Zoom out Search Issue

ManualsBrandsContents Manualsaudio & home theatreZoom in

IEEE SIGNAL PROCESSING MAGAZINE [68] MARCH 2015

language skills at a young age. In many countries, a single CI is

reimbursed by health insurance organizations, and in some coun-

tries, the cost of a second CI is also reimbursed, primarily for chil-

dren. About 80% of normally developing, severely

hearing-impaired children with a CI eventually participate in the

mainstream educational system.

Apart from the technological and surgical progress that has

made cochlear implantation the success it is today, the preformed

cochlear duct and the ease of surgical access via the middle ear

have played a role in its proliferation and progress. How CIs work

has been described before in several articles; e.g., in [2]–[4]. This

article focuses on a review of stimulation strategies. These are the

techniques that convert sound signals picked up by a microphone

into patterns of electric stimuli that activate the auditory nerve.

The remainder of this section provides a short overview on how

we hear and how a CI works.

In the normal auditory system, sound is captured and trans-

mitted by the outer ear, predominantly the pinna (external ear)

and ear canal, and then transformed in the middle ear (via the

ossicles—small bones that have a mechanical impedance-

matching function) to movement of the fluids and membranes

in the cochlea, or inner ear. The cochlea has a spiral structure

typically about 10 mm wide and 5 mm high. Within the cochlea,

there are numerous transducer structures—the inner and outer

hair cells—which have stereocilia that are deflected in response

to incoming sound waves. In a healthy ear, movement of the

stereocilia of inner hair cells leads to streams of action poten-

tials in the auditory nerve fibers. This electrical activity has pat-

terns with temporal and tonotopic characteristics that

ultimately enable identification and interpretation of sounds,

including music, speech, and language, at higher neural levels

[5]. Temporal information about sound signals is carried

through the precise timing of action potentials both within and

between nerve fibers, whereas spectral information is repre-

sented mainly in the spatial distribution of activity across the

neural population; the latter is referred to as the tonotopic orga-

nization of auditory nerve.

The most common cause of deafness is damage to or loss of the

stereocilia and hair cells, resulting from infections, trauma, expo-

sure to high levels of noise, side effects of certain drugs, and a

range of physiological disorders. Hearing impairment may be

acquired by adults who previously had normal hearing, or it may

be present at birth. In many cases, the degree of hearing loss

becomes progressively worse over time. When the hair cells are

absent or extensively damaged, the transduction of the acousti-

cally induced motion in the cochlea to neural action potentials is

disrupted. If the resulting hearing loss is severe, the amplification

that can be provided by acoustic hearing aids may be insufficient

to restore satisfactory perception of sounds.

A CI bypasses the deficient transducer structures and produces

action potentials at the auditory nerve sites (or the residual neu-

rons, depending on the degree and type of pathology) using direct

electrical stimulation. Most of today’s CI systems have an external

and an internal part. The external part consists of a behind-the-ear

(BTE) device connected to an external transmission coil, which

provides a radio-frequency (RF) link to a matching coil in the

internal part, the implant. The implant consists of a miniature

enclosure containing electronics connected to a number of elec-

trodes. There are one or more reference electrodes on the enclo-

sure or on a separate lead, and there is an array of multiple

intracochlear electrodes, between 12 and 22 depending on the

manufacturer and implant type. The stimulation currents flow

between selected electrodes to activate the neural structures near

the electrode-neuron interface. The electrode array is surgically

inserted into the cochlea. Implantation of the complete internal

system takes approximately three hours.

As illustrated in Figure 1, sound is captured in the external BTE

device by a microphone system (one or more microphones). Pre-

processing is applied, for example, to optimize the input dynamic

range relative to input signal levels and to adjust the spectrum

shape using a pre-emphasis filter. In some systems, there is also

fixed or adaptive beamforming or other types of noise-reduction

processing that typically exploit the differences between signals

obtained from several microphones to enhance desired sounds

while suppressing competing noise. The stimulation “strategy”

refers to the transformation of the input sound signal into a pattern

of electrical pulses. Digital specifications of the required stimula-

tion patterns produced by the stimulation strategy are coded in the

transcutaneous RF transmission. The RF signal also provides

power to the internal part. The specifications of the stimulation are

decoded from the RF signal. The electronics of the implant include

one or more current source(s) to deliver the electrical stimulation

pattern to the electrode channels. A channel is defined as a set of

two or more electrodes with currents flowing between them. The

term monopolar stimulation is used to describe current passing

between an intracochlear electrode and a remote reference elec-

trode, whereas bipolar refers to stimulation current passing

between two intracochlear electrodes. The implant also has mea-

surement amplifiers on-chip for the recording of evoked neural

activity from nonstimulating electrodes via outward telemetry.

A few weeks after implantation and at regular intervals thereafter,

stimulation levels are adjusted (“fitted”) to the individual patient. In

each fitting session, a patient-specific “map” is set up containing all

stimulation parameters. For each channel, minimal levels of stimula-

tion (min) and levels of maximal comfortable loudness (max) are

determined. In some cases, the shape of the growth function

between min and max that converts the input acoustic levels to elec-

tric stimulation levels is also determined. During a fitting session,

Microphone

Preprocessing

Stimulation

Strategy

Transmission

Decoder

Pulse

Generation

Electrode

Array

[FIG1] A block diagram of a complete CI system.

THE WORLD’S NEWSSTAND

THE WORLD’S NEWSSTAND