1/'\ VeteransAdministration

Journal of Rehabilitation Researchand Development Vol . 23 No . 2BPR 10-44 Pages 27-33

Nucleus 22-channel cochlear implant:

Preliminary observations

MICHAEL S . HIRSHORN, M .D . : DIANNE J . MECKLENBURG, PH .D.;

JUDITH A . BRIMACOMBE, M .A.

Cochlear Corporation, Englewood, Colorado ; and Nucleus Limited, Sydney, Australia

technical note*

Abstract A carefully designed study was undertaken in 1982to evaluate the performance of individuals who received theNucleus 22-channel cochlear implant . All patients were pro-found-totally deaf, adults with a postlingual onset of impair-ment . The preoperative evaluation, prosthesis fitting, training,and postoperative testing were consistent across clinics . Sin-gle-subject studies, where each patient acted as his/her owncontrol, revealed that of the 37 subjects, 16–24 obtainedsignificant improvement (P~0 .001) on unpracticed, unfamil-iar recorded speech tests from the Minimal Auditory Capabili-ties (MAC) Battery, when using hearing alone (no lipreading).In addition, virtually all patients showed improvement inrecognition of speech material with lipreading . The data supportthe efficacy of a feature extraction coding system wherespecific formant and amplitude information are transmitted viadirect electrical stimulation to the cochlea.

INTRODUCTION

Profound deafness is a very severe handicap for which,until recently, very little treatment has been available.Since the 1970's clinical studies of several differentdesigns of cochlear implants have been undertaken toprovide sound to the profoundly deaf and have yielded

promising results . The objective of these devices hasbeen to bypass the damaged hair cells of the cochlea and

stimulate the cochlear nerve endings directly.The two most common approaches have been to

stimulate the cochlea with one electrode (single channel)

* "Technical Notes" are published in the Journal as a means ofexchanging information concerning an investigator's use of a particu-lar scientific instrumentation or procedure, which might further thecourse of research . While these original notes are subject to peerreview and represent an important contribution to the researchliterature, they lack controlled comparison studies and are thusdifferent from "scientific articles ."

27

or more than one electrode (multichannel) . Both systems

can be divided into those that deliver the speech signalwithout speech specific processing or those that extractthe speech features in some way (2) . The primaryrationales for multichannel stimulation are to providemore information and to take advantage of the tonotopicorganization of the cochlea.

The purpose of this paper is to provide a description ofa technologically advanced 22-electrode speech featureextraction type cochlear prosthesis and to present somepreliminary results of the clinical study to date.

THE CLINICAL PROGRAM

This 22-electrode system, which has been designedspecifically for enhancing speech understanding, hasnow been implanted in over 68 patients worldwideincluding 32 in North America, 28 in Australia, and 8 inWest Germany.

Preoperative Patient Selection

The primary criteria for selection of cochlear implantcandidates are 1) postlingually deafened ; 2) profound

deafness, bilaterally ; 3) 18 years of age or older ; 4) no

benefit from any sensory device (tactile or hearing aid) asdefined by less than 1 percent open-set discriminationwhen aided; and 5) positive CAT scan or tomogramdemonstrating patency of the basal turn of the scalatympani.

There are three stages of evaluation . If the patient isconsidered suitable after completing the first stage, he orshe progresses to the next . The initial steps (stage I) are

28

Journal of Rehabilitation Research and Development Vol . 23 No. 2 April 1986

FIGURE 1Implantable receiver/stimulator.

those taken to evaluate patients before they are consid-ered as implant candidates . During this stage the degreeof hearing loss is established and trials with new hearingaids are conducted if inappropriate amplification hasbeen provided in the past . It is in stage I that promontorystimulation is performed . It is administered by placementof an electrocochleography needle electrode on thepromontory. Those patients who do not obtain anysensation of sound in response to electrical stimulationare not considered candidates, at this time.

Stage II considers results obtained from speech dis-crimination testing using the Minimal Auditory Capabili-ties Battery and selected subtests from the Iowa CochlearImplant Battery (4, 7) . These tests are prerecorded by anunfamiliar speaker and are presented to potential candi-dates in a controlled fashion (i .e ., using standard au-diometric equipment in a sound field presentation) . Alltests are performed with a sensory device . For baselinepurposes a measure of speechreading is also obtained.

Stage III is the final step in patient selection and occursonly if the candidate meets stages I and II criteria . Itinvolves counseling for appropriate expectations, bothwith the patient and the family. Further, a tinnitusquestionnaire is administered, the surgical procedure isfully explained, and the patient is scheduled for implanta-tion. A more complete description of the patient selectionnrnracc oan }u. fn111-1A in

,,«,i D«; 1.,. /')\

Postoperative Testing and Training

Typically, 2 to 2½ weeks after surgery, fitting of thespeech processor, counseling, orientation, and training ofthe patient commences . Approximately 10 sessions ofadditional patient contact occur over the next 10 weeks.

The Patient System

The patient's device consists of the implantable receiv-er/stimulator (Fig . 1) with its 22-electrode array and theexternal electret microphone, transmitting coil, andwearable speech processor.

The smooth, flexible, free-fit electrode array carries 22pure platinum electrodes and 10 support rings on asilastic carrier. Each electrode is independently pro-grammable . The array tapers from a maximum of a0.6 mm diameter to a 0 .4 mm diameter ; it may beinserted up to 25 mm. The equally spaced activeelectrodes are located on the distal 17 mm of the silasticcarrier, and 10 support rings are located on the proximal8 mm . The receiver/stimulator receives the externallycoded stimulus information and generates charge-bal-anced, constant-current, biphasic stimulus pulses to

29

HIRSHORN ET AL . : Technical Note

SIMULATOR

MICROPHONE -HEADSETSPEECH

PROCESSOR

SPEECHPROCESSORINTERFACE

PATIENTCONTROLKNOB

SOFTWARE

KEYBOARD

DPUPRINTER

FIGURE 2Wearable speech processor.

Surgery

The surgery for implantation involves a 2- to 3-hourprocedure which is essentially a modified mastoidectomyand posterior tympanotomy with the addition of thedrilling of a bed for the receiver-stimulator in the mastoidand the insertion of the 22-electrode array into the scalatympani.

Figure 2 shows the speech processor, which is130 x 75 x 18 mm and weighs 210 gm when the batteriesare included. There are controls for adjusting the sen-sitivity and applying a squelch feature available to thepatient, as well as a variable LED output which providesboth a means of estimating battery strength and thepresence or absence of an input signal . Alignment of theexternal coil over the internal coil in the implant isfacilitated by a continuous test tone when the speechprocessor sensitivity dial is switched to the "T" position.

Sound is picked up through the microphone andtransmitted to the speech processor, where certain fea-tures of speech are extracted, passed through an analog-to-digital converter, coded, and sent along the headsetcable to a transmitter coil which is located in closeapproximation to the internal receiver coil . Electricalpower and the coded signals are transmitted through theintact skin to the receiver/stimulator by magnetic induc-tion, where the signal is decoded and electrodes areselected for stimulation.

Speech Processing Strategy

The feature extraction coding strategy estimates threenarameterc of cnperh . 11 The fundamental freauencv

FIGURE 3Block diagram of the complete Nucleus Diagnostic and ProgrammingSystem . The modified microcomputer is called the DiagnosticProgramming Unit (DPU).

(FO) determines the rate of stimulation at the electrode.2) The place of simulation is determined by an estimateof the frequency of the second formant . 3) The overallamplitude of the speech sample is estimated and convert-ed into current . For example, a high-frequency, secondformant speech segment will cause a more basal elec-trode to be selected . This coding strategy is based onstudies by Clark et al . (1) and Tong et al . (5). Theydemonstrated that patients were able to accurately associ-ate place of stimulation with different vowels and torelate changes in the rate (frequency) of stimulation tothe voice fundamental.

This strategy allows use of the normal frequencyorganization of the cochlea with its relative progressionof high to low frequency sensitivity from the basal to theapical end . Bipolar stimulation allows for better controlof the electrical stimulation site by passing currentbetween two specific locations . When stimulation occursat different points along the cochlea, different pitchpercepts are elicited.

Computer-Based Testing

The audiologist's computer-based testing system,called the Diagnostic and Programming System (DPS)(Fig . 3), is used to determine optimum parameters and toprogram the patient's speech processor to his or herindividual needs . The system consists of a speciallymodified microcomputer, a speech processor interface(SPI) designed to control the operation of the signalprocessor during psychophysical and speech testing, apatient control knob for altering the amplitude or rate ofstimulation of a given electrode, a simulator which

30

Journal of Rehabilitation Research and Development Vol . 23 No. 2 April 1986

provides a visual display of the electrode selected forstimulation and an acoustic simulation of the stimulusdelivered, and a printer for providing hard copies of thedata gathered for medical records.

Fitting the Device

The three primary psychophysical measurementsneeded to program the device are effective dynamicrange for each electrode pair, loudness balancing, andplace-pitch ranking.

Once these psychophysical measurements are ob-tained, they are compiled for programming and transfer-ring onto the memory chip (EPROM : erasable, pro-grammable, read-only memory) of the speech processor.All of the psychophysical testing is done with the samespeech processor, which the patient takes home.

Specially designed subroutines of the software are alsoavailable which allow the device to be fine tuned . Thepurpose of "fine tuning" is to optimally adjust thestimulation parameters for each of the 22 electrodes foreach patient ; this satisfies quality judgment as patientsgain more listening experience and become more de-scriptive reporters.

Psychophysical tests such as place-pitch ranking andloudness growth for each electrode allow for optimiza-tion of the speech processor program . In these tests, thespeech processor is connected to the diagnostic andprogramming system and is directly controlled by themicrocomputer keyboard or patient control knob . In thecase of pitch ranking, the objective is to ensure that thepatient perceives the pitch of the electrodes in order, fromhigh to low, as they are stimulated from the basal to theapical end of the cochlea.

The objective of loudness growth programming is togive the patient the most natural speech perception . Inother words, the stimulation amplitude which will occurfor a given input signal can be adjusted. The effectivedynamic range defines the limits of a patient's useablestimulation for sound sensation from the softest sensation(100 percent critical points) to the maximum comfortableloudness . Typically, this varies from electrode to elec-trode.

Training With the Speech Processor

To introduce the patients to the task of integrating andtransferring newly received electrically stimulated soundpatterns, exposure to normal listening conditions andnarticination in interactive communication is nee csary

This is accomplished through structured listening tasksprovided throughout a 40-hour rehabilitation program.During the training sessions patients are presented avariety of materials, both with and without the aid oflipreading . Careful introduction of specially designedspeech materials provides the patient with the opportu-nity to associate words and sentences with patterns ofelectrical stimulation . Rehabilitation designed for mul-tichannel implant systems differs from that used forsingle-channel devices . The primary difference is theincreased number of channels of information available tothe patient ; these allow for more complex patternrecognition by the multichannel cochlear implant reci-pient.

The main elements of the training program are aformal program of counseling, follow-up checks of theequipment, appropriate level practice materials, andintensive practice in the use of the processed signals in anormal communication situation, with the telephone andwith environmental sounds . Materials include closed setsof words and sentences ; contextually based open-setsentences ; and completely open-set presentations ofwords, phrases, and sentences.

There are 6 to 10 3-hour sessions . Two to threesessions are concerned with providing patient-specificprogramming, introduction to sound sensation, and coun-seling as to the use of the device in different listeningsituations . The remainder of the sessions are directedtoward specific sound experience exercises, speechtracking (with and without lipreading), continued equip-ment monitoring, and counseling.

RESULTS OF SPEECH DISCRIMINATIONTESTING

Patient Profile

All patients were postlingually deafened adults, rang-ing in age from 22 to 74 years, who had bilateral,profound hearing loss . The onset of profound deafnesswas 6 months to 50 years for these patients . All patientshad more than 90 dB loss unaided . None could receiveany benefit from an alternate sensory aid . No medical orradiological contraindications to surgery were noted, andof those tested with promontory stimulation, all demon-strated positive results . Although 68 patients have beenimplanted to date, data from only 35 to 38 patients arepresented here as not all individuals have undergonecmmnlete nnctsnruical evahintinnc

A

31

HIRSHORN ET AL. : Technical Note

TABLE 1Paired comparison of mean minimal auditory capabilities (MAC)

results for selected closed-set subtests.

N

Presurgical, Postsurgical, N Subjects withSignificant Improvement t(4 correct % correct

mean range mean range

Noise/Voice 37 68 0-98 86* 34-100 21Four-choice spondee 37 42 0-85 72* 0-100 21Vowel 37 29 0-55 46* 0-68 18Final consonant 37 32 0-63 49* 0-67 20

* Significant difference, P < 0 .001.t Binomial model, P < 0 .05.

Results

Presurgical and postsurgical speech results for record-ed materials with multichannel users were obtained fromsubtests of the Minimal Auditory Capabilities Battery(MAC Battery) and subtests of the Iowa CochlearImplant Battery. Essentially three categories of tests wereused: those that emphasize prosodic features in a closed-set format, closed-set speech tests, and open-set speechtests . The test conditions for the pre- and postsurgicalevaluations were the same . Detailed results of sevenselected subtests of the MAC Battery are found inTables 1—3.

All postsurgical data were obtained at three months ofexperience with the speech processor. Mean data for oneprosodic test and three speech measures of the MACBattery are shown in Table 1 ; all are in a closed-setformat . The noise/voice test was selected as beingrepresentative of prosodic elements as it appears to assessthe perception of the broad features of speech, such asamplitude variations and temporal cues . For this study,patients consistently scored above chance levels of 50percent on such tests for both presurgical and postsurgic-al conditions . An example of patient group performanceon these measures is shown by the noise/voice test . Themean score for 37 patients was 68 percent preoperatively

(range, 0—98 percent) and 86 percent postoperatively(range, 34—100 percent) . Most tests of suprasegmental(prosodic) features of speech can be performed bysubjects who receive basic auditory or tactile cues.

Closed-set segmental tests present a more difficulttask; in this case the subject is required to correctlyidentify an item from a group of selected alternatives.Each has a chance score of 25 percent since the task is toselect the correct item out of a set of four. Although somepatients demonstrate the ability to achieve high scores onthese subtests, results are not necessarily indicative ofdiscrimination of unfamiliar speech . Table 1 shows

results of three closed-set segmental tests . The four-choice spondee test is included here rather than spondeesame/different test as it may indicate the ability to makedecisions based on changes in the fundamental frequency.Mean score on the four-choice spondee for 37 patientswas 42 percent preoperatively, with a range of 0 to 85percent ; it was 72 percent, postoperatively with a rangeof 0 to 100 percent . Significant differences between thepresurgical and postsurgical measures using a t test werefound (P < 0 .001).

The mean vowel test score for 37 subjects was 29percent before implantation (range, 0—55 percent) and 46percent postsurgically (range, 0—68 percent) . The differ-ence between the preoperative and postoperative condi-tions was found to be significant (P < 0 .001) . For 37subjects, the final consonant subtest yielded a mean scoreof 32 percent preoperatively (range, 0—63 percent) and49 percent postoperatively (range, 0—67 percent)with a significant mean difference of 17 percent (t testyielded P < 0.001).

A summary of the presurgical and postsurgical open-set speech recognition test results obtained from thosepatients who have completed their evaluations is present-ed (Table 2) . These are most important measures as theynot only provide a basis for patient selection during thepresurgical evaluations, but they may be more indicativeof patient's abilities to use electrically elicited auditorysensations for the understanding of some conversationalspeech, postsurgically. For these open-set subtests, re-corded speech materials are completely unfamiliar, un-practiced items presented with no alternative choices.The patient repeats a spondee, monosyllabic word, orsentence.

Table 2 shows the means and range of scores on threemeasures : spondee, monosyllabic, and phoneme recogni-tion. For the 20-item spondee recognition test, thepreoperative mean for the group was 0 .9 percent with arange of 0 to 4 percent . Postoperatively, 17 of 37 showed

32

Journal of Rehabilitation Research and Development Vol . 23 No. 2 April 1986

TABLE 2Paired comparison of spondee recognition and NU #6 subtestsof minimal auditory capabilities battery.

Presurgical,% correct

Postsurgical, N Subjects withSignificant Improvement+% correct

mean range mean range

Spondee recognition 37 0 .9 0—4 10 .4* 0—36 17NU #6

word score 37 0 .2 0—4 3 .7* 0—22 16phoneme score$ 37 4 .1 0—13 16 .6* 0—39 24

* Significant difference, P < 0.001.t Binominal model, P < 0 .05.$ Chance = 4 .5 percent.

TABLE 3CID sentences.

Condition N Mean Range

Lipreading only 37 54 .9 14—85Implant only 37 12 .4 0—58Lipreading and

implant 37 82 .6 18—100

a significant improvement [binomial model, P < 0 .05(ref. 6)] with a mean of 10 .4 percent and a range of 0 to36 percent . The NU #6 word list is phoneticallybalanced and scored both by word correct (50 items) andphoneme correct (150 items) ; thus two sets of scores areobtained from one presentation of the NU #6 list.Patients were encouraged to guess at each item . Themonosyllabic word test was the most difficult of theMAC subtests ; patients typically were unable to obtainany score preoperatively, with a mean of 0 .2 percent anda range of 0 to 4 percent . Postoperatively significantimprovement was demonstrated by 16 out of 37 (bino-mial model, P < 0 .05) . The mean was 3 .7 percent, andthe range was 0 to 22 percent . The phoneme meanscores, derived from responses to the word test, representresults from an inherently closed-set task since there are alimited number of English phonemes from which tochoose . The chance score is 4 .5 percent and preoperativepatients scored a mean of 4 .1 percent with a range of 0 to13 percent . These scores increased in the postoperativecondition to a mean of 16 .6 percent with a range of 0 to39 percent . With the use of the binomial modeling of thetest scores, 24 of 37 patients demonstrated a significantimprovement when comparing the preoperative aidedbest-ear response to that of the cochlear implant(Table 2).

The results of testing when using CID sentencematerials as the stimuli are found in Table 3 . Threeconditions are shown• linrearlino alnne implant a1nnP

and lipreading combined with the cochlear implant . Theresults demonstrate that 22 out of 37 were able to achievea significantly higher score when comparing lipreadingalone to lipreading with the prosthesis, with a mean of54.9 percent and 82.6 percent, respectively . Most of thepatients were unable to obtain any score preoperativelywhen using the implant only (mean, 0 .2 percent),whereas postoperatively, 20 out of 37 obtained a signifi-cantly improved score (mean, 12 .4 percent ; range, 0—58percent) with hearing only (no lipreading).

Of the patients reported here, none achieved presurgi-cally a discrimination score on more than one of the fouropen-set tests (spondee, monosyllabic word, phoneme,and sentence recognition) . Fifteen of the thirty-sevenpatients obtained scores on all four open-set measures,postoperatively. The preoperative results for the open-setmeasures show an ability to recognize speech for virtual-ly all patients . Postoperatively, scores ranged from 0 to58 percent, depending on the type of speech stimuli used.

Further, evaluation of the patients informally and withspeech-tracking tests have also demonstrated that approx-imately 30 percent of the patients have significant speechunderstanding without lipreading.

CONCLUSION AND FUTURE DIRECTIONS

Severe and profound deafness is a very isolatinghandicap ; its presence may have far reaching effects onthe person's family, work, and social interaction . Today'sadvanced cochlear implant systems help to overcomemany of the difficulties which result from such hearingimpairments.

Presently, a 22-channel cochlear implant system de-signed to extract the most important features of speechhas been designed and tested by Nucleus Limited andPnrhlPar \ nrnnratinn A ctarnrlarrl Pvahlatinn nrnnram_

33

HIRSHORN ET AL . : Technical Note

ming, training, and postsurgical evaluation protocol hasbeen used by 13 centers on 68 patients . Of these patients,none failed to achieve hearing sensation and no devicehas failed to function . All but one patient listens with thedevice for an average of 10 hours per day. For virtuallyall patients who have completed the presurgical andpostsurgical speech battery, speech recognition scoreswere found to be significantly improved. Particularlynotable is the fact that approximately one-third of thepatients were able to achieve significant speech recogni-tion without lipreading.

In the future, more effective speech processing strate-gies and coding schemes will make it possible for even

REFERENCES

I . CLARK GM, BLACK R, FORSTER IC, PATRICK JF, TONG YC : De-sign criteria of a multiple-electrode cochlear implant hearingprosthesis . J Acoust Soc Am 63 : 631-633, 1978.

2. LEVITT H : Sensory aids for persons with hearing impairment : Atutorial review, in press.

3. MECKLENBURG DJ, BRIMACOMBE JA : An overview of the Nucleuscochlear implant program . Sem Hearing 41-51, 1985.

4. OWENS E, TELLEEN C : Speech perception with hearing aids andcochlear implants . Arch Otolaryn 107 : 160-163, 1981 .

greater improvements in the recognition abilities ofcochlear implant recipients for speech . Also, with ad-vancing technology, it is likely that cochlear prostheseswill become useful to a wider group of hearing-impairedindividuals who obtain very limited benefit from hearingaids.

ACKNOWLEDGMENT

Special thanks to Jim Patrick, Richard Dowell, and Graeme Clarkfor their continued support and advice ; and to Leah Dunn and MaryMintz for their secretarial assistance.

5. TONG YC, BLACK RC, CLARK GM, FORSTER IC, MILLAR JB,O'LOUGLIN BJ, PATRICK JF : A preliminary report on a multiple-channel cochlear implant operation . J Laryn Otol 93 : 679-695,1979.

6. THORNTON AR, RAFFIN MJM : Speech discrimination scores mod-eled as a binomial variable . J Speech Hear Res 21 : 507-518,1978.

7. TYLER RS, PREECE JP, LOWDER MW : The Iowa Cochlear ImplantTest Battery . Department of Otolaryngology, Head and NeckSurgery, University of Iowa, 1983.

„