Memoire Online - Contrainte Psycho-Physiques et Electrophysiologiques sur le codage de la stimulation électrique chez les sujets porteurs d'un implant cochléaire

Nous avons comparé le pourcentage de reconnaissance sans lecture labiale avec les latences et intervalles des ondes II, III et V des PEAEP des 17 sujets.

Les résultats montrent de fortes corrélations entre la performance d'une part et, la latence de l'onde V (R=0.57 p=0.017), l'intervalle II-V (R=0.68 p=0.003), l'intervalle III-V (R=0.69 p=0.002) d'autre part. Nous avons ensuite modélisé ces corrélations par une fonction linéaire avec une régression linéaire multiple pas à pas. Seul l'intervalle III-V à été choisi comme modèle. La relation entre les deux paramètres est Intervalle III-V (ms) = 2.01 - 0.378 Reconnaissance (N=17 R= 0.69 p=0.002). Cette relation explique 48 % de la variance (R²).

La relation qui existe entre l'intervalle III-V et la reconnaissance de la liste de mots sans lecture labiale peut s'expliquer de la manière suivante. Un allongement anormal du temps de propagation entre l'onde III (reflétant principalement l'activité du noyau cochléaire homolatéral) et le complexe IV-V (reflétant principalement l'activité du colliculus controlatéral) se traduit par une mauvaise compréhension de la parole.

Le fait que le modèle linéaire décrivant les relations entre l'intervalle III-V et les performances des sujets porteurs de l'implant cochléaire sans lecture labiale explique environ 48 % de la variance est intéressant. La mise en place de la technique de recueil de PEAEP avant l'implantation (avec une électrode temporaire sur le promontoire ou sur la fenêtre ronde) permettrait de pouvoir estimer les performances des sujets après l'implantation.


ELSEVIER	Electroencephalography and clinical Neurophysiology 108 (1998) 521-525

Cochlear implant performance and electrically-evoked auditory brain-stem
response characteristics

^dUPRESA--CNRS 5020, Pavillon U, Hôpital E. Herriot, 3 place d'Arsonval, 69437 Lyon, France
^bLaboratoires MXM, 2720 chem. St Bernard, Vallauris, France
`Service ORL, Hôpital Avicenne, Bobigny, France
^dService ORL., Pavillon U, Hôpital E. Herriot, Lyon, France

Objectives: The purpose of this study was to find a correlation between cochlear implant performances in phoneme discrimination and activity of the brain-stem.

Methods: Electrically-evoked auditory brain-stem responses (EABRs) and speech recognition performances were measured in 17 patients implanted with an MXM Digisonic DX10 cochlear implant. Speech recognition performances without lip-reading were tested using lists of isolated French words containing 3 phonemes.

Results: The results indicated statistically significant correlations between phoneme correct-identification scores and the following EABR variables: wave V latency, wave II-V latency interval and wave latency interval. These results, indicate that up to about 48% of the variance in isolated word recognition without lip-reading can be accounted for by EABR variables.

Keywords: Cochlear implant; Electrically evoked auditory brain-stem response; Speech recognition

Speech recognition performances are largely variable across cochlear implant (CI) subjects: while some patients can understand running speech in noisy backgrounds without the aid of lip-reading, others cannot recognize simple sentences spoken slowly and quietly. The speech recognition performances of a given CI subject may depend on variable factors, such as the etiology of deafness, the duration of sensory deprivation, and the language-acquisition level before implantation and motivation, and it is difficult to find a reliable predictor. The physiological state of auditory system structures appears to be an important underlying factor of overall performance with the implant. The number

* Corresponding author. UPRESA--CNRS 5020, Pavillon U, Hôpital E. Herriot, 3 place d'Arsonval, 69437 Lyon, France. Tel.: +33 4 72110530; fax: +33 4 72110504. of surviving cells in the spiral ganglion, which generaily correlates well with the duration of deafness (Otte et al., 1978; Schmidt, 1985), has been pointed out as a possible factor, not only of thresholds for electrical stimulation (Pfingst et al., 1980; Shannon, 1983; Pfingst, 1984), but also of speech recognition performances (Pauler et al., 1986) in CI subjects.

However, although results in the literature support the notion that neural survival might be reflected in the growth of electrically evoked brain-stem responses (EABRs) (Smith and Simmons, 1983; Lusted et al., 1984; Brightwell et al., 1985; Hall, 1990), the results of studies in which relationships between speech recognition performances and measures derived from EABR amplitudes, such as threshold, response growth, refractoriness and strengthduration functions have been investigated, have failed to evidence strong correlations (Abbas and Brown, 1991a,b). The results of recent studies suggest that other estimates of the neurophysiological state of peripheral and intermediate

auditory nervous structures, which are related to speech recognition performances, might be derived from late EABR wave latencies (Herman and Thornton, 1992; Gallégo et al., 1997a).

The present study sought to characterize further the relationship between speech recognition performances and EABR wave latencies in subjects implanted with the Digisonic DX10 cochlear implant.

Seventeen adult subjects (10 female, 7 male; aged between 24 and 67 years), implanted with an MXM Digisonic DX10 cochlear implant, participated in the study. Except for the fact that a minimum of 12 out of the 15 electrodes of the implant had to be active, i.e. to elicit auditory sensations when stimulated, no particular inclusion criterion was defined regarding the etiology of deafness, the duration of deafness and the time since implantation, so as to allow for a wide range of speech recognition performances in the study sample.

The MXM Digisonic DX10 (Fig. 1) is a multi-electrode transcutaneous cochlear implant (Beliaeff et al., 1994). The implanted part of the device consists of a coil receiver and 15 0.5 mm diameter electrodes spaced 0.7 mm apart. The electrode carrier is inserted surgically along the first turn of the cochlea, generally between the 5th and the 20th mm of the basal extremity. The stimulation mode used was the `common ground' mode in which one electrode is activated at a time, all the others being connected to the ground.

Stimuli used for ABR recordings were generated using an MXM Digistim portable stimulator powered with batteries and run from a PC optically coupled to a serial port. The stimulus generator contained in this system can produce electric pulses having an intensity of up to 3 mA with a 2

load (Gallégo et al., 1998a); in the present study, the pulse amplitude was generally fixed to 1 mA. Pulse durations were varied and ranged from 5 to 310 fis. The pulses were biphasic, the positive and negative phases being asymmetric.

Each stimulation channel can be modeled as a capacitor and a resistor in series. Precise setting of the capacitor allows accurate equilibration of the positive and negative charges, which is important in order to preserve the physiological integrity of the stimulated tissues (Gallégo et al., 1998b). EABR recordings were synchronized on the stimulus using the trigger signal provided by the stimulation system.

An ipsilateral 3-electrode A1/2-Fpz set-up was used, with the earth electrode placed onto the contralateral ear lobe. The stimulation and recording techniques differed slightly from those used for acoustically evoked auditory brain-stem responses. The stimulation rate was 60 Hz; though such a rate would be too high for acoustically evoked auditory brain-stem responses (Suzuki et al., 1986), it has been shown to be adequate for EABRs (Gallégo and Micheyl, 1998).

The stimuli were delivered to 4 different electrodes; generally, electrode numbers 4, 7, 10 and 13. Eight stimulation levels were used; as the threshold and maximum comfort levels for electrical stimulation may vary widely across electrodes and patients, these levels were specified as percentages of the dynamic range of the tested electrode, rather

high-frequency filtering is used for the early waves whilst low-frequency filtering is used for the late waves. Fig. 2 shows the filter transfer function for the 2 ms latency wave. An algorithm for automatic wave-latency measurement based on the search for inflection points in the first derivative of the signal, combined with knowledge-based rules (Gabriel et al., 1980; Fridman et al., 1982), was then applied. Waves II, III and V latencies and the interval between waves II and III, III and V, and II and V, were considered.

To assess phonemic recognition performances, lists of 34 3-phoneme French words (`Listes cochleaires de Lafon') were used. The words recorded on CD were generated using a CD-player and presented via speakers located about 50 cm from the implant microphone. Each word was presented only once and the subject had to repeat what he/she had heard. The number of phonemes correctly identified was counted by the experimenter and expressed in percentages.

Fig. 2. Example of digital filter transfer function used for the processing of EABR signais.

than as absolute intensities. They corresponded to 5%, 10%, 20%, 30%, 40%, 50%, 70% and 90% of the dynamic range, defined as the difference between the intensities corresponding to maximum comfort and threshold level, respectively.

Fig. 3. EABRs before (a) and after (b) digital processing. The automatic wave-latency estimation program indicated: LII = 1.22 ms, LIII = 1.98 ms, LV = 3.74 ms.

The full-scale range used for EABR recording was #177;50 p.V. Responses were filtered using a wide band (0.1-8000 Hz) analog bandpass filter. The averaging involved 1024 sweeps. In order to estimate test/re-test reproducibility, EABRs were recorded 3 times at each stimulation level. EABR reproducibility for a given stimulus intensity and electrode was calculated as the inter-correlation between 3 averages from 1 to 7 ms. EABR traces were considered reliable when the resulting inter-correlation coefficient was superior or equal to 0.15 and significant (P < 0.01); otherwise, the traces were rejected and another recording was performed.

Following recording, the EABR traces were submitted to further analysis consisting of digital signal processing. A first problem, which is particularly acute in the case of electrically evoked ABRs, is that of the stimulus artifact, the amplitude of which can be a million-fold larger than that of the response to be measured. Therefore, special care must be taken to avoid the saturation of the pre-amplification stage. A second general problem with EABR recordings is the fact that ABR are of very small amplitude (in the sub-microvolt range), and thus generally vulnerable to interference from a variety of physical and physiological sources of noise.

Digital filters can improve greatly the ABR signal-tonoise ratio (Urbach and Pratt, 1986; Moller, 1988), with the advantage of zero phase-shift (Boston and Ainslie, 1980). In the present study, the quality of ABR traces was enhanced using an original digital-processing scheme described in a previous paper (Gallégo et al., 1997b, 1998c). This scheme is based on the use of filters with different transfer functions; for each of the ABR waves

Fig. 4. Relationship between EABR wave III--V latency interval and speech-recognition scores.

Fig. 2 shows examples of EABR traces. The 3 upper curves (Fig. 3a) correspond to recordings prior to digital processing; the lower curve (Fig. 3b) represents the EABR after filtering and summing. Waves II, III and V can easily be identified.

Using the Kolmogorov-Smirnov normality test, the wave latencies and latency intervals, averaged across electrodes and levels in each patient, were found to be normally distributed; therefore, Pearson' s parametric correlation was used to test for relationships between correct phoneme identification scores and EABR wave latencies. Significant cor- relations were found between correct phoneme recognition scores and wave V latency (r = 0.57, P < 0.05, n = 17), wave II--V latency interval (r = 0.68, P < 0.005, n = 17) and wave III--V latency interval (r = 0.69, P < 0.005, n = 17). Wave V latency and wave II--V latency interval proved to be highly correlated with wave III--V latency interval (r = 0.77 and r = 0.90, respectively; P < 0.001 and n = 17 in both cases).

Stepwise multiple regression analysis involving all EABR variables indicated wave III--V latency interval as a powerful predictor of speech recognition scores, being able to account for about 48% of the variance in the data (r = 0.69, P < 0.005, n = 17) (Fig. 4). The two variables were related by the following equation: y = 2.83 -- 1.27x, where x and y correspond to the latency interval in ms and to the speech recognition score in percentages. Another expression for the relationship between the two variables was found to be: y = 1 -- 1.27(x -- 1.44); according to this equation, the speech recognition score was 100% when the

wave III--V latency interval was as small as 1.44, and it decreased towards 0% as the interval increased up to 2.33 ms.

The EABR wave latency values obtained in this study are in agreement with those available in the literature (Abbas and Brown, 1988). The variable-filter processing scheme used in the present and previous studies (Gallégo et al., 1996, 1997a) leads to clear EABR traces, allowing for accu- rate estimation of EABR wave latencies using an automatic detection algorithm.

The main result of the present study consists in a relationship between correct phoneme recognition scores on the one hand, and the latency of EABR wave V and the intervals between this latency and that of the two previous waves (II and III). The strongest predictor of speech recognition performances proved to be the wave III--V latency interval. These results generally agree with results from previous studies: in particular, Herman and Thornton (1992) evidenced a relationship between phoneme recognition performances and EABR wave V latency. More recently, Gallégo et al. (1997a) evidenced a correlation between speech recognition and wave III and V intervals.

One interpretation of the observed relationship between correct phoneme recognition percentages and EABR wave III--V latency intervals is that poor speech comprehension is connected with abnormal lengthening of the propagation time between the nervous structures that generate waves III and V. Wave III is commonly associated with the activity of the ipsilateral cochlear nucleus, while the generator of wave IV--V complex is thought to be more diffuse, involving in particular the contralateral inferior colliculus activity. A likely origin for such a lengthening in neural propagation time consists of neural degeneration. However, the fact that a relationship between EABR wave latencies and speech recognition performances was obtained specifically with wave V latency and the wave III--V latency inter- val, and not with earlier latencies or latency intervals, suggests a specific origin of the wave III--V latency interval increase. One such origin might be a reduced temporal synchrony of neural discharges at the output of the cochlear nucleus, leading to reduced temporal summation of inputs by inferior colliculus units, and consequently, increased response latencies of these units. The present results cannot be interpreted as an indication for the fact that the neurophysiological state of structures before the level of the cochlear nucleus is not an important factor of speech recognition performances in CI subjects; simply, the latencies of EABR waves associated to structures below the cochlear nucleus do not appear as a significant predictor of speech recognition performances. Because of the complexity of the neural processes involved in speech processing, relationships between electrophysiological variables and speech

recognition performances are more likely to be found at higher than at lower levels of the auditory system. Relationships between phoneme recognition performances and electrically evoked middle latency responses have recently been evidenced (Groenen et al., 1997); the results of the present study suggest that such relationships might find their origin at lower levels.

Whatever the mechanisms underlying the observed relationship between EABR wave V latency and wave III--V latency interval, the results of the present study indicate that speech recognition performances can be predicted to some extent by EABR variables. A possible application of these results might consist of pre-implantation prognostic of post- implantation speech recognition performances (Gallégo et al., 1998d). However, given that EABR characteristics appear to vary with the duration of implantation in both animais (Miller et al., 1995) and humans (Gallégo et al., 1998d), nothing yet warrants that EABR characteristics obtained using a temporary electrode installed on the promontorium or on the round window will show similar relationships with speech recognition performances.

Abbas, P.J. and Brown, C.J. Electrically evoked brainstem potentials in cochlear implant patients with multi-electrode stimulation. Hear. Res., 1988, 36: 153-162.

Abbas, P.J. and Brown, C.J. Electrically evoked auditory brainstem response: growth of response with current level. Hear. Res., 1991a, 51: 123-138.

Abbas, P.J. and Brown, C.J. Electrically evoked auditory brainstem response: refractory properties and strength-duration functions. Hear. Res., 1991b, 51: 139-148.

Beliaeff, M., Dubus, P., Leveau, J.M., Repetto, J.C. and Vincent, P. Sound processing and stimulation coding of Digisonic DX10 15-channel cochlear implant. In: I.N. Hochmair (Ed.), Advances in Cochlear Implant. Proceedings of the 3rd International Cochlear Implant Conference, Innsbruck, 1994, pp. 198-203.

Boston, J.R. and Ainslie, P.J. Effects of analog and digital filtering on brain stem auditory evoked potentials. Electroenceph. clin. Neurophysiol., 1980, 48: 361-364.

Brightwell, A., Rothera, M., Conway, M. and Graham, J. Evaluation of status of the auditory nerve: psychophysical test and ABR. In: R.A. Schindler and M.M. Merzenich (Eds.), Cochlear Implants. Raven, New York, 1985, pp. 343-349.

Fridman, J., John, E.R., Bergelson, M., Kaiser, J.B. and Baird, H.W. Application of digital filtering and automatic peak detection to brain stem auditory evoked potential. Electroenceph. clin. Neurophysiol., 1982, 53: 405-416.

Gabriel, S., Durrant, J.D., Dickter, A.E. and Kephart, J.E. Computer identification of waves in auditory brainstem evoked potentials. Electroenceph. clin. Neurophysiol., 1980, 49: 421-425.

Gallego, S., Micheyl, C., Berger-Vachon, C., Truy, E., Morgon, A. and Collet, L. Ipsilateral ABR with cochlear implant. Acta Otolaryngol. (Stockh.), 1996, 116: 228-233.

Gallégo, S., Truy, E., Morgon, A. and Collet, L. EABRs and surface potentials with a transcutaneous multielectrode cochlear implant. Acta Otolaryngol. (Stockh.), 1997a, 117: 164-168.

Gallégo, S., Collet, L. and Berger-Vachon, C. Electrically auditory brainstem responses (EABR): contribution of a filter adapted to the auditory system. J. Int. Fed. Med. Biol. Eng., 1997b, 35 (suppl. 1): 304.

Gallégo, S. and Micheyl, C. Relationship between auditory brainstem responses and intensity discrimination in cochlear implant patients. Behav. Neurosci., 1998 (in press).

Gallégo, S., Luu, B.L. and Berger-Vachon, C. Modelling of the electrical stimulation delivered by the Digisonic Multichannel cochlear implant. Adv. Modelling Anal., 1998a, 39(1): 39-53.

Gallégo, S., Beliaeff, M., Frachet, B., Ouayoun, M., Berger-Vachon, C. and Collet, L. Long-term change in threshold and comfort levels and dynamics in Digisonic cochlear implant bearers. 1998b (submitted).

Gallégo, S., Durrant, J., Collet, L. and Berger-Vachon, C. Numeric timevariant filters adapted to the recording of electrically evoked auditory brainstem responses (EABR). 1998c (submitted).

Gallégo, S., Truy, E., Berger-Vachon, C. and Collet, L. Electrically auditory brainstem responses in cochlear implant assessment: possibility and interest. 1998d (submitted).

Groenen, P., Snik, A. and van den Broek, P. Electrically evoked auditory middle latency responses versus perception abilities in cochlear implant users. Audiology, 1997, 36: 83-97.

Hall, R.D. Estimation of surviving spiral ganglion cells in the deaf rat using the electrically evoked auditory brainstem response. Hear. Res., 1990, 45: 123-136.

Herman, B. and Thornton, A. Electrically evoked-evoked auditory brainstem responses in cochlear implanted subjects (abstract). In: The Second International Cochlear Implant Symposium, Iowa City, IA, 1992, p. 57.

Lusted, H., Shelton, C. and Simmons, S. Comparison of electrode sites in electrical stimulation of the cochlea. Laryngoscope, 1984, 94: 878882.

Miller, C.A., Faulkner, M.J. and Pfingst, B.E. Functional responses from guinea pigs with cochlear implants. II. Changes in electrophysiological and psychophysical measures over time. Hear. Res., 1995, 92: 100-111.

Moller, A.R. Use of zero-phase digital filters to enhance brain-stem auditory evoked potentials (BAEPs). Electroenceph. clin. Neurophysiol., 1988, 71: 226-232.

Otte, J., Schuknecht, H. and Kerr, A. Ganglion cell populations in normal and pathological human cochleae. Implications for cochlear implantation. Laryngoscope, 1978, 88: 1231-1246.

Pauler, M., Schuknecht, H. and Thornton, R. Correlative studies of cochlear neuronal loss with speech discrimination and pure-tone thresholds. Arch. Otorhinolaryngol., 1986, 243: 200-206.

Pfingst, B. Operating ranges and intensity psychophysics for cochlear implants. Arch. Otolaryngol., 1984, 110: 140-144.

Pfingst, B., Telman, S. and Sutton, D. Operating ranges for cochlear implants. Ann. Otol. Rhinol. Laryngol., 1980, 89 (suppl. 66): 1-4. Schmidt, J. Cochlear neuronal populations in developmental defects of the inner ear: implications for cochlear implantation. Acta Otolaryngol. (Stockholm), 1985, 99: 14-20.

Shannon, R.V. Multichannel electrical stimulation of the auditory nerve in man. I. Basic psychophysics. Hear. Res., 1983, 11: 157-189.

Smith, L. and Simmons, F.B. Estimating eighth nerve survival by electrical stimulation. Ann. Otol. Rhinol. Laryngol., 1983, 92: 19-23.

Suzuki, T., Kobayashi, K. and Takagi, N. Effects of stimulus repetition rate on slow and fast components of auditory brainstem responses. Electroenceph. clin. Neurophysiol., 1986, 65: 150-156.

Urbach, D. and Pratt, H. Application of finite impulse response digital filters to auditory brain-stem evoked potentials. Electroenceph. clin. Neurophysiol., 1986, 64: 269-273.

Les performances auditives des sujets porteurs d'un implant cochléaire présentent une grande variabilité. Les facteurs intervenant dans le succès de l'implantation sont le caractère pré- ou post- lingual de la surdité, le type de surdité, l'état de la cochlée, le nombre de fibres nerveuses survivantes, les capacités psychophysiques. La grande diversité des résultats peut également résulter de différences dans les capacités d'adaptation à la stimulation électrique, mais aussi dans les capacités cognitives des implantés.

La perception d'infimes différences auditives, essentielle pour adapter le traitement des signaux de parole, est sans doute un facteur primordial dans le succès d'une implantation cochléaire. L'utilisation de mesures électrophysiologiques qui évaluent les capacités de discrimination sont très intéressantes pour obtenir une adaptation objective de l'implant cochléaire aux capacités du sujet.

L'objectif principal des études précédentes a été d'évaluer l'intérêt des PEAEP en routine clinique pour aider au réglage de l'implant cochléaire. Il est vrai que les PEAEP ne sont pas les uniques et les plus fiables mesures électrophysiologiques qui permettent d'évaluer les caractéristiques psycho-physiques de chaque sujet implanté testé. Ils ont néanmoins l'avantage d'être fiables, robustes et rapides à obtenir.

Nous avons chercher à relier les caractéristiques des PEAEP à des seuils de détection, des seuils de confort, des discriminations en intensité et des reconnaissances phonétiques. Des études en cours, non présentés dans ce mémoire tentent de corréler des mesures de la tonie temporelle et de la tonotopie avec des recueils de PEAEP.

Les PEAEP sont des indicateurs partiels des certains paramètres tels que la tonie, l'intégration temporelle et la reconnaissance de la parole qui font intervenir des traitements plus complexes du système auditif. Les PEAEP doivent donc être couplées dans certains cas à d'autres mesures électrophysiologiques pour arriver à obtenir des estimations fiables des caractéristiques psycho- physiques.

L'utilisation des potentiels évoqués auditifs tardifs est tout aussi possible en routine clinique que les PEAEP:

L'amplitude des ondes N1P2 est ample (plusieurs pV), reproductible et assez robuste ; le nombre de stimulation nécessaire pour avoir une courbe fiable ne dépasse pas les 100, ce qui correspond à une durée d'environ 1 min 30 s.

Des études ont montré qu'il était possible d'estimer les performances des sujets implantés cochléaires avec l'analyse des ondes tardives (ref.).

Contrainte Psycho-Physiques et Electrophysiologiques sur le codage de la stimulation électrique chez les sujets porteurs d'un implant cochléaire

- Article 20 :

COCHLEAR IMPLANT PERFORMANCE AND ELECTRICALLY-EVOKED AUDITORY BRAIN-STEM
RESPONSE CHARACTERISTICS

Contrainte Psycho-Physiques et Electrophysiologiques sur le codage de la stimulation électrique chez les sujets porteurs d'un implant cochléaire

- Article 20 :

COCHLEAR IMPLANT PERFORMANCE AND ELECTRICALLY-EVOKED AUDITORY BRAIN-STEMRESPONSE CHARACTERISTICS

COCHLEAR IMPLANT PERFORMANCE AND ELECTRICALLY-EVOKED AUDITORY BRAIN-STEM
RESPONSE CHARACTERISTICS