CI Technology Explained

Question 1

Are there several common components among all CI devices?

Answer 1

Yes, even though there are differences in appearance and features

Question 2

What are the essential parts of the external sound processor?

Answer 2

Microphone (pick up acoustic signal and convert it to an electrical signal for input to the speech processor)
Digital speech processor (converts input into patterns of electrical stimulation; makes a code)
Also a cord that transmits the information from the speech processor to the coil
Coil/antenna (to transmit data to internal component of the implant via radio frequency)
Magnet (to maintain connectivity between internal and external components)
Power source (rechargeable or disposable batteries)

Question 3

What are the components of the internal implant?

Answer 3

Magnet (to maintain connectivity with external device)
Receiving/transmitting coil (receives data delivered by external transmitter via radio frequency)
Digital signal processor (receives signal from the speech processor and converts then into electrical impulses; receives code and then creates pulses)
Stimulator for electric pulse generation (decodes, analyzes, and delivers data to electrode array)
Electrode leads (deliver the electric current from the stimulator to the electrode array; ends at insertion marker)
Electrode array (stimulate the auditory nerve fibers in the cochlea)

Question 4

Are there battery packs that can be hooked up to a CI to supply more charge?

Answer 4

Yes, even available for children and babies

Question 5

Does the coil send both power and information about the auditory signal?

Answer 5

Yes, the implant itself doesn’t have a power source, so it needs to use the one for the external processor

Question 6

What is an example of how to counsel patients about CIs?

Answer 6

The device helps you hear by sending sound straight to your hearing nerve. It starts with a small microphone that sits behind your ear like the hearing aid. That microphone picks up sounds around you and sends them to a processor, which is like a tiny computer. The processor cleans up the sound and turns it into a special electric signal. That signal travels through a small coil on your head to the part of the implant under your skin. This coil stays in place because it has a small magnet that connects to another magnet inside the internal implant under your skin. From there, the internal implant sends the signal to tiny wires called electrodes that will be implanted inside your inner ear. These wires gently send electrical signals to your hearing nerve. Your brain picks up those signals and turns them into sound you can understand. That’s how you hear with a cochlear implant. It doesn’t restore normal hearing, but it can help you understand speech and sounds more clearly. It takes time and practice to get used to hearing this way—listening therapy and follow-up appointments are important for the best results.

Question 7

What are the processes that are used in the speech processor?

Answer 7

Amplification (exact amount of increase is determined by the gain of the amplifier)
Filtering (divides the signal into multiple frequency bands using a filter bank, designed to mimic the natural tonotopic organization of the human cochlea; this separation also allows each band to be processed independently)
Envelope detection (generates the envelope of the decomposed signal using a rectifier and low-pass filtering)
Compression (used to compress the incoming signal into the narrow electrical dynamic range to make it suitable for stimulation; compressing cuts out quieter sound levels considered not useful and resets louder sounds to a constant level)
Modulation (trains of biphasic pulses are generated with amplitudes proportional to the energy of each channel)

Question 8

Do current cochlear implants rely on the Fourier Analysis?

Answer 8

Yes
It is a mathematical method that decomposes complex functions or signal into simpler, periodic components
Typically represented as a sum of sine and cosine waves of different frequencies

Question 9

Is the CI speech processor essentially a fourier analyzer?

Answer 9

Yes
Incoming signal is divided into “n” different components
Signals are then sent to the electrodes distributed evenly along the basilar membrane to stimulate the appropriate nerves

Question 10

Does each channel have their own band-pass filter?

Answer 10

Yes

Question 11

What is the rectifier/rectification?

Answer 11

It takes the negative parts of the speech waves and makes them positive so they can be processed
Makes it so all of the wave properties are able to be detected

Question 12

Does compression vary between manufacturers?

Answer 12

Yes, therefore different types are associated with different speech processing strategies

Question 13

What do signal coding strategies do?

Answer 13

Decides which information goes to which electrode
Decides the priority of different pieces of information (such as pitch and timing)

Question 14

What are speech coding strategies?

Answer 14

It describes the algorithm used by the speech processor to transform the important features of the incoming acoustical signal (amplitude, frequency, and temporal cues) into an electrical code that can be delivered to the implant
All modern multi-channel systems divide sounds into different channels depending on its frequency and convert this information into electrical signals that are sent to the internal implant
It determines how the auditory signal is processed and delivered to the auditory nerve

Question 15

Why do we need a speech coding strategy?

Answer 15

CIs have limits and they cannot transmit all of the details of incoming speech
They need a strategy to decide what information to keep, what to simplify, and how to deliver it
Speech coding strategies attempt to condense the incoming signal into a form suitable for transmission as an electrical signal, while preserving the important information

Question 16

What makes speech so hard to code?

Answer 16

Speech is a complex signal that is information-rich
Different phonemes can have overlapping acoustic features
Phonemes influence each other in connected speech (co-articulation)
Talker variability
Speech is often embedded in competing noise
There are other sounds of interest present, such as music

Question 17

Where are the speech coding strategies applied to the signal?

Answer 17

During the filtering process

Question 18

What are the two main types of information in a speech signal?

Answer 18

Spectral information - which frequencies are present and at what intensity
Temporal information - how the pattern of speech changes over time (can be categorized into envelope and fine structure cues)

Question 19

What are envelope cues?

Answer 19

Slow fluctuations in amplitude
2-50 Hz

Question 20

What are fine structure cues?

Answer 20

Rapid changes in waveform
150-several thousand Hz

Question 21

What are feature extraction strategies?

Answer 21

Algorithms to extract type of spectral information, such as formants, to use this information to generate the stimulus to the electrodes
F0/F2 strategy and F0/F1/F2 strategy
Average scores on word recognition increased from 30% correct with the F0/F2 processor to 63% correct with the F0/F1/F2 processor

Question 22

What is a limitation of the feature extraction strategies?

Answer 22

Did not yield significant improvements on consonant recognition scores
Consonants are at higher frequencies, and they are harder to distinguish - and they are important for understanding speech

Question 23

What is the MPEAK or multipeak strategy?

Answer 23

Designed to improve the representation and perception of consonants by augmenting the F0/F1/F2 strategy
In addition to formant information, the new processing strategy extracted high-frequency information from the input speech signal
The important feature of this strategy that it emphasizes the high frequency range (800-4000 Hz)
MPEAK strategy yielded a mean improvement of 17% on consonant identification and a mean improvement of 28% on open-set sentence recognition

Question 24

What is the limitation of MPEAK strategy?

Answer 24

It tends to make errors in formant extraction, especially in situations where the speech signal is embedded in noise

Question 25

What are the waveform/envelope based strategies?

Answer 25

Aim to present some type of waveform (in analog or pulsatile form) derived by filtering the speech signal into different frequency bands They concentrate mainly on the timing information Examples: N of M - Cochlear Advanced Combination Encoder (ACE) Continuous Interleaved Sampling (CIS) - used by Med-El and Advanced Bionics Fine Structure Processing (FSP)

Question 26

What are N of M strategies?

Answer 26

Spectral maxima sound processor (SMSP) is an early on n of m developed in the early 1990s It uses a bank of 16 bandpass filters; in each stimulation cycle, the six electrodes with largest amplitude (maxima) are selected and stimulated at a constant rate Later refined and incorporated in the Nucleus Spectra 22 processor as SPEAK (spectral peak) - 2 extracochlear electrodes The incoming sound is divided into up to 20 frequency bands (channels) Signal is filtered into 20 bands with center frequencies ranging 250-10 KHz The SPEAK processor continuously estimates the outputs of 20 filters and selects 6 to 10 bands with the largest amplitude (maxima)

Question 27

What is ACE (advanced combined encoder)?

Answer 27

An N-of-M strategy in Cochlear Ltd. implants Currently the default coding strategy for most users Filters speech into a number of channels and then selects the highest envelope signals for each cycle of stimulation Similar to SPEAK but it uses much higher stimulation rates Most recipients perform better with ACE than with SPEAK Cochlear also offers a variation of ACE known as ACE(RE) or HighACE. ACE(RE) which uses a total stimulation rate is 32,000 pps rather than 14,400 pps

Question 28

What is CIS (continuous interleaved sampling)?

Answer 28

Strategy employed by AB and Med-El The acoustic signal is sent through a bank of bandpass filters that separates the input signal into discrete frequency bands Temporal envelope is extracted from each band, then compressed and converted into interleaved electric pulses The CIS strategy uses high-rate pulsatile stimuli to capture the fine temporal details of speech

Question 29

Does AB offer a variation of CIS?

Answer 29

Yes, known as Multiple Pulsatile Sampler (MPS)

Question 30

What is a Multiple Pulsatile Sampler (MPS)?

Answer 30

Provides partially simultaneous stimulation where two electrodes that are remotely spaced from one another are stimulated at the same time E.g. electrodes 1 and 5, then 2 and 6, 4 and 8…. The MPS strategy allows for a doubling in the stimulation rate which should theoretically improve speech recognition Channel interaction has been reported which could offset any performance benefits Limited evidence suggesting better performance with MPS compared with CIS - no longer really in use *Stimulate two electrodes at a time - but there needs to be distance as to not cause channel interaction and possibly distortion

Question 31

What did AB move onto after MPS?

Answer 31

HiResolution (HiRes) sound processing Commercially released in 2003 16 electrodes Higher max stimulation rates (up to 5156 pps) Higher cutoff frequencies for low-pass filters More sophisticated automatic gain control Available in HiResS (sequential stimulation) and HiResP (partial stimulation) Clinical trials showed HiRes led to better speech recognition and was preferred over conventional strategies

Question 32

What is HiRes Fidelity 120?

Answer 32

Released in 2006 Incorporates current steering to create up to 120 virtual channels - not actually in the array Works due to current steering - stimulate 2 electrodes in different proportions (20% of one and 80% of another) as to avoid interaction If you stimulate 50% of one and 50% of the other, the point in the middle will be stimulated Going to be stimulated closer to the electrode that is being stimulated at a higher percentage Stimulate in-between which is how they create the virtual channels Problems with battery dying fast

Question 33

Why was HiRes Fidelity 120 created?

Answer 33

There is limited space in the device Need to find a way to create more channels without creating a larger device with more space

Question 34

What is HiRes Optima?

Answer 34

The newest strategy in the HiResolution sound processing family Primary objective is to reduce the power requirements for the processor - so the battery doesn't die as fast

Question 35

What variation of CIS did Med-El develop?

Answer 35

CIS+ and High Definition CIS (HDCIS) CIS+ operates similarly to conventional CIS, but uses a Hilbert transformation in place of the conventional wave rectification and low-pass filtering Takes original signal and creates another one 90 degrees out of phase, which creates a complex signal - the envelope is taken from this complex signal (makes it easier to predict where the envelope is going to be) HDCIS allows for a broader frequency range (250–8000 Hz), with the option to extend the lower limit to 70 Hz, and a total stimulation rate of 50,704 pps Intermediate pitch percept can be created in both strategies (create more channels in-between - creates overlapped bands in small regions) Played with timing of interleaving of channels so it stimulates somewhere in-between (creating virtual channels) Possibility of 250 channels

Question 36

How do virtual channels differ between HiRes and HDCIS?

Answer 36

HiRes creates virtual channels by steering current between electrodes, resulting in a physical virtual stimulation site between electrodes (physical) HDCIS creates virtual spectral channels by using overlapping frequency filters during signal processing This results in closely timed, sequential stimulation of adjacent electrodes, which the brain interprets as an intermediate pitch between the two (perceptual)

Question 37

Does Med-El offer another novel CIS-based strategy?

Answer 37

Yes, known as Fine Structure Processing (FSP) Uses Hilbert transformation to extract spectral, envelope, and fine temporal structure information from the input signal Thought to enhance CI users’ music appreciation, and speech understanding in noise Variants include FS4 and FS4-p FS4 stimulates sequentially, whereas FS4-p allows simultaneous stimulation of the four fine structure channels (up to 1000 Hz) The longer electrode array of MED-EL arrays may further support FSP by targeting low-frequency nerve fibers in the apical cochlea

Question 38

Does the term "electrode" used inconsistently?

Answer 38

Yes, can be used to refer to individual contacts, the entire array, the lead wire connecting the array to the receiver-stimulator unit, or the combination of all three An electrode is the physical structure that injects current into the tissue

Question 39

What is the electrode array?

Answer 39

Refers to the distal part of the device that carrier the electrode contacts and is inserted into the cochlea

Question 40

What are the parts of a cochlear implant?

Answer 40

Electrode array Electronics Titanium housing Gold coil (receiver) Magnet

Question 41

What is the basic role of the electrode?

Answer 41

To do what the speech processor tells it to do Doesn't matter if the implant is really old, it will still do what the speech processor tells it to

Question 42

Does the array provide an electrical interface between the electronics package of the implant and neural tissue of the cochlea?

Answer 42

Yes, by delivering electrical stimulation to spatially distinct neural structures inside the inner ear in a tonotopically appropriate manner This is accomplished using a series of metal contacts linearly arranged on a flexible silicone carrier inserted along the cochlear spiral The carrier also encloses fine wires connecting each of the contacts to a receiver/stimulator unit located outside the inner ear

Question 43

What are some of the issues associated with electrode design?

Answer 43

Electrode placement Number of electrodes and spacing of contacts Orientation of electrodes with respect to the excitable tissue Electrode configuration

Question 44

Do intracochlear electrode contacts come in many shapes?

Answer 44

Yes Shape and design of each electrode contact can affect the shape of the resulting current fields In general, smaller electrode contacts that face the modiolus will yield more focused stimulation Examples: radial pairs (one on top, one on bottom - turns into 24 contacts instead of 12), full band electrodes, half-band electrodes

Question 45

Do electrode arrays differ across manufacturer?

Answer 45

Yes, in regard to the number of intracochlear electrode contracts contained within the array All commercially available Cochlear Nucleus electrode arrays have 22 intracochlear electrode contacts Advanced Bionics cochlear implants have 16 intracochlear electrode contacts MED-EL cochlear implants have 12 different stimulating sites *The differences in the number of intracochlear electrode contacts across manufacturers are largely based on philosophical differences of each company

Question 46

Does electrical stimulation always require a pair of electrodes?

Answer 46

Yes, one active and one return (ground)

Question 47

What is monopolar stimulation?

Answer 47

The ground stimulation is usually much larger and is preferably placed in a region where sensory tissues are absent (on the temporalis muscle or on top of the stimulator housing in typical CIs When current is applied to an electrode placed in the inner ear it flows across all tissue between it and the ground electrode, with the pattern of flow determined by the resistivity and other characteristics of the intervening tissues *or integrated in the speech processor

Question 48

What is bipolar stimulation?

Answer 48

The ground electrode is another electrode on the cochlear array (the electrodes on the array can function alternately as active or ground electrodes) Theoretically, such an arrangement would limit current flow to those tissues between the two electrodes and allow for more selective stimulation In practice, however, this type of stimulation has not proven useful and is rarely employed Only one electrode from the electrode array *The ground is just another electrode in the array

Question 49

What does modiolar mean?

Answer 49

Curved electrode array Gets the electrode contacts closer to the modiolus

Question 50

What manufacturer has the longest electrodes?

Answer 50

Med-El Every manufacturer has different lengths

Question 51

What are the things that you need to consider when trying to improve speech quality?

Answer 51

More channels increase spectral information (increasing channels from 1 to 12) Channel interaction will distort the signal, which may be more common with more channels bc they are closer together in the array