ENT Department of the University Medical School of Pécs, Hungary

**ENT Department of the Semmelweis University of Budapest, Hungary

1. Introduction

Reproducibility is a crucial criterion for the evaluation of the transient otoacoustic emissions (TEOAEs). In order to establish whether a true echo is present or not, Kemp (1978) introduced an overall reproducibility value, calculated from the cross-over correlation between the two curves recorded under the same conditions. Below +50%, the two curves give poor overall reproducibility, and so it is reasonable to believe that no echo can be demonstrated. However, the overall correlation can be low in spite of the existence of short-time emissions shown by transient correlations between the two traces in certain time domains. Employing only an overall reproducibility value as a criterion, such temporary, short-time emissions may be overlooked.

Thornton (1991) reported on the moving time window method, which seems to be helpful in detecting temporary correlations in the emission curves by analysing their short time-segments. This method can detect parts with emission even if overall reproducibility is below 50%. 

We have introduced a similar technique during off-line analysis of TEOAE, enabling the determination of TEOAE duration. The aim of this study was to test moving time window analysis (MTWA) for clinical application.

We have studied the duration of TEOAE in different groups of patients to find the cause of the difference in the TEOAE duration and to define a critical value for minimum OAE duration, which could be used as a new criterion for the evaluation of TEOAE.

2.    Method

Subjects

1. Adults with normal hearing (36 ears, male/female = 1/1; mean age 39 years). In this group middle ear function was normal by tympanometry, and hearing thresholds were lower than 10 dB nHL at all frequencies by conventional audiometry.

2. Adults with sensorineural hearing loss of varying severity (75 ears, male/female = 1/1; mean age:  44 years). Tympanograms were  normal, and at one or more frequencies sensorineural hearing loss (> 25 dB nHL) was found. Ears selected from this group were classified into subgroups according to the width of the missing frequency band measured by conventional audiometry. Fifty ears satisfied the criteria for this selection.

Subgroup 1: sensorineural hearing loss (> 25 dB nHL) at 4 and 8 kHz. Subgroup 2: hearing loss at 2,  4 and 8 kHz. Subgroup 3: hearing loss at 1, 2, 4 and 8 kHz. Subgroup 4: hearing loss at 0.5, 1, 2, 4 and 8 kHz.

Tympanometry was carried out with a GSI middle ear analyser, threshold audiometry with Peters 6 standard clinical audiometer. TOAE measurements were made, using a standard ILO88 device (Otodynamics Ltd., England) in combination with an IBM compatible AT personal computer. The subjects were tested in a quiet laboratory environment with a standard probe generating differential clicks of 0.1 ms duration. The options were set as detailed by Harris, Probst and Wenger (1990). Nonlinear mode was used, stimulus level and probe position were constantly monitored, and gain was set so that the stimulus level reached 86 dB SPL.

Moving time window analysis (MTWA)

The ILO88 analyser measures two independent emission curves, which are displayed and their correlation is given by the original software as the reproducibility of the TOAE measured. For an off-line analysis of the curves recorded by the ILO88 device we compiled a programme. During scanning of the double emission curve with 1 ms windows, in every section a correlation coefficient was computed and graphically displayed along the time axis, wich resulted a curve running between +1 and -1 (Fig.1.). Above +50% the correlation was considered acceptable. The time domains where the values of the correlation curve were above this limit were regarded as containing emission. Mean values of the original curves, where emissions were demonstrated according to the MTWA, were displayed separately below the correlation curve. Below this the original ILO88 curve is shown. Our software calculates the longest continuous time-interval where the correlation is above 50% (relative TEOAE-duration). The last point of the continuous curve can be regarded as the end of TEOAE or absolute TEOAE duration. Absolute duration is calculated from the beginning of the stimulus. Relative duration is calculated from the first CC value above 0.5 (50%) after the 2.5 ms Kemp-window. Generally the relative duration starts at 2.5 ms, but occasionally later. An MTWA curve broken by a short valley (shorter than 0.4 ms) was regarded as a continuous MTWA curve. Effective duration is calculated after 2.5 ms, from the sum of all parts having CC higher then .5 (50%). The moving window concepts are illustrated in Figure 1.

Figure 1: Moving time window analysis (A) of a TOAE measured with the ILO88 device(C).
Where the moving time window correlation curve runs over 50% (A), the TOAE time-domain is plotted as a real emission (B).

3.    Results

1. Adults with normal hearing (36 ears):

Absolute TOAE duration in this group was (± SE) 15.1 ± 0.69 ms. Minimal duration was 8.2 ms, maximal duration 19 ms, full sweeptime.
Relative TOAE-duration was 13.1 + 0.56 ms. Minimal duration, 5.72 ms, maximal duration, 16.5 ms. Four ears showed full sweeptime OAE duration.


2. Adults with sensorineural hearing loss (75 ears):

Mean absolute TOAE duration (± SE) was 9.34 ± 0.43 ms. Mean relative TOAE duration (± SE) was 6.1 ± 0.41 ms. Both values differ significantly from the values in normals (p < 0.05).The absolute and relative duration values of TOAE in adults with normal hearing and in 4 subgroups with different width of frequency band, where sensorineural hearing loss was measured, are shown in Figure 2. Figure 3 presents the values for absolute and relative duration as a function of each missing frequency. For this figure the mean values of duration were calculated for every single missing frequency.

Figure 2:TEOAE absolute duration as a function of hearing loss at different frequencies
* = significant difference at the 0.05 level between two adjacent bars.

Figure 3:Absolute and relative duration of TOAE versus single missing frequencies. * = significant difference at the 0.05 level between two adjacent bars.
(Dark bars=absolute duration, lined bars=relative duration)

4.    Discussion

We used a moving time-window correlation calculation to determine the duration of transient otoacoustic emission in subjects with normal hearing and patients with sensorineural hearing loss. According to the normal adult group, MTWA always lasted longer than 8 ms.

In patients with sensorineural hearing loss, both the relative and absolute duration of TEOAE were significantly shortened. The breakdown of this group into subgroups according to the width of the missing high frequency band showed a correlation between the severity of the hearing loss and TEOAE duration. The correlation is not linear, a sudden decrease being observable in the group where the hearing loss extends to the 1 kHz frequency band. We plotted the values of absolute and relative duration for each missing frequency independently of other missing frequencies (fig. 3). In this figure the short values hovered around the speech frequencies, in the range of 0.5-l-2 kHz. The shortest duration was found in cases where the hearing threshold was higher at 1 kHz. These data suggest that the integrity of that part of Corti's organ, where perception of the speech frequencies (especially 1 kHz) is located, is very important in the generation of TEOAE.

Other authors analysing the correlation of TEOAE threshold and frequency, with subjective audiograms have reported similar results. According to Collet, Gartner, Moulin, Kauffmann, Disant, and Morgon (1989) the presence of TOAE indicates middle frequency functional integrity of the outer hair cells of Corti's organ. Avan, Bonfils, Loth, Narcy and Trotoux (1991) have found that TEOAE threshold and amplitude are altered only "when the hearing loss is already significant and associated with a change in audiometric threshold at 2 kHz".

By introducing a new parameter, namely the duration of TEOAE, we obtained the same results. The greatest duration was seen in cases of sensorineural hearing loss around 1 kHz. This fact suggests that TEOAEs are generated at definite sites, but they also need a good global cochlear status to propagate back to the oval window and become detectable (Avan, Bonfils, Loth, Narcy and Trotoux, 1991). Tavartkiladze (1991) assumed that the sources of TEOAE are dominantly in a narrow range in the cochlea, where the perception of 1 kHz occurs. Our results agree with his observation.

Regarding the clinical significance of duration as a parameter for the evaluation of TEOAE, we believe that it should be used for the identification of high correlation, emission positive parts on the TEOAE curve.

In summary :

1. If the absolute duration of OAE is greater than 8 ms, then the hearing threshold at the speech frequencies must be lower than 30 dB.

2. If the absolute duration of OAE is greater than 6.5 ms, then the cochlea must have a short normally functioning part, which holds out the prospect of a good prognosis as to hearing level, that is, there must be some frequencies where the hearing threshold is below 30 dB. This is of particular importance in childhood.

3. The most important site of OAE generation is that part of Corti's organ which is tuned to speech frequencies.

6.    References

Avan, P., Bonfils, P., Loth, D., Narcy, Ph., Trotoux, J.: Quantitative assessment of human cochlear function by evoked otoacoustic emissions. Hearing Research, 52:99-112 (1991)

Collet, L., Gartner, M., Moulin, A., Kauffmann, I., Disant, F., Morgon A: Evoked otoacoustic emission and sensorineural hearing loss. Arch Otolaryngol Head and Neck Surg, 115:1060-1062 (1989)

Harris, F.P., Probst, R., Wenger, R.: Repeatability of transiently evoked otoacoustic emissions in normally hearing humans.Audiology  30:135-141 (1990)

Kemp, D. T.: Stimulated otoacoustic emissions from within the human auditory system. J Acoust Soc Am,  64:1386-91 (1978)

Pytel J, Bauer M: A self-developed method for measuring otoacoustic emission. Fül-orr-gégegyógyászat,  38:9-16 (1992)

Tavartkiladze, G.A., Frolenkov, G.I., Kruglov, A.V.: On the site of the otoacoustic emission generation. XII. IERASG Symposium Abstracts p.58. (1991)

Thornton, A.R.D.: Click evoked otoacoustic emissions in hearing loss XII. IERASG Symposium, Abstracts p. 57. (1991)