Basic Design

The key elements in the auditory prosthesis are illustrated schematically in Figure 1.1. Acoustic signals (A) are picked up by a microphone (B) that converts them into electric signals and delivers them to an electronic signal processor (C). The processor typically divides the signal into multiple components using filters or other electronic processors and sends each component to a separate output channel as illustrated in Figure 1.2. The signal on each of these processor output channels is then converted into some sort of electrical waveform. Options for the electrical waveform include analog waveforms, which faithfully represent the temporal characteristics of the analog output from each channel of the processor; pulse trains, which can be fixed in rate or rate-modulated according to some temporal characteristic of the processor output channel; or amplitude-modulated pulse trains for which the modulation frequency is controlled by the processor channel output characteristics.

FIGURE 1.1 Schematic depiction of the components of an auditory prosthesis. A: Amplitude vs. time waveform of the acoustic speech signal. B: Microphone to capture the speech signal for delivery to the processor. C. Box containing the processor (See Figure 1.2), and in some cases the controlled-current stimulator. D: Batteries to power the processor and the stimulator. E: Transmission systems. The transcutaneous transmission system depicted on the left consists of an external antenna for transmitting signals and power across the skin and an implanted electronics package for receiving the signals and power and delivering controlled currents to the electrode array. The percutaneous connector depicted on the right can serve as an alternative for the transcutaneous transmission system. When the percutaneous connector is used, the controlled-current stimulator is housed with the processor. F: Electrode array implanted near the target neurons. In this depiction the electrode array is implanted in the scala tympani of the cochlea near the auditory nerve fibers (also see Figure 1.4). Possible alternative approaches include thin-film electrode arrays implanted directly into the auditory nerve bundle and surface or penetrating electrode arrays implanted on or within the cochlear nucleus in the auditory brainstem. G: Axons of the stimulated auditory neurons carrying nerve impulses to the central auditory system. This depiction illustrates the auditory nerve. H: The brain, including central auditory pathways, auditory cortex, and higher brain centers involved in speech recognition, speech production, and behavior. I: Detection, recognition, and discrimination of the electrical stimulation are indicated by subject responses, which can include oral and motor responses. This depiction shows a control knob that the subject can use to adjust the level of the signal during calibration sessions until it is just detectable or until it is at maximum comfortable loudness.

Examples of some stimulus waveforms are illustrated in Figure 1.3. The electrical signal must be compressed to conform to the dynamic range of electrical hearing. The signal must then be sent to the implanted electrodes.

Power for the electronics and for the stimulator is provided by batteries. Typically, a rechargable battery pack is attached to the unit containing the processor (Figure 1.1D).

The most common scheme for delivering current to the electrodes is to implant a stimulator under the skin (Figure 1.1E: left drawing) behind the external ear and to lead wires from there to the electrode array that is implanted in or near auditory

ASignal

FREQUENCY

FIGURE 1.2 Schematic representation of a generic processing strategy. A: Depiction of the amplitude vs. frequency spectrum of a speech signal, in this case a vowel. For vowels, the spectrum is characterized by peaks called formants (Fx, F2 and F3). The horizontal arrows above the formant peaks and near the fundamental frequency (F0) indicate that the frequencies of these formants change as a function of time during the speech utterance. B: Schematic representation of a processor. The processor extracts certain features from the acoustic signal and then these features are assigned to specific output "channels" based on their frequency content. In this illustration, bandpass filters are used to divide the acoustic signal into four specific frequency-limited bands and the output of each filter then comprises a channel. C: The output of each channel is sent to a specific place in or near the neural array, depending on its frequency. In a cochlear implant, channels that carry high frequency information are sent to the basal region of the cochlea and channels carrying low frequency information are sent to more apical regions, as illustrated in this example.

FREQUENCY

FIGURE 1.2 Schematic representation of a generic processing strategy. A: Depiction of the amplitude vs. frequency spectrum of a speech signal, in this case a vowel. For vowels, the spectrum is characterized by peaks called formants (Fx, F2 and F3). The horizontal arrows above the formant peaks and near the fundamental frequency (F0) indicate that the frequencies of these formants change as a function of time during the speech utterance. B: Schematic representation of a processor. The processor extracts certain features from the acoustic signal and then these features are assigned to specific output "channels" based on their frequency content. In this illustration, bandpass filters are used to divide the acoustic signal into four specific frequency-limited bands and the output of each filter then comprises a channel. C: The output of each channel is sent to a specific place in or near the neural array, depending on its frequency. In a cochlear implant, channels that carry high frequency information are sent to the basal region of the cochlea and channels carrying low frequency information are sent to more apical regions, as illustrated in this example.

neurons. Instructions and power are then transmitted to the implanted receiver-stimulator by an antenna placed on the skin overlying the implanted receiver. Typically, the antenna is aligned and held in place by a magnet. The antenna outside the skin receives its information by cable from the battery-powered signal processor worn externally. The microphone is positioned near the external ear and is hardwired to the signal processor. Recently developed implanted electronics packages include the capability of transmitting information from the implanted electrodes in the cochlea, back out across the skin, to an external receiver so that electrical activity within the inner ear that is picked up by the implanted stimulating electrodes can be monitored, as can the electrode impedances. Some earlier designs of auditory prostheses for human patients and some current designs used for experimental studies

FIGURE 1.3(A,B,C) Illustration of various patterns of electrical stimulation used in auditory prostheses. A: Acoustic amplitude (ordinate) vs. time (abscissa) waveform for the word "ears." The time axis is shown below Figure 1.3C. B: Analog electrical waveforms for the word in Figure 1.3A following band-pass filtering. The waveform amplitudes would be compressed to fall within the subject's dynamic range for electrical hearing and then sent to sites in the

FIGURE 1.3(A,B,C) Illustration of various patterns of electrical stimulation used in auditory prostheses. A: Acoustic amplitude (ordinate) vs. time (abscissa) waveform for the word "ears." The time axis is shown below Figure 1.3C. B: Analog electrical waveforms for the word in Figure 1.3A following band-pass filtering. The waveform amplitudes would be compressed to fall within the subject's dynamic range for electrical hearing and then sent to sites in the in animals use a percutaneous connector (Figure 1.1E: right hand drawing) instead of a transcutaneous transmission system. With a percutaneous connector, the stimulator can be connected directly to the internal electrodes, and direct monitoring of those electrodes is possible.

The output of the stimulator is sent to an array of electrodes that is implanted near neurons in the auditory pathway. Since most regions of the auditory pathway are tonotopically organized, similar processing strategies can be used regardless of the location of the electrode array.

The most common site for placement of the electrodes by far is in the inner ear, inside the snail-shaped cochlea (Figure 1.1F and Figure 1.4). These prostheses are commonly called cochlear implants. Typically, an array of electrodes is placed in the cochlea through the round window or through a fenestration made near the round window. Thus, barring an accidental deviation from the intended path, the electrode array will lie in the scala tympani and will occupy approximately the first 11/2 turns of the cochlea. The human cochlea is approximately 25/s turns with the scala tympani having a length of approximately 35 mm, so the electrode array lies in the basal portion, which comprises the mid- to high-frequency region of a normal cochlea. The smaller, tighter turns of the apical region of the cochlea make it more difficult to place electrodes in that region without causing severe mechanical trauma to the remaining cochlear structures.

The scala tympani of the cochlea is an appropriate location for the electrode array because auditory nerve fibers are systematically arrayed along the longitudinal axis of the spiraling cochlea (Figure 1.4). In theory, specific groups of neurons can be stimulated by passing current between two electrodes placed near those neurons. Frequently, particularly in more recent implants, one or two extra-cochlear electrodes are implanted to serve as remote return electrodes for individual intracochlear electrodes. Stimulation in this configuration (monopolar stimulation) produces a larger current field and requires lower currents than bipolar configurations where both

FIGURE 1.3(A,B,C) (continued) electrode array, with the lowest frequency channel being sent to the most apical site and the highest frequency channel being sent to the most basal site. In this example, filter passbands (in Hz) for the eight rows (top to bottom) are: 187-437, 437-687, 687-1062, 1062-1562, 1562-2312, 2312-3437, 3437-5187, and 5187-7937. Analog waveforms such as these would be used in a compressed-analog (CA) or simultaneous-analog stimulation (SAS) strategy (Advanced Bionics Corporation). C: Fixed-channel amplitude-modulated interleaved pulse trains for the word in Figure 1.3 A. Fixed-rate trains of symmetric biphasic pulses are amplitude modulated by the envelopes of filtered analog waveforms. These envelopes are extracted from filtered waveforms, such as those shown in Figure 1.3B, by half-wave rectification and low-pass filtering. The pulses are interleaved; that is, they are staggered in time so that no two channels are stimulated simultaneously. This interleaving is illustrated in Figure 1.3F. The pulse amplitudes would be compressed to fall within the subject's dynamic range for electrical hearing and then sent to sites in the electrode array, with the lowest frequency channel being sent to the most apical site and the highest frequency channel being sent to the most basal site. In this example, filter passbands are the same as those in Figure 1.3B, and pulse rate is 1200 pulses per second per channel. Amplitude-modulated pulse trains such as these are used in the continuous-interleaved-sampling (CIS) processing strategy.66

FIGURE 1.3(D,E,F) D: Acoustic amplitude versus time waveform of the word "ears" duplicated from Figure 1.3A. The time axis is shown below Figure 1.3E. E: Varied-channel amplitude-modulated interleaved pulse trains for the word in Figure 1.3D. Trains of symmetric biphasic pulses are amplitude modulated by the envelopes of the filtered analog waveforms in a subset of the number of available channels of stimulation. The subset of channels to be

FIGURE 1.3(D,E,F) D: Acoustic amplitude versus time waveform of the word "ears" duplicated from Figure 1.3A. The time axis is shown below Figure 1.3E. E: Varied-channel amplitude-modulated interleaved pulse trains for the word in Figure 1.3D. Trains of symmetric biphasic pulses are amplitude modulated by the envelopes of the filtered analog waveforms in a subset of the number of available channels of stimulation. The subset of channels to be electrodes are close together within the scala tympani. The monopolar configurations may have some additional advantages, as described later in this chapter.

The nerve fibers most likely stimulated by electrodes in the scala tympani are the auditory nerve fibers. In a normal cochlea these fibers have their cell bodies in the spiral ganglia within the cochlea and send peripheral processes along the basilar membrane, which forms the roof of the scala tympani (Figure 1.4). The central processes of these neurons exit the cochlea through the modiolus, which is located in the center of the snail-shaped cochlea. Following deafness, the peripheral processes of the neurons frequently degenerate, leaving the cell bodies and central process intact,22,23 so the site of action potential initiation in many cases is probably at the central process.

Patients who are deaf due to bilateral auditory or vestibular nerve tumors (e.g., neurofibromatosis) may have the auditory nerves removed, in which case it is possible to place an array of electrodes on the cochlear nucleus (the first brainstem nucleus in the auditory pathway) where auditory nerve normally terminates. The current generation of auditory brainstem implants uses an array of electrodes that is placed on the surface of the cochlear nucleus.24,25 However, very thin multicontact stimulating electrodes that penetrate the cochlear nucleus are currently under devel-opment.26-28 These penetrating implants are manufactured using thin-film technology, and they potentially offer several significant advantages over surface electrode arrays. First, many more contact sites can be achieved because the electrodes penetrate through the layers of the nucleus and because the design technology allows a much more efficient and dense placement of stimulation sites on the electrode array.

FIGURE 1.3(D,E,F) (continued) stimulated on a given cycle is determined by the largest peaks in the outputs of the available filters. The pulses are interleaved; that is, they are staggered in time so that no two channels are stimulated simultaneously. This interleaving is illustrated in Figure 1.3F. The pulse amplitudes would be compressed to fall within the subject's dynamic range for electrical hearing and then sent to sites in the electrode array, with the lowest frequency channel being sent to the most apical site and the highest frequency channel being sent to the most basal site. In this example, the acoustic signal is passed through a bank of 22 band-pass filters spanning frequencies from 187 Hz to 7937 Hz. Between 0 and

9 channels, with 8 channels as the default average value, are stimulated at 900 pulses per second per channel. Varied-channel amplitude-modulated pulse trains such as these are used in the Advanced Combined Encoding (ACE) and SPEAK processing strategies (Cochlear Ltd.), and n-of-m processing strategy (Med-El). F: Expanded display pulse trains from Figure 1.3E, channels 1-5 in the time-range from 350 to 375 msec. Outputs from the active channels occur in round-robin fashion in rapid succession so that no two electrodes are stimulated simultaneously. Short pulse durations must be used so that the rate of the carrier pulse train on each electrode will be sufficiently high to faithfully carry the envelope waveform. In this example the pulse duration is 0.1 msec/phase, and the pulse rate on each stimulated electrode is normally 900 pulses/sec. Figures 1.3C, E, and F were constructed using the sCILab software program written by Lai and Dillier.166 The pulse trains illustrated in Figures 1.3C, E, and F are fixed-rate pulse trains that are used as carriers for analog envelope waveforms. Another option, not illustrated here, is to modulate the rate of a pulse train based on some low-frequency feature of the signal, as is done in the early versions of the Nucleus processing strategies12,54 and in the Laura phase-locked continuous interleaved strategy.68

FIGURE 1.4 Artist's depiction of a dissected cochlea with a banded 22-electrode array inserted in the scala tympani. A wedge-shaped cut has been made in one side of the cochlea to reveal the internal structures. A: Outer bony shell of the apex of the cochlea. The apical region of the cochlea normally processes low-frequency sounds. B: Outer bony shell of the basal end of the cochlea. The basal region of the cochlea normally processes high-frequency sounds. C: Scala tympani. D: Scala vestibuli. E: Scala media, which contains the organ of Corti (F) in the normal cochlea. The organ of Corti normally contains about 3,000 inner hair cells that transduce mechanical vibrations caused by sounds into nerve impulses. In most profoundly deaf individuals, the hair cells have died. Implanted electrodes can be used in these cases to stimulate the auditory nerve. G: Parts of the spiral ganglion, which contains the cell bodies of the auditory nerve. Peripheral processes from these cell bodies project to hair cells in the organ of Corti (F) in an intact ear. Central processes from these cell bodies comprise the auditory nerve (H), which passes through the bony cone-shaped modiolus in the center of the cochlea and projects to the cochlear nucleus in the auditory brainstem. I: Multicontact electrode array inserted through a fenestration in the bone near the round window into the scala tympani and inserted for the first one and one-half turns of the cochlea. The Nucleus-22 electrode array depicted in this illustration consists of 22 platinum band electrodes supported by a silicone rubber carrier. Wires attached to the electrodes are located inside the silicone carrier and proceed to the receiver stimulator, which is implanted in the mastoid bone behind the external ear. Ten additional platinum bands (stiffening rings) at the basal end of the array are not connected to wires, but serve to improve the mechanical characteristics of the array to facilitate implantation. The cochlear structures overlying the implant are depicted as transparent so that the implant can be seen along its full length. J: Temporal bone surrounding the cochlea. This is the densest bone of the body. K: A portion of the vestibular nerve, which runs parallel to the auditory nerve and the facial nerve (hidden in this view) in the internal auditory meatus. L: The stapes, one of the middle ear ossicles that conduct sound from the ear drum to the fluids in a cochlear scalae in the normal ear. This drawing is reproduced with permission from Cochlear Corporation.

Second, the penetrating electrode arrays allow very close contact between the stimulation sites and the target neurons, resulting in lower current requirements and more specific activation of small populations of neurons. Third, the manufacturing process allows more flexibility in the design of the electrode arrays, batch fabrication of thousands of identical electrode arrays, and much lower production costs.

The potential advantages achieved with a penetrating cochlear nucleus implant could also apply to patients with an intact auditory nerve. A thin-film penetrating electrode could be placed in the auditory nerve in the modiolus. The surgical feasibility of such a modiolar implant was demonstrated very early in the history of cochlear implants by Simmons, who placed a bundle of wires into the auditory nerve. However, the amount of damage done by such a gross set of electrodes was, no doubt, considerably more than would be done by a thin-film penetrating electrode array. This modiolar implant is still in the early experimental stages.29,30

Thin-film technology can also be applied to scala tympani electrode arrays (cochlear implants). The technical requirements in this case are more difficult than with the penetrating brainstem or modiolar implants. The implant must be flexible so that it can follow the turns of the scala tympani, but it must be sufficiently stiff so that it can be inserted through the round window and pushed along the longitudinal axis of the scala tympani without buckling or sticking to the tissue surfaces. The current concept is to produce the electrode array and contact leads on a silicon substrate and then mount the thin-film array on other materials.31 This would achieve the desired stiffness, curvature and bulk to facilitate insertion of the electrode array and to hold it in position once it is inserted. Currently most commercial scala tympani implants are manufactured by hand, which is a labor-intensive process. Thin-film technology offers the advantage of greatly reduced cost, as well as the potential for inserting a larger number of electrodes in a greater variety of spatial configurations.

Was this article helpful?

0 0

Post a comment