Cochlear Implants: Fundamentals and Application - part 7 pptx

484 8. Engineering F IGURE 8.18. Top: The Cochlear Limited Nucleus 24 body-worn SPrint for the regular and research speech-processing strategies (Reprinted with permission from Cochlear Limited). Bottom: The Clarion Platinum body-worn speech processor (Reprinted with permission from Advanced Bionics Corporation). The Clarion behind-the-ear speech processor used rechargeable batteries, but this limited their running time. It is illustrated in Figure 8.19. The Med El Tempoם speech processor used three zinc-air batteries for approximately 36 hours operation. It dimensions were 6.6 ן 1.3 ן 0.9 cm., and came in straight, angled and children’s configurations. Receiver-Stimulators The receiver-stimulator needed not only to be designed with circuitry for low power consumption and hence long battery life, but also to allow a number of speech-processing strategies to be evaluated. If, for example, certain rates, wave- forms and stimulus patterns were found to be helpful, this should be possible without having to explant the device and replace it with a new one. The device should also provide charge-balanced, biphasic pulses to minimize corrosion of electrodes. Electronic and Communications Engineering 485 F IGURE 8.19. Left: The Cochlear Limited Nucleus 24 behind-the-ear ESPrit-3G for the regular speech-processing strategies (SPEAK, CIS, ACE) (Reprinted with permission from Cochlear Limited). Right: The Clarion behind-the-ear speech processor for the CIS strategy (Reprinted with permission from Advanced Bionics Corporation). Design Principles Constant Current Stimulation The first consideration was whether constant current or constant voltage stimu- lation should be employed. With constant voltage stimulation the interface im- pedance between electrode and tissue may change, in which case the current flowing through the nerve could vary. An increase in electrode impedance will result in reduction in current unless the voltage is increased (I ס V/R; Ohm’s law). The current amplitude, and therefore the amount of charge per phase, needed to remain constant for reliable stimulation. This could be avoided by constant current stimulation. Constant current stimulation allows the stimulating current to be specified, and ensures charge balance between the two phases of a biphasic pulse. Constant current stimulators are effective where there are small or moderate changes in electrode impedance. With high impedances they may not be able to provide the necessary voltage (i.e., they lose voltage compliance). This could lead to asymmetrical current pulses, and deliver net DC charge to the adjacent tissue with possible neural damage. Charge Balance The electrical stimuli produced by the receiver-stimulator should be charge-bal- anced, biphasic pulses to minimize the buildup of a DC current at the electrode– tissue interface. The DC current will lead to corrosion of electrodes, the produc- tion of toxic products at the electrode–tissue interface, and neural and other tissue damage. With the Nucleus 22 systems this was possible at high rates as they had circuitry to short any residual current between pulses, and at high rates there would not be time for this to occur. Capacitors in the circuit are effective in 486 8. Engineering preventing a buildup in the DC as their impedance is frequency dependent and approaches infinity for a zero frequency. As discussed in Chapter 4, with a si- nusoidal current the relationship between voltage (V ), current (I ), and capacitance (C ) and angular velocity (x) which is 2pf where f is frequency, is V ס I/2pfC. The term 1/2pfC is the capacitive reactance, and if the frequency becomes very low the impedance is proportionately high. Thus current flow will become infin- itesimally low. At the time when the Nucleus 22 implant was developed, capac- itors were large and high stimulus rates were not needed. Now they are smaller and with the Nucleus 24 implant there are two capacitors for the extracochlear electrode, so that for monopolar stimulation at high rates there is shorting between electrodes as well as capacitors to prevent the buildup of charge. The Clarion S and Combi-40ם have capacitors for each electrode, but do not use shorting be- tween pulses. Simultaneous Versus Nonsimultaneous A fixed-filter strategy that modeled the physiology of the cochlea and the neural coding of sound in the auditory nerve was tested (Laird 1979), but unsatisfactory results were obtained due to simultaneous stimulation of electrodes leading to channel interaction and unpredictable variations in loudness (Laird 1979; Clark, Blamey et al 1987). To avoid this the Nucleus F0/F1/F2 and Multipeak (Dowell et al 1987; Seligman 1987) strategies presented formant and spectral information nonsimultaneously (the pulses were separated by 0.8 ms) so that there was no summation of the electrical field. This also occurred with the IP strategy (Wilson et al 1988). An alternative explored was to stimulate with each phase of, say, three biphasic stimuli broken into nonoverlapping monophasic pulses (quasi- simultaneous stimulation). This was tested on an initial patient for the estimation of the first and second formants and gave similar results to the standard processor (McDermott 1989). Number of Stimulus Channels The place coding of frequency with multiple-channel implants is the main reason that results with these systems are superior to those of single-channel implants. The number of stimulus channels to be incorporated in the receiver-stimulator was therefore a matter of importance, but the optimal number has not been fully established. Multiple-channel devices were implanted with the number of channels varying from four with the Ineraid (Eddington 1983), to seven to eight with the Clarion S and Chorimac 8, to 12 with the Chorimac 12 and Combi-40 (Fugain et al 1984), and to 22 with the Nucleus 22 and 24 systems (Clark, Black et al 1978). Holmes et al (1987) found that open-set word recognition and continuous discourse track- ing results for the Nucleus F0/F1/F2 speech processor improved for the use of up to 15 electrodes. In a study by Blamey et al (1992) there was a positive correlation between speech perception and the number of electrodes in use, up to Electronic and Communications Engineering 487 21. For more details see Chapter 7. As there can be variations in the density of auditory neurons due to pathology, a further advantage for the Nucleus systems in having 22 electrodes is that there are more electrodes available in areas of the cochlea where place of stimulation is more effective. In determining the number of stimulus channels there is an interaction between mode of stimulation, elec- trode geometry, and cochlear anatomy for the optimal place coding of frequency. In contrast, Dorman et al (1989) showed that the number of stimulus channels for a fixed-filter (modified channel vocoder) system should be at least four, and Wilson et al (1992) and Battmer et al (1994) reported that with CIS the upper limit was seven or eight. This was consistent with the development of the Clarion SAS system with seven channels (electrodes). It used bipolar stimulation with the array that originated in the Coleman Research Laboratory at UCSF, as distinct from monopolar stimulation as undertaken by Eddington (1980, 1983). However, radial stimulation with this system could not always reach the dynamic ranges required on each electrode pair for place coding of frequency, so eight electrodes were connected longitudinally to make seven pairs (“enhanced” bipolar stimu- lation). Subsequently, monopolar stimulation was used as discussed above. Mode of Stimulation Another important design question was the mode of stimulation–bipolar, common ground, or monopolar. The stimulus mode should be the one that gives the best localization of current to distinct groups of auditory nerve fibers for the place coding of frequency. The second aim was to achieve low stimulus current thresh- old levels to increase battery life. These modes were discussed and illustrated in Chapter 5. Bipolar stimulation occurs when a potential difference is created be- tween neighboring electrodes to allow current to flow between the two. This was shown by Merzenich (1975) and Black and Clark (1977, 1978, 1980) to produce localized stimulation of the cochlear nerve fibers. It was demonstrated by Black and Clark (1977, 1978, 1980) that common ground stimulation would also lo- calize the current to separate groups of cochlear nerve fibers. Common ground stimulation occurs when there is one active electrode, with the others all con- nected electronically to form a common ground. Bipolar and common ground stimuli were shown in experimental animal studies to provide more localized stimulation than monopolar pulses. However, Busby et al (1994) found in a psy- chophysical study that monopolar stimulation between an active and distant ref- erence electrode in the cochlea allowed pitch percepts for each electrode to be scaled as well as for bipolar or common ground stimulation. Monopolar stimu- lation occurs when a potential difference is created between an active electrode and a distant ground usually outside the cochlea. The ground electrode is usually placed underneath the temporalis muscle. Studies in the CRC for Cochlear Im- plant, Speech and Hearing Research in Melbourne in 1995 showed this location was a suitable sink for the current to avoid stimulating the facial nerve or pain fibers in the vessels around the dura. The modes of stimulation were discussed in more detail in Chapter 5. 488 8. Engineering Current Levels The minimum current level for a T-level stimulus depends on the pulse width, cochlear pathology, electrode geometry, stimulus mode, and stimulus rate. For an individual patient the T level varies with electrodes and depends on the properties of the nerve membrane, as evidenced by the strength duration curve. With a shorter pulse a higher current is required to reach a T level. The current required for the MC level, is just below the minimum acceptable discomfort level (MDL). However, with a high rate, and therefore a shorter pulse width, a lower current is needed to excite the neuron as a greater electrical charge is produced. The relationship among stimulus rate, pulse width, and charge delivery is complex and was discussed in Chapters 5 and 6. Furthermore, a lower current output is needed if a speech-processing strategy uses subthreshold stimuli. With high im- pedance due to fibrous tissue and bone, high stimulus levels are required to main- tain the current required for neural excitation. The output current levels depend on the discriminable steps in loudness and their effect on speech perception. The discrimination of electrical current was studied in patients by Simmons (1966), Douek et al (1977), Eddington et al (1978), Fourcin et al (1979), Aran (1981), House and Edgerton (1982), Hochmair and Hochmair-Desoyer (1983), Dillier et al (1983), Shannon (1983), Tong et al (1988), and Nelson et al (1995). The just discriminable differences in electrical current varied from 1% to 8% of the dy- namic range. The loudness growth or dynamic range for stimulation at 200 pulses/s with the University of Melbourne/Nucleus banded electrode array in the scala tympani was found to vary from 5 to 10 dB (Clark, Tong et al 1978; Tong et al 1979). With the University of Melbourne’s first receiver-stimulator, the intensity con- trol allowed independent variation of the stimulus current of each channel in increments of 70 lA from a minimum of approximately 70 lA to a maximum of approximately of 1000 lA in 15 steps. These limitations in current steps were imposed by the integrated circuit technology at the time. The advances in speech processing requiring more frequency and intensity control coincided with im- proved integrated circuit technology. This allowed more precise control of inten- sity and rate of stimulation. The Nucleus 24 implant had a minimum stimulus current of 10 lA and a maximum of 1750 lA. There is a logarithmic relationship between the current level and the discriminable steps in loudness. Thus the current steps (I n ) can be calculated according to the following formula: (n/225) I ס 10 ן 175 , n ס 0, 255 n e.g., for n ס 0, I 0 ס 10 ן 175 0 ס 10 lA; and for n ס 255, I 255 ס 10 ן 175 1 ס 1750 lA. The current step ratio I mם1 /I m is 1.02; that is, each current step is typically 2% greater than the previous current level. As current level can be traded for pulse width to maintain equal loudness, greater flexibility was achieved by designing the Nucleus 24 as well as the Clarion S and Combi-40ם devices so that the pulse width could be varied. Typically widths of 20 to 400 ls per phase were used. The interphase gap could be varied in increments of 0.2 ls. At least 8 ls should be used, and typically 8 to 50 ls. Electronic and Communications Engineering 489 Charge Density and Charge Per Phase Charge density was shown to lead to electrolytic changes at the electrode tissue interface, and the release of gas and toxic products for short-duration (100 to 200 ls) biphasic current pulses, for a charge density of 300 lCcm מ2 geometric/phase and above. Charge density and charge per phase covaried in producing neural damage when electrodes were in contact with the cortex (McCreery et al 1988, 1990, 1994). The effects of electrical stimulation on the auditory nerve is dis- cussed in more detail in Chapter 4. Increases in the extracellular (K ם ) concentra- tion in the cortex were seen at high charge densities per phase (100 lCcm מ2 / phase or 1 lC/phase at 50 Hz) as well as high rates (Heinemann and Lux 1977; Nicholson et al 1978; Urbanics et al 1978; Stockle and Ten Bruggencate 1980; Agnew et al 1983; McCreery and Agnew 1983). The acute findings do not necessarily apply to long-term stimulation and may vary with the tissue and the distance the electrode is from the tissue. For that reason studies were carried out long-term on the experimental animal to ensure that electrical stimulation with the University of Melbourne/Nucleus banded array did not produce charge densities that could be damaging. With the Nucleus banded-electrode, animal studies (Shepherd et al 1983) showed that continuous stimulation at charge densities of 18 to 32 lCcm מ2 geometric/phase did not lead to damage of spiral ganglion cells. It was also shown that charge densities of 20 to 40 lCcm מ2 geometric/phase were within biologi- cally safe limits (Leake-Jones et al 1981a,b). Although the above in vivo study by Shepherd et al (1983) and the in vitro studies by Brummer and Turner (1977a–c) showed that charge densities below 32 lCcm מ2 geometric/phase were safe, the upper limit for safety was not estab- lished. The electrodes on the original University of Melbourne/Nucleus array had a relatively large surface area (0.44–0.66 mm 2 ). With a pulse width of up to 50 ls (normally 25 ls) and the highest current (1.75 mA) delivered through the smallest band on the Nucleus array, the maximum charge density possible was 19.9 lCcm מ2 geometric/phase. So with the worst-case scenario the charge density for the Nucleus banded array was well within the safe level. With the Nucleus perimodiolar array (Contour) the electrodes were half (see Fig. 8.49) rather than full bands, and their area varied from 0.283 to 0.306 mm 2 geometric. This would mean the density could double, but this would be counterbalanced by the lower thresholds with the electrodes closer to the spiral ganglion cells. In contrast, the surface areas of the electrodes of the Med El and Clarion devices were up to five times smaller (0.14 mm 2 ) than for the Nucleus array (Med El Combi-40 Manual; Clarion Device Description, Advanced Bionic Corp.). The Clarion system could produce up to 2.5 mA, and at its minimum pulse width of 77 ls results in a charge density of 137.5 lCcm מ2 geometric/phase. The pulse width could be increased resulting in an even greater charge density. The dimen- sions of the Clarion High Focus II electrode pads have increased to a size and shape comparable to that of the Nucleus half-band Contour array. The Med El Combi-40ם could deliver a current of 2.5 mA for pulse widths between 40 and 490 8. Engineering 640 ls, so the maximum charge density could range from 80 to 914 lCcm מ2 geometric/phase. These are well above those that have been shown to be safe (32 lCcm מ2 geometric/phase). It is therefore important to establish the safe upper levels for charge density (Clark and Lawrence 2000). Stimulus Rate The perception of pitch, due to rate of stimulation, is important for speech un- derstanding, and thus it is necessary to provide the facility to vary the rate. The upper limit on the perception of variations in the rate was shown to be about 400 to 800 pulses/s in humans (Simmons 1966), and 200 to 800 pulses/s in animal experimental studies (Clark 1969b; Clark, Kranz et al 1973). There is no solid evidence to provide stimulus rates for pitch discrimination in excess of 1500 pulses/s. Although there are limits on the perception of variations in pulse rate as dis- cussed above, the fine time structure of electric pulses delivered to the electrodes may be important. For this reason there was a need to have adequate control of stimulus timing especially between channels. The Nucleus 24, Clarion S, and Combi-40ם systems provided the CIS strat- egies at stimulus rates higher than 1000 pulses/s. This also applied to the Nucleus 24 ACE strategy. The auditory nerve fibers have an absolute refractory period of approximately 0.5 ms during which time they cannot respond to another stimulus. They also have a relative refractory period of 0.2 ms when their responsiveness is markedly reduced and a stronger stimulus is required to produce excitation. How frequently the neurons respond to each stimulus at different rates can be measured with interspike interval histograms (see Chapter 5 for further details). The data of Paolini and Clark (1997) showed that at 1800 pulses/s where the period is 0.55 ms and close to the absolute refractory period the firing pattern was a Poisson distribution, and thus the response of the unit was not related to the stimulus rate. It is also thought possible to convey temporal information by amplitude vari- ations at high rates of stimulation through altering the rate and population of nerves excited (Rubinstein et al 1999), although psychophysical studies (Vie- meister 1979; Shannon 1992; Busby, Tong et al 1993) showed that only low rates of modulation (100 to 200 Hz) could be detected. Hong et al (submitted) found that a subthreshold conditioning stimulus of 5000 Hz increased the dynamic range up to 6.7 dB on average. High rates of stimulation (1000 to 2000 pulses/s) used within the clinically acceptable intensity levels are safe (Xu et al 1997); however, the use of a high rate may damage auditory neurons at current levels and charge densities above normal clinical levels (Huang et al 1996, 1998a,b). In addition, the devices should be engineered to allow for charge recovery at the electrode/tissue interface be- tween pulses. This prevents a buildup of DC current that can damage nervous and cochlear tissue at levels greater than 2 lA (Tykocinski et al 1997). As dis- cussed above, this can be avoided by the use of capacitors with the extracochlear Electronic and Communications Engineering 491 electrode for the Nucleus 24 system, and for each electrode with the other systems. Past neurobiological safety studies demonstrated the importance of evaluating, in animal experimental studies, any significantly altered rate of stimulation as well as the electronics to be used in patients to deliver the high pulse rates. All sig- nificant changes in stimulus parameters and electrode geometry in the Nucleus system have been accompanied by animal studies. Nonsimultaneous stimulation was provided with the Nucleus 22 F0/F1/F2 and Multipeak strategies, the Nucleus 24 system for the SPEAK, ACE, and CIS strat- egies, and the Clarion S and Combi-40/40ם devices for CIS. With the Nucleus 24 the maximum overall stimulus rate was 14,400 pulses/s at 25 ls/phase, and minimum 8 ls gap and 12 ls for shorting. When this overall rate was distributed across electrodes, the maximum rate on each of 10 could be approximately 1440 pulses/s per electrode. The Combi-40ם could produce 18,180 pulses/s (I. Hochmair, personal com- munication), and thus for 12 electrodes could stimulate at up to 1515 pulses/s on each electrode. The Clarion S was reported to produce 104,000 samples/s (Clarion Device Description). The term sample rate must not be confused with biphasic pulse rate, but rather refers to the voltages used for the simultaneous analog representation of the speech signal; 91,000 samples/s were available to stimulate seven electrodes. In addition, because it took two samples to make a biphasic pulse and due to the limitations on update time, the device could produce 6500 biphasic pulses/s for distribution (Schulman et al 1996). Design Realization With the first studies undertaken by research groups it was thought there should be almost complete flexibility with the stimuli so that a patient’s percepts for different stimulus parameters, such as current levels, pulse widths, and pulse rates, could be determined. Now that more is known about the range of percepts pos- sible, a receiver-stimulator can be designed to provide the appropriate stimuli without having to be quite so flexible or require a plug and socket. Power and Data Transmission After speech is transformed into electrical signals, the signals are transmitted to electrodes in the cochlea to excite the residual auditory nerves. A high-frequency electromagnetic carrier wave was shown to be the best for transmitting power and speech data (Forster 1978; Clark, Black et al 1977), and the wave was mod- ulated by the coded speech signal. The code specified the electrode to be stimu- lated, the amplitude of the current, and the start time for each electrode within 1 ms. Since that time the Nucleus 24, Clarion S, Combi 40/40ם, and MXM receiver- stimulator devices have codes that specify the electrode to be stimulated, mode (bipolar with various electrode spacing, common ground, monopolar), rate, cur- rent amplitude, pulse width, and interpulse separation which are transmitted se- rially as pulses in a radiofrequency signal. The circuits in the receiver-stimulator 492 8. Engineering must implement the instructions from the speech processor, and the digital infor- mation is finally converted into a current to stimulate the cochlear nerve fibers. For the transcutaneous transmission of information a high-frequency modulated wave is desirable as it permits a small aerial to be used, and a large amount of speech data to be transmitted efficiently. Examples of a carrier wave modulated by varying either its amplitude or frequency are illustrated in Figure 8.3. Digital Versus Analog Circuitry In designing the receiver-stimulator, as with the speech processor, an important decision was whether to use analog or digital circuitry or a combination of both. As discussed above (see Digital Versus Analog Circuitry), analog circuits are those in which continuously varying physical parameters such as voltages can be altered or combined. With analog circuits the instantaneous amplitude of speech could be converted into a voltage proportional to the amplitude. The voltage could be transmitted indirectly to the receiver-stimulator with the induced current used to excite nerve fibers near electrodes. As a number of electrodes required stimu- lation, however, it was more attractive to use digital circuitry. It was thus more straightforward to combine (multiplex) the control information for each electrode into a single signal, and to recover this information in the stimulator. A single transmission path could then be used for a multiple-channel implant. Digitally controlled current sources deliver well-defined stimuli, and the speech processor could be precisely adjusted to suit individual patients. Finally, integrated circuit silicon chip technology has become available for low-power digital designs. Receiver-Stimulator Circuitry The circuitry of the receiver-stimulator was designed first to receive the induc- tively coupled RF signal. A format for the transmitted signal was discussed above (see Encoder and Transmitter) and illustrated in Figure 8.11. As illustrated in Figure 8.20, the signal from the inductor coil is directed to power converter and data receiver sections. These are the output current generator (OCG) and the data decoder (DDE). The output current generator produces a steady voltage that drives current through the selected electrodes. The data in the decoder are sent via a clock unit to control timing to a decoder, the stimulus output controller (SOC), to determine the instructions for the stimulus parameters. These are then fed through an electrode decoder (ED) to control the output switches (OS) that govern the mode and duration of the stimulation. This illustrated in Figure 8.21 for the Nucleus 22. As shown for the first phase of a bipolar pulse, when switch S2a and S3b are on, the voltage between the rail V dd and V ss causes current to flow in one direction from electrode E2 to a sink of current at electrode E3. V dd is the drain supply voltage, and V ss the source voltage for a FET in a CMOS circuit. The current flows through the constant current source before returning to V ss . V dd is typically ם9toם11 V in the cochlear implant. V ss is the source supply voltage, and is almost always ground or zero voltage in digital circuits. For the second phase electrode E3 is connected to V dd when switch S3a is closed and then the Electronic and Communications Engineering 493 1 22 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 OS ED TC C SOC OCG V dd I out RC Electrodes DDE F IGURE 8.20. A block diagram of the sections of the receiver-stimulator circuitry RC, receiving coil; DDE, data decoder/encoder; C, clock; SOC, stimulus output controller; ED, electrode decoder; OCG, output current generator; OS, switches; TC, telemetry controller. V E22 E3 E1 E2 E4 Power Supply V dd 123 4 V ss 22 I S1a S1b S2a S2b S3a S3b S4a S4b S22a S22b F IGURE 8.21. Switching circuitry for the Nucleus CI 22 system V dd is typically ם9toם11 volts; V ss is 0 volts; Sa is the switch between the electrode and V dd ; Sb is the switch between the electrode and V ss ; E is the electrode. [...]... cells (Simmons and Glattke 1 970 ; Schindler and Merzenich 1 974 ; Clark 1 977 ) In contrast, a free-fitting electrode was encapsulated locally with fibrous tissue, and there was little mechanical damage or degeneration of spiral ganglion cells An extracochlear multiple-electrode array was considered a possible alternative Bioengineering 509 to an intracochlear array by Banfai et al (1984b) and Banfai et al... bipolar stimulation with electrodes in the scala tympani would localize current to separate groups of neurons, without it short-circuiting along the fluid compartments of the cochlea (Merzenich 1 975 ; Black and Clark 1 977 , 1 978 , 1980) It was also demonstrated by Black and Clark (1 977 , 1 978 , 1980) that common ground stimulation would localize the current Both studies on the acute experimental animal showed... contract No 1-NS- 7- 2 342 from 19 87 to 1992, and the Coleman Laboratories for studies on pediatric auditory prosthesis implants, contract No DC- 7- 2 391 An analysis of human temporal bones was made by Dahm et al (1993) on 60 specimens from people ranging in age from 0.16 to 84 years Key findings were that growth between the sinodural angle (representing the site of the receiverstimulator placement) and the round... infraorbital foramen and the external auditory meatus) The skull is flatter at 45 degrees to this plane, and this is the preferred orientation of the device (Fig 8. 27) The receiver-stimulator packages made from ceramic had different dimensions л16 22 6.9 4 .7 2.15 50.5 75 R10 3.8 80 FIGURE 8.26 A diagram of the Nucleus CI 24 R receiver-stimulator and dimensions (reprinted with permission from Cochlear Limited)... had shown that the banded electrode array could be easily removed and another inserted if it had to be replaced (Clark, Pyman et al 19 87) It then had a maximum thickness of 6.5 mm This CI 22 receiver-stimulator replaced the clinical trial device for both children and adults The Nucleus 24 M and R receiver-stimulators (CI 24) (Fig 8.23) were designed to be smaller for use in infants and children under... lining fibrous tissue and bone formed beneath the electrode bed and would have increased the impedance, and the trauma caused loss of hair and spiral ganglion cells The tissue response did not justify the use of the extracochlear multiple-electrode especially considering the poorer speech perception performance (Banfai et al 1984a) A stiff electrode wire was used by House and Urban (1 973 ) for insertion... Melbourne/Nucleus multiple-electrode free-fitting array could be kept to a minimum, its mechanical properties were examined and compared with those of the 3M/House single-electrode array, especially in view of the trauma that had been observed in human cochleae with their single- and multiple-electrode array (Johnsson et al 1982) Single platinum wires with a diameter of 0.21 mm had been implanted (House and Edgerton... (Chouard 1 978 ) A study comparing the effects of implantation through holes in the otic capsule, and through the round window and along the scala tympani was Bioengineering 5 07 made in deafened cats by Clark (1 973 ), and a preliminary report indicated degeneration of spiral ganglion cells and peripheral processes was most severe if infection occurred An analysis of the data (Clark, Kranz et al 1 975 ) showed... elements (Rebscher et al 1981; Leake-Jones et al 1985) The Clarion array was precoiled to hug the modiolus, had eight pairs of embedded ball electrodes, and required right- and left-hand models It had a wider diameter than the Nucleus straight array The electrode was placed in an insertion tool, and a large cochleostomy was required to accommodate both insertion tool and array They were both inserted... Clark and Hallworth 1 976 ) Studies in the experimental animal by Clark (1 977 ) showed that it was effective, although there was more trauma than with passing an electrode from below upward A subsequent study by Clark, Patrick et al (1 979 ) demonstrated that if the array had graded stiffness, being flexible at the tip and stiffer toward the base, it could pass upward into the cochlea with minimal trauma, and . was shown by Merzenich (1 975 ) and Black and Clark (1 977 , 1 978 , 1980) to produce localized stimulation of the cochlear nerve fibers. It was demonstrated by Black and Clark (1 977 , 1 978 , 1980) that common. without it short-circuiting along the fluid compartments of the cochlea (Merzenich 1 975 ; Black and Clark 1 977 , 1 978 , 1980). It was also demonstrated by Black and Clark (1 977 , 1 978 , 1980) that common. (1 977 ), Eddington et al (1 978 ), Fourcin et al (1 979 ), Aran (1981), House and Edgerton (1982), Hochmair and Hochmair-Desoyer (1983), Dillier et al (1983), Shannon (1983), Tong et al (1988), and