21; L = L = 75-dB SPL; 128 time averages; 2 spectral averages). 1 2 D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 19

23 region and again in the 3.6 khz region and also for the frequencies greater than 5 khz. Only the very low frequencies demonstrate response amplitudes greater than 0-dB SPL and these are from 9 db to 10 db greater than the noise floor. The amplitude of these response at 8 khz are 13 db to 17 db greater than the noise floor. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 20

24 Figure 14 illustrates a comparison of the data shown in Figure 13 with that seen in Figure 2. In Figure 14, the data have been normalized relative to the noise level. The noise level, being less in the rubber-ear recording, may be able to show some lowfrequency DPs that were not observable in the human-ear recording due to its relatively high noise level in that Figure 14 Comparison of distortion product observed in a rubber ear with those obtained in an adult afflicted with high-frequency sensorineural hearing loss. Response amplitude is expressed relative to the noise amplitude. frequency region. Thus, the relative amplitude of the DPs in the rubber ear may only appear to be greater than that of the human ear for the very low frequencies (less than 1.6 khz). The human ear shows a broad range of response in the 2-kHz region that is sustained from approximately 1.6 to 3.2 khz. In comparison, the responses measured in the rubber ear show only sharp peaks whose amplitudes are about 5 to 10 db nearer the noise levels than those of the human ear. Both clinical entities demonstrate a highamplitude response at 8 khz. While there are differences between the presumed emissions observed in the human patient and the DPs observed in the rubber ear, one could anticipate that some clinicians would be tempted to call responses similar to those exhibited by the rubber ear as islands of normal hair-cell functions when the responses are observed in patients afflicted with hearing loss. Clinical Entities: Posthumous Human Ear D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 21

25 Figure 15 illustrates the results from measurements obtained on a cadaver (these may be called otoacoustic distortion products, but are probably not emissions). No response peak exceeds 0-dB SPL, and only in the region of 2 khz do the response peaks exhibit amplitudes more than 10 db greater than the noise amplitude (by subtracting the noise amplitude from the DP Figure 15 Distortion products measured in a post-autopsy cadaver as a function of the geometric-mean frequency of f 1and f 2(f 2/f 1ratio = 1.21; L 1 = L 2 = 75-dB SPL; 64 time averages; 2 spectral averages). amplitude). A single exception is noted to this observation. At 8 khz, the response peak is 18.8 db greater than the noise floor and ranges from 7.5- to 9-dB SPL. Figure 16 illustrates a comparison of the data shown in Figure 2 with that seen in Figure 15. In Figure 16, the data have been normalized relative to the noise level. The relative amplitude of the DPs in the live patient appears only marginally greater than that measured in the cadaver. The differences between these two subjects are differences which clinically might be considered Figure 16 Comparison of distortion products observed in a cadaver with those obtained in an adult afflicted with high-frequency sensorineural hearing loss. Response amplitude is expressed relative to the noise amplitude. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 22

26 within the test-retest measurement error. The largest difference (at 1.8 khz) is about 13 db, and that appears to be due in large part to the differences in the width of the response area, with the live patient exhibiting a larger width. An examination of published research indicates that responses similar to those obtained in the present investigation in this, and several other, post-autopsy cadaver ears tend to be identified as responses. Moreover, when such responses are observed in individuals with significant auditory dysfunction (shown by other tests), then the validity of the DPs may not be questioned. Rather, a theory may be postulated whereby a given patient probably had normal peripheral (inner ear) function, accompanied by significant (previously unsuspected) retrocochlear or nervous-system dysfunction (see for example Katona, et al., 1993 ; Prieve, et al., 1991 ). The findings of the present study cannot refute those theories. Rather, the results of the present investigation indicate that such rare events should not be postulated until the effects of artifacts are ruled out. Input-Output Functions Some investigators (for example: Arjmand, 1994 ; Gorga, 1994 ) have suggested that the characteristics of the I/O functions are clinically indispensable, and may be sometimes useful in detecting artifacts thus allowing for the differentiation of genuine DPOAEs from artifacts. Figure 17 exhibits an I/O function in a small (0.2 cc) hardwalled cavity characterized by a sudden increase in amplitude as the input sound pressure attains 67-dB SPL, followed by a sharp rollover or perhaps Figure 17 Input-output function at 2 khz for a 0.2-cc, hard-walled cavity (Amplaid calibration reference for Model 702 aural acoustic immittance measurement system) (f 2/f 1ratio = 1.21; L 1= L 2; 32 time averages; 4 spectral averages). D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 23

27 saturation. The level of the noise floor also increases sharply at greater sound-pressure levels. Figure 18 illustrates results obtained on a normalhearing subject (Lasky, et al., ). The comparison of the data illustrated in Figure 17 with that shown in Figure 18 is interesting because it demonstrates a resemblance of an I/O pattern obtained in a cavity with that obtained in a human. The pattern is an unusual one in that there is a build up to a peak, then a Figure 18 Detail of an input-output function obtained on a patient (after Lasky, R.E., Snodgrass, E. and Hecox, K. [1994] Distortionproduct otoacoustic emission input/output functions as a function of frequency in human adults. J. Am. Acad. Audiol., Vol. 5, Page 192. reduction in amplitude of the DPs to a sharply defined notch that is within about 5 db of the DP amplitude observed at low-levels of stimulation, and this is followed by a marked increase in DP amplitude to a level approximating the peak amplitude observed just prior to the notch. In the human subject the dynamic range of the high-intensity, non-monotonic notch is about 15 to 20 db, with some slight variability between trials. In the cavity the dynamic range of the high-intensity, non-monotonic notch pattern observed in Figure 17 is only 10 to 15 db. There are other, perhaps more significant differences (since similar 37 findings were obtained from two different instruments). Lasky, et al. (1994 ) used totally different hardware (Ariel, DSP-16+ digital signal processing board with Etymotic, ER2 earphones and Etymotic, ER-10B low-noise ear-canal microphone) and different software (CUBDISP developed by Allen and colleagues). In addition, their data (displayed in Figure 18) were obtained with an f 2 of 3760 Hz (f Hz, f e 3432 Hz). Moreover they have identified a similar pattern for yet higher frequencies. D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 24

28 Figure 19 shows an I/O function obtained in a 2-cc hard-walled cavity characterized by a gain function greater than unity (unity gain is defined as an I/O function for which every increase in input level is accompanied by an equal increase in output level). That is, as the input changed by 12 db (from 63- to 75-dB SPL), the output increased by 38 db (from -29-dB SPL to 8.7-dB SPL). Figure 19 Input-output function at 2 khz for a 2-cc, hard-walled cavity (Amplaid calibration reference for Model 702 aural acoustic immittance measurement system) (f 2/f 1ratio = 1.21; L 1= L 2; 32 time averages; 4 spectral averages). Such a pattern might be called a recruiting pattern which, when observed in live patients, might prompt one to speculate about a hair-cell level mechanism for recruitment. A more 41 simple explanation, perhaps, was predictable from von Helmholtz's (1856 ) calculations of nonlinearities of combination tones barely more than a quarter-century after the premiere in Paris of Hector Berlioz's Symphonie Fantastique and five years before Prosper Ménière published his memorable series of articles. These would predict that the magnitude of the DPs should increase with the level of the primaries (f 1and f 2) cubed. The slope of 3.16 (db-output/db-input) shown in Figure 19 is not inconsistent with the slope 42 of 3.4 reported for one normal-hearing subject by McFadden and Pasanen (1994 ). D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 25

29 Figure 20 Illustrates yet another pattern of I/O functions which has been observed in systems which do not have hair cells. These data were obtained at a geometric-mean frequency of 4 khz in a 2-cc hard-walled cavity. This pattern shows a less-than-unity gain (characterized by an output growth which is less than the input change), as the output changes by about 12 db (from - Figure 20 Input-output function at 4 khz for a 2-cc, hard-walled cavity (Amplaid calibration reference for Model 720 aural acoustic immittance measurement system) (f 2/f 1ratio = 1.21; L 1= L 2; 32 time averages; 4 spectral averages). 18-dB SPL to -6-dB SPL) with an input change of 24 db (from 51-dB SPL to 75-dB SPL). Less-than-unity gain functions have not been reported in live humans. Thus, this kind of I/O function may be unique to distortion product non-emissions. Figure 21 also shows an unusual I/O pattern in that the output is irregular until the input level attains 67-dB SPL. There is also a slight increase in the noise level at the high output levels. For the input range of 65- to 75-dB SPL, the slope (db output per db input) of the I/O function is For the input range of 67- to 75-dB SPL, the slope of the I/O function is Figure 21 Input-output function in a 1-cc capacity syringe adjusted to a 0.9-cc volume for a 2-kHz stimulus (f 2/f 1 ratio = 1.21; L 1= L 2; 32 time averages; 2 spectral averages). D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 26

30 Finally, for the input range of 53- to 75-dB SPL, the slope of the I/O function is 1.3. All of these slopes are within the range of 1.0 to 2.1 reported in adult human subjects by Wier, 43 Pasanen and McFadden (1988 ). The I/O pattern exhibited in Figure 22 shows a highly variable output which seems to demonstrate almost a step function characterized by a sharp gain between input levels of 63- and 65-dB SPL after which the output stabilizes near -15-dB SPL. Figure 22 Input-output function in a 3-cc capacity syringe adjusted to a 0.9-cc volume for a 2-kHz stimulus (f 2/f 1 ratio = 1.21; L 1= L 2; 32 time averages; 2 spectral averages). A stable saturation pattern is exemplified in Figure 23. While the input increases from 65-dB SPL to 75-dB SPL, the output remains stable at about -13-dB SPL. I/O functions may exhibit different patterns under different measurement conditions. It is quite possible that some of the patterns Figure 23 Input-output function in a 5-cc capacity syringe adjusted to a 0.9-cc volume for an 8-kHz stimulus (f 2/f 1 ratio = 1.21; L 1 = L 2; 32 time averages; 2 spectral averages). D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 27

31 documented as distortion product non-emissions have not before been documented in patient populations. However, other patterns seen in the present study are clearly not differentiable from patterns exhibited by normal-hearing patients, or by patients afflicted with mild hearing impairment. I/O functions in cavities are highly reproducible within session (repeated testing without removing the probe from the cavity) where the standard deviations typically are less than 2 db (the standard deviation is related to the number of time averages used to measure the average DP), as well as between session spanning months where the standard deviation is less than 5 db. Unfortunately, a high degree of reproducibility (and its accompanying low measurement error) is one of the factors which makes this measurement clinically attractive, and this characteristic cannot be used to differentiate DP non-emissions from DPOAEs. When studied in detail, the I/O function of acoustic distortion products is characterized by a series of peaks and troughs spanning as much as 20 db with primary 44 frequency changes of as little as 1/32nd. octave (He and Schmiedt (1994 ). He and 44 Schmiedt (1994 ) have postulated that one major cause of I/O variability is related to the behavior of the distortion-product fine structure at high levels. If that is the case, then one possible mechanism for this variability may be the result of a slight frequency shift that occurs with changes in level. In this context, increasing stimulus levels may produce downward shifts in stimulus and/or DP frequency. When the downward frequency shift is into a DPOAE trough, a non-monotonic notch pattern (such as that seen in Figure 17 and Figure 18) may be expected. When the downward shift in frequency is toward a DPOAE peak, a change in slope with a higher-gain I/O function is predictable. Moreover, for frequencies exceeding 3 khz, the ratio of the measurement-plane sound-pressure level to the sound-pressure level at the ear drum changes very markedly with only slight changes 34 in frequency (Siegel, 1994 ). Distortion Products as a Function of L 1 and L2 D:\wp\doc\OAE.doc bytes rev.: Thursday 30 June 1994, at 04:12:07 Hours Page 28