Far-field cochlear microphonics in man and their relation to cochlear integrity

Far-field cochlear microphonics in man and their relation to cochlear integrity

86 Electroencephalography and clinical Neurophysiology, 1983, 5 6 : 8 6 - 8 9 Elsevier Scientific Publishers Ireland. Ltd. Short communication FAR-~...

368KB Sizes 0 Downloads 37 Views

86

Electroencephalography and clinical Neurophysiology, 1983, 5 6 : 8 6 - 8 9 Elsevier Scientific Publishers Ireland. Ltd.

Short communication FAR-~ COCHI.gAR MICROPHONICS IN MAN AND ~ R INTF~RITY

RELATION TO COCHLEAR

M A N F R E D E U L E R * and J U R G E N KIESSLING **

• Fb.lO Physik-Technologie, UniversitiR Duisburg, D-4100 Duisburg, and ** University ENT-Clinic, Universiti~t Giessen, D-6300 Giessen (F.R. G.) (Accepted for publication: February 28, 1983)

Although the relation of cochlear microphonics (CM) to the integrity of outer haircells is well established (Dallos et al. 1972), its use for detecting hearing disorders caused by cochlear damage is generally regarded as limited (Eggermont 1976). C M records from the promontory, the ear canal or from more remote sites do not appear to be simply related to cochlear function, because contributions from haircell generators along the basilar membrane interfere in a complicated way (Aran and Charlet de Sauvag¢ 1976; Hoke 1976; Elbefling 1977). Recent work has led to the impression that the CM recorded by skin electrodes is generated mainly in the basal turn, even in response to low frequency stimuli (Sohmer et al. 1980). All these conclusions are based on C M responses to isolated test frequencies and must be re-evaluated if one wishes to interpret CM characteristics measured continuously with a sweeping frequency. The present results on far-field CM components point to apically located frequency-specific haircell sources, at least at certain frequency windows, and allow the clinically useful application of CM signals.

Method Instead of the conventional time domain averaging we use frequency selective lock-in analysis to extract CM signals from background noise. Our experimental arrangement is comparable to the one used previously in frequency following potential (FFP) measurements (Euler a n d Kiessling 1981). Compared to our earlier work, we have improved the stimulus section, so that acoustic stimulation by a closed acoustic system is possible. A double wall (2 × 1 ram) mu-metal case effectively shields the headphone. Tests with deaf ears established that no electromagnetic artifacts are generated up to 120 dB SPL. Additionally we took care to correct headphone and external ear transfer functions by measuring stimulus amplitude and phase close to the eardrum with a probe microphone. Thus, only middle ear transfer effects remain uncorrected in the present data. Stimulation is performed by a continuous tone, the frequency of which is slowly swept from 0.3 to 4.5 kHz. C M signals are picked up by surface electrodes between the ipsilateral mastoid and the forehead with the contralateral mastoid as ground.

From our earlier measurements (Euler and Kiessling 1981) it is known, that below 1 kHz neural F F P sources interfere with C M signals. As the neural F F P latencies are about a factor of 10 longer than CM delay times a quickly oscillating neural F F P pattern is created on the smoothly varying C M vector components. By using sweep rates appropriate for CM detection (5-10 Hz/sec), these neural F F P patterns are smeared out with lock-in bandwidth settings of 0.03 Hz (4 see time constant).

Results and Discussion Fig. la shows the two CM vector components A - s i n q0 and A. cos q0 of a normally hearing subject. From these components the frequency-dependent amplitude response A(w) and phase lag ¢p(w) can be calculated. By subtracting the phase response of the transducer-ear canal system the corrected phase lag of Fig. 2 is obtained. The phase function exhibits a two-step structure with two sections of relatively invariant phase. For the range in between a linear approximation yields C M group latencies t~ = - d c p / d ~ of I msec at 1.2 kHz and 0.55 msec at 3 kHz. This frequency dependence of group latencies (dispersion) has been rq~roducod in 10 normal hearing subjects with a variation of +0.1 msec. Obviously, normal cochleae can be characterized by such a ' normal' dispersion effect and by phase functions of the type shown in Fig. 2. We are able to pick up C M s up to frequencies of 10 kHz for stimulus intensities between 90 and 100 dB. However, above 5 kHz phase correction becomes difficult because of sensitivity to probe microphone position. There are indications of a third linear phase lag section between 6 and 8 kHz that yields group latencies of 0.3 msec. Due to the inherent difficulties of appropriate phase correction its reproducibility is poorer. That is why we have concentrated our interest on the frequency range below 4.5 kHz. The phase function is rather insensitive to the stimulus level in the range investigated (70-100 dB SPL). A few subjects show a slight reduction o f phase lag below 1.5 kHz at levels of 95 dB SPL or more. The CM amplitude-frequency dependence in Fig. 2 shows a periodic structure in normal hearing subjects. The minima

0013-4649/83/0000-0000/$03.00 © 1983 Elsevier Scientific Publishers Ireland, Ltd.

FAR-FIELD COCHLEAR MICROPHONICS IN MAN

ul

87

-- U-A.sin? 1.5

@

--- U = A .cos/,

A orb. units

+5011!~' r~V

1 I

0

!

• I i

!

I

0.5

!

! #

!

- 50

© *50

I

/

II

o

o.51

2 4[kH4

40 A

phase tag

~a

2¢(

i -50

I

d

I 1

I

I

I

I

1

:2

3 kHz

/~

frequency Fig. 1. Vector components of cochlear microphonics measured continuously as a function of frequency (95 dB mean SPL). a: normal hearing, b: noise-induced hearing impairment; the insert shows the typical dip in the pure tone audiogram.

coincide with the constant phase sections, whereas the regions of maximum delay with approximately linearly increasing phase lag sections are in close vicinity to the amplitude maxima. The amplitude function shown has been corrected for an equal sound pressure of 95 dB close to the eardrum, according to the mostly non-linear input-output (IO) functions. The type of IO function is strongly frequency dependent and exhibits a periodic recurrence with frequency in a close correlation with the periodic variation of phase and amplitude functions. IO curves with minimum slopes of 0.2-0.3 (double logarithmic plot) are measured in the regions of amplitude and latency minima. At the frequencies of maximum latencies IO functions with slopes

I 2

I 3 kHz

I

A

frequency Fig. 2. Amplitude and phase functions of cochlear microphonics derived from Fig. la for a normally hearing subject. The curves have been corrected according to stimulus amplitude (95 dB SPL) and phase close to the eardrum. close to unity (0.9-1) occur, which tend to saturation at 90-95 dB SPL. Yoshie and Yamaura (1969) reported the general trend that non-linearity (slopes different from unity) is common in recordings with electrodes more remote from CM sources, indicating that the amount of non-linearity depends on the extent of the region sampled by the electrodes. In the present typical far-field configuration of skin electrodes we find linearity coinciding with CM latency and amplitude maxima. This is an indication of local specifity of the CM responses at these frequency bands. Furthermore, the present group latencies of 1 msec at 1.2 kHz and 0.55 msec at 3 kHz prove the existence of more apically located frequency-specificCM sources. Our CM phase and latency data represent mean values derived from an averaging process over an unknown weighted distribution of haircell generators along the basilar membrane. The above maximum

88 latencies agree well with frequency-dependent cochlear delay times. Eggermont (1979) arrives at an estimation of t = f-f0 5(fef in kHz, t in msec) based on worst case approximations of narrow band action potential data in recruiting ears, where the second filter mechanism is apparently not functioning. This agreement allows the conclusion that in the present far-field electrode arrangement certain frequency windows exist, where the averaging process is performed in such a way that approximately frequency-specific responses prevail. Thus, we can expect the responses of 1.2 and 3 kHz to reflect activity mainly from haircells tuned to these frequencies, provided the CM phase is in the normal range. To avoid misinterpretations concerning the actual frequency specificity obtained by the present method, we point out that the window concept indicates not absolute but only a certain degree of place specificity due to the averaging process and the high stimulus levels, which are necessary to elicit a measurable far-field response well above the noise level. There is some uncertainty in judging the actual amount of frequency specificity, as the travelling wave propagation time in h u m a n s is far from being well established (Eggermont 1979). The above formula is an upper limit. A more conservative estimate by the same author ( t = 1.9. f~-0.7) yields latencies increased by roughly a factor of 2. As a consequence this approximation, together with the present group delays, would point to an averaging process from a haircell distribution with its centre of gravity actually 1-2 octaves higher than the measuring frequency. There are two different mechanisms by which cochlear filter broadening at high stimulus levels can influence the far-field CM phase: the CM excitation pattern makes dramatic excursions into more basal parts; thus, contributions from short latency high frequency haircells increase and lower the apparent group delay; the non-linear level-dependent broadening of the cochlear filter is correlated with a flattening of the phase response. Accordingly group delays will decrease (Goldstein et al. 1971). As both effects go in the same direction it is difficult to separate them experimentally. More specific responses can be expected by reducing the stimulus level far below 70 dB SPL. Recent measurements on basilar membrane tuning in the cat cochlea ( K h a n n a and Leonard 1982) indicate mechanical tuning almost as sharp as neural tuning. Thus we should expect a further increase in C M delays at lower levels. Unfortunately, the rapidly deteriorating signal-to-noise ratio does not allow us to reduce the stimulus level any further than 70 dB. We have tested the assumption of nearly frequency specific windows (defined in the above sense) with pathological cochleae and found excellent agreement between deviations in the pure tone audiogram and abnormal CM vector components in those cases where outer haircell damage is obvious (n0ise-induced hearing impairments), Fig. lb shows a typical C M response from a cochlea with a restricted region of pathological change (dip in the 4 - 5 kHz region due to noise exposure). Below 1.5 kHz C M vector components are normal with slightly increased phase lag. Above 2 kHz CM amplitude is reduced and the phase relation is abnormal, indicating latencies of 0.2 msec or

M. EULER, J. KIESSLING less. From these short latencies it is clear that the responses are generated in more basal cochlear regions beyond the audiometric notch, where outer haircells are not damaged. This small amplitude component of short latency is completely absem in those cases where the audiogram shows a steep decrease instead of a dip. In addition to noise-induced hearing loss we also investigated several cases of presbycusis, showing considerably reduced CM amplitudes and phase lag increases of ¢r and more. In cases of Meni~re's disease reduced CM amplitudes with normal phase curves have been found. So we get a kind of typical fingerprint for different pathological cochlear states. Apart from the periodic CM dispersion effect another fact forces us to give up the model of coherent movement of the basal basilar membrane section as the main C M source. The presence of CM responses to frequencies above 5 kHz demonstrates that even short wave length excitation patterns, confined to only part of the basal turn. are detected by electrodes in the far-field configuration. This result has led us to a simple model of far-field CM generation, which qualitatively explains the periodic variauon of C M amplitude and phase characteristics with frequency and the existence of frequency specific windows, where the recording electrodes effectively 'see' CM responses from the basilar membrane vibration maximum, or at least from regions in the vicinity of it. The basilar membrane vibration pattern, transformed into an electrical pattern by haircell activity, represents a multipole source of complicated geometrical structure. We assume for its far field that only its dipole moment is detected, because higher multipole moments vanish more rapidly with distance. As the stimulus frequency is decreased, the centre of the vibration pattern moves farther apically into the cochlear spiral and the resulting dipole moment will rotate. Depending on frequency the dipole vector is oriented parallel to the axis given by the two electrodes or it points in a perpendicular direction, resulting in m a x i m u m or m i n i m u m response. In the case of the CM maximum, contributions from the best frequency region prevail. causing m a x i m u m CM latencies. Conversely, in the minima, contributions from the basal basilar membrane region predominate and the average latency is shifted towards lower values. The above dipole explanation can only give a qualitative impression and is intended to clarify the periodic ups and downs of CM amplitude and group delays, in addition to this qualitative model we have performed some preliminary model calculations of far-field CM responses based on multipole electrical patterns arranged along the cochlear spiral and using a homogeneous conductor approach. The travelling wave envelope and its phase response are parameters which can be varied to study different excitation patterns. The results confirm the periodic variations of CM amplitude and phase, in good agreement with the measurements. Thus our fundamental assumption of far-field CM contributions spread over the cochlea according to the excitation pattern and not only restricted to the basal turn is beyond doubt. Furthermore the calculations predict a rather high sensitivity especially of the ipsilateral electrode position. This was experimentally tested by moving the electrode, which altered the

FAR-FIELD COCHLEAR MICROPHONICS IN MAN amplitude and phase curves systematically. We found maximum responses with the ipsilateral electrode not exactly at mastoid centre but about 3 cm above. The results of the model calculations in comparison with the actual CM distribution and frequency, as well as level dependence, are in the state of iterative adaption and will be published in a forthcoming paper.

Conclusion

We are optimistic that a detailed analysis of CM components over the skull will lead to a complete evaluation of haircell function within different frequency windows. Furthermore, quantitative assessment of intact haircell distribution appears to be possible, provided that the problem of space specificity is solved by lowering the stimulus level and increasing the sensitivity of the method, or by a better theoretical understanding of the basilar membrane excitation pattern. In the present state the measurements provide typical abnormal dispersion and amplitude 'fingerprints' in cases of outer haircell damage. The method is highly promising in detecting early stages of inner ear damage by noise or ototoxic drugs and in discriminating between haircell and neural lesions. Another advantage is its routine non-invasive applicability which does not require sedation and works under normal laboratory conditions. As the integrity of the most peripheral stages of the auditory system (outer hairceils) is accessible to objective analysis especially in the low frequency speech region the method fills a gap in conventional brain stem audiometry, which is known to possess only poor frequency specificity. In combination with frequency selective testing of the neural input via frequency following responses (Euler and KJessling 1981) the low frequency auditory function can be tested. Summary Far-field cochlear microphonics (CM) in man have been measured as a function of frequency by phase-sensitive detection. CM amplitude, dispersion and input-output functions vary periodically with frequency and exhibit certain frequency windows, where responses are approximately frequency specific, in contrast to the commonly accepted model of basal turn CM generators as main sources. Pathological cochleae show abnormal amplitude and phase functions. The results offer new diagnostic possibilities, because haircell function can be evaluated by a routine non-invasive procedure down to the low frequency speech range.

Rrsum6 Potentiels microphoniques cochliaires en enregistrement lointain chez I'homme, et leur relation avec l'intdgritd de la cochlde

On a enregistrr, chez l'homme, les potentiels microphoniques cochlraires (CM) /I distance, en fonction de la

89 frrquence, ceci par drtection phase-sensible. L'amplitude du CM, sa dispersion et les fonctions entrre-sortie varient prriodiquement avec la frrquence et trmoignent de certaines fen~tres frb.quentielles, dans lesquelles les rrponses sent pratiquement sprcifiques de la frrquence. Cette observation va h I'encontre du modrle classique, selon lequel ce sent les grnrrateurs du tour basal qui seraient sa source essentielle. Des cochlres pathologiques ont foumi des fonctions d'amplitude et de phase anormales. Ces rrsultats apportent de nouvelles possibilitrs de diagnostic, dans la mesure o6 le fonctionnement des cellules cilires peut 6tre apprrcire par une mrthode non invasive, jusque darts le domaine des basses frrquences de la parole.

References

Aran, J.M. and Charlet de Sauvage, R. Clinical values of cochlear microphonic recordings. In: R.J. Ruben, C. Elberling and G. Salomon (Eds.), Electrocochleography. University Park Press, Baltimore, Md., 1976: 55-65. Dallos, P., Billone, M.C., Durrant, J.D., Wang, C.Y. and Raynor, S. Cochlear inner and outer haircells: functional differences. Science, 1972, 177: 356-358. Eggermont, J.J. Electrocochleography. In: W.D. Keidel and W.D. Neff (Eds.), Handbook of Sensory Physiology, Vol. V/3. Springer, Berlin, 1976: 626-705. Eggermont, J.J. Narrow-band AP latencies in normal and recruiting human ears. J. acoust. Soc. Amer., 1979, 65: 463-470. Elberling, C. Some Aspects of Electrocochleography (ECoG). Fadl's Forlag, Copenhagen, 1977. Euler, M. and Kiessling, J. Frequency-following potentials in man by lock-in technique. Electroenceph. clin. Neurophysiol., 1981, 52: 400-404. Goldstein, J.L., Baer, T. and Kiang, N.Y.S. A theoretical treatment of latency, group delay, and tuning characteristics for auditory-nerve responses to clicks and tones. In: M.B. Sachs (Ed.), Physiology of the Auditory System. National Educational Consultants, Baltimore, Md., 1971: 133-14 I. Hoke, M. Cochlear microphonics in man and its probable importance in objective audiometry. In: R.J. Ruben, C. Elbeding and O. Salomon (Eds.), Electrocochleography. University Park Press, Baltimore, Md., 1976: 41-54. Khanna, S.M. and Leonard, D.G.B. Basilar membrane tuning in the cat cochlea. Science, 1982, 215: 305-306. Sohmer, R., Kinarfi, R. and Gafni, M. The source along the basilar membrane of the cochlear microphonic potential recorded by surface electrodes in man. Electroenceph. olin. Neurophysiol., 1980, 49: 506-514. Yoshie, N. and Yamaura, K. Cochlear microphonic responses to pure tones in man recorded by a non-surgical method. Acta otolaryng. (Stockh.), 1969, 252 (Suppl.): 37-69.