TECHNOLOGY
by Graeme M. Clark
Building the Bionic Ear
for replicating the encoding of sound, the process by which electrical signals that correspond to sound are created in the nerves and brain cells; • an implant in the inner ear would cause the loss of the very nerves it was intended to stimulate; • speech was too complex for electrical stimulation to present it to the nervous system in a way that it could be understood; • too few residual hearing nerves would remain in the inner ear as a result of deafness to transm it essential speech information; and • children born deaf would never develop the appropriate neural connections in response to electrical s ti mul ation t hat they nee d to acquire adequate hearing. To resolve the first issue, my colleagues and I did animal studies, first at the University of Sydney and then, starting in 1970, at the Univ e rsity of Melbourne. We initially used a single-electrode cochlear implant to determine how well electrical stimulation could reproduce the normal ear’s encoding of sound. These studies showed a significant difference in timing between the brain’s responses to sound and to electrical stimuli, and that animals could not distinguish changes in electrical frequency. These findings me ant th at a s i ngl e-el ect rode Figure 1. The younger the child at the time of cochlear implant would not allow the implant operation, the more hearing perforthe encoding of speech frequenmance improves. Today, children as young as 9 cies up to 3,000 Hz. The speechfrequency range in a child is 125 to to 12 months receive cochlear implants. 4,000 Hz, a fact that led to our deaf, however, have lost their hair cells, so decision in 1973 to begin developing a even amplifying sound with a hearing aid multiple-electrode cochlear implant. Hearin g re quires place c oding, t h e fails to transmit information that the brain process by which electrical current is directcan interpret as sound. Those of us who envisioned building a ed appropriately to the different groups of bionic ear had to confront five fundamental hearing nerve fibers that convey various and quite reasonable objections from skep- pitches. Physiological and behavioral studies demonstrated that electrical stimulation tics, who argued that: • the inner ear was too complex to be could only partially encode sound. Because replaced by a small number of electrodes it is not as efficient as the normal ear, electrihen I began my cochlear-implant research in 1967, one could do nothing to help profoundly deaf people hear. In normal hearing, hair cells in the cochlea, the snail-shaped inner ear, transduce sound vibrations into electrical signals. These signals produce patterns of electrical responses in the auditory pathways that convey the frequency and intensity of the sound to the brain. The profoundly
FEBRUARY 2000 © American Institute of Physics
cal stimulation can transmit some but cannot transmit all of a sound’s frequency and intensity information.
Mutiple electrode
This finding reinforced the need to develop a multiple-electrode cochlear implant. Multiple-electrode stimulation is necessar y for electrodes to correctly stimulate the various hearing nerves that present different pitches to the brain. This ensures that separate channels of information go to the brain. To achieve optimal place coding, we began research to determine where to place electrodes in the inner ear so that bipolar and common-ground currents could be localized to discrete groups of auditory nerve fibers. Our research showed that the lowe r compartment of the inner ear was the correct location. But would inserting a bundle of electrodes lead to the loss of the very nerves we hoped to stimulate? Further animal studies showed that it would not. Barring a serious inner ear infection, and provided no trauma had damaged the inner ear’s vibrating membrane or fractured the bone supporting the membrane, we found that the auditory nerve was preserved in the deaf, so that hearing loss could be reversed by electrical stimulation. We also discovered that injury to the inner ear by an implant, which can cause deterioration and loss of the auditory nerve, was minimized if the electrode bundle had the right mechanical properties. It needed to be smooth, tapered, flexible at the tip, and stiffer toward the proximal end. In addition, we found that no long-term damage occurred to the inner ear or auditory nerve if we used two-phase pulses that were balanced to prevent a buildup of dc current, and if the charge density was kept low.
Receiver–stimulator
To establish that multiple-electrode stimulation could indeed transmit sufficient information for understanding speech required developing a fully implantable receiver–stimulator. And because speech perception is an
12
The Industrial Physicist
2
Auditory cortex
5
10
20
5 kHz 2 kHz
Transmitter coil
Banded electrode array in the cochlea
When sensory hair cells on a vibrating
10 kHz 20 kHz
Microphone Behind-the-ear speech processor
membrane move to and fro, sound vibrations are converted into electrical currents that excite nerve fibers leading to the brain.
Sensory hair cells
Vibrating membrane Auditory nerve
Receiver– stimulator Wearable speech processor
Cochlea
Electrode array
Figure 2. Elements of the Nucleus 24 multiple-electrode cochlear prosthesis. Speech is detected by the microphone, modulated by the wearable speech processor, and transmitted to the receiver–stimulator, which sends electrical signals to the electrode array in the cochlea. Nearby auditory nerve fibers respond to stimulation from corresponding electrodes and convey the message to the auditory cortex that specific types of sound are being heard.
especially human skill, the device could not be evaluated in experimental animals. A prototype receiver–stimulator was developed beginning in 1974 with funding from an Australian telethon, trusts, and foundations. It was first implanted in a profoundly deaf adult on Aug. 1, 1978. The device was placed in the mastoid bone behind the ear, and the
electrodes were threaded into the inner ear to lie near, but not in direct contact with, the nerves relaying speech frequency to the brain (Figure 2). Our research received further support when the company that became Cochlear Ltd. (Lane Cove, NSW, Australia) joined us as an industrial partner. Psychophysical studies began in 1978
with our first implant patient. Each patient who received an implant was asked to describe what he or she perceived. It was discovered that patients could distinguish different frequencies or hear different pitches for a rate of stimulation up to only 300 pulses/s. This rate was much less than the 3,000 pulses/s needed for under-
13 The Industrial Physicist
Te c h n o l o g y
standing speech. On the other hand, when the different electrodes were stimulated, patients experienced a change in timbre from sharp to dull and from high to low, but they could not perceive pitch. Our first implantable receiver–stimulator filtered out the different speech frequencies and presented them to various electrodes, which then stimulated the corresponding frequency areas of the auditory nerve and speech-processing areas of the brain. This approach was not successful because the electrical stimulation to the auditory nerve was simultaneous. Electrical currents that are presented simultaneously to two or more electrodes can produce a combined electrical field that is unpredictable in its effects on nerves. Electrical voltages produce electrical field interactions that cause unpredictable variations in loudness. We then developed an alter n a t i v e speech-processing strategy that enabled patients to understand conv e rs a t i o n a l speech, either by electrical stimulation alone or in combination with lip reading. This new strategy took advantage of the fact that when different groups of nerves are stimulated nonsimultaneously, electrical fields do not ov e rlap in an unpredictable way. The clue to this speech-processin g system came when the firs t patient to receive an implant report e d vowel sounds when each electrode was stimulated separately. These vowels corresponded to those perceived in normalhearing people when similar areas of the inner ear were excited by formants, which are vocal-tract resonances important to understanding speech. As a result of this research, we developed a device that filtered out the frequencies in the second formant—the most important for understanding speech—and presented the voltage from this filter to the appropriate electrode in the inner ear. The sound pressures at this frequency were converted to various current levels, which patients perceived as different degrees of loudness. A device incorporating this speech-processing strategy was tested at
14
The Industrial Physicist
Te c h n o l o g y
centers in the United States and Europe beginning in 1983 and 1984, respectively. In 1985, it became the first multiple-electrode cochlear implant to be approved by the U.S. Food and Drug Administration (FDA) for the treatment of deafness. Our research then focused on the first and third formants, which are important in th e l owe r an d h igh er f r equenci es of speech. Our most recent implantable receiver–stimulator, the Nucleus 24 syst e m , selects the six to eight frequency bands with the greatest energy from a bank of 20 band-pass filters, and presents that information for sound encoding. However, the sites of stimulation within the cochlea may lie close together, which can lead to electrical interactions and unpredictable variations in loudness. Using a constant rate of stimulation to all electrodes has minimized this. Our studies indicate that most people using Cochlear Ltd.’s Nucleus 24 implant can communicate effectively over the telephone.
Children
The final major objection was that children born deaf might not be able to develop the right neural connections for speech understanding through electrical stimulation. It was of critical importance,therefore, to learn whether speechperception performance in children who had hearing before becoming deaf was comparable to that of children born without hearing. We wanted to know, in particular, whether exposure to sound during a critical period when brain connections are still being established was necessary for adequate perception, or whether appropriate connections could develop in the absence of any exposure to sound. The speech-perception abilities of two groups of children—one born deaf and another that lost hearing after exposure to sound—were tested and found to be similar. Consequently, prior exposure to sound was not necessary for good speech perception. In 1990, the FDA approved the marketing of a Cochlear Ltd. implant for children aged 2 and older. This was the first time any World Health Organization regulatory
body had approved a cochlear implant as safe and effective for children. Finally, we examined the importance of electrically stimulating the auditory nerve at a young age,when the brain is still forming connections important to hearing. The results showed that there is considerable variability in responses, but that the younger the age of the child at the time of the implant operation, the more hearing performance improves. However, special safety issues must be considered for implantation in children younger than 2 years. These include the effects of head growth; middle ear infections, which are common at this age; and electrical stimulation of a maturing nervous system. We conducted research supported by the U.S. National Institutes of Health that showed no cause for concern in operating on this group of children. We also learned how to configure the electrode leading from the receiver–stimulator so that the normal growth of the head did not pull on the electrode in the inner ear. Pulling on the electrode would lead to poor performance and prevent the insertion of another electrode later in life, if necessary. Today, children as young as 9 to 12 months receive cochlear implants (Figure 1). Cochlear Ltd.’s implants, which rely on the results of speech and biological research undertaken by the University of Melbourne and the nonprofit Bionic Ear Institute, have been implanted in approximately 25,000 people in 50 countries, at least half of them children. Further challenges lie ahead: to improve the fidelity of sound, increase the numbers of people able to benefit, improve performance in the presence of background noise, and make implants ultimately invisible with a totally implantable system that includes a rechargeable battery.
Graeme M. Clark is professor of otolaryngology at the University of Melbourne and the founder and director of the Bionic Ear Institute in East Melbourne, Australia (g.clark@medoto.unimelb.edu.au).
15
The Industrial Physicist