8 - Cochlear Implants

Hair cells are our sensors, on basilar membrane in cochlea (Inner Ear):

• Inner Hair Cells (IHCs): converting sound vibrations into electrical signals, sent to brain

• Outer Hair Cells (OHCs): act as biological amplifiers -> enhance sensitivity and frequency slectivity of cochlea

Outer Hair Cells damaged:

—> loss of compression, reducing dynamic range and ability to detect low-intensity sounds

Inner Hair Cells damaged:

—> ‘deaf regions’ in cochlea, certain frequency ranges that cannot be heared

81 dB or greater

most succesfull neuronal prosthesis

given to deaf and hard-of hearing

—> provides good speech understanding

—> deaf born children learn to hear and speak

• External Processor = captures sound via microphone and processes into digital signals —> sent to internal receiver through coil

• Internal Receiver/Stimulator = decodes digital signal and generates electrical pulses —> sent to electrodes implented in cochlea

• Electrode Array = delivers electrical stimulation to auditory nerve fibers, bypassing damaged parts -> freq.-specific stimulation

one of the early cochlear designs

• small dynamic range (~10 dB) required compression to prevent distortion

• most patients could not achieve open-set speech recognition

• exceptional cases only 2-4% consonant recognition

transforms sound frequencies into spatial activation patterns along the basilar membrane —> used by implants to stimulate specific areas corresponding to frequency

• Feature-Extraction Strategies: F0/F2, F0/F1/F2, MPEAK

• Waveform Strategies

  • compressed analogue

  • continous interleaved Sampling (CIS), ACE

  • PDT, FS4 with pulse timing

F0/F2 strategy: F0 encoded in rate and F2 in place and amplitude

quasi-random stimulation used for unvoiced intervals

1. processes signals in different frequency bands

2. each band corrected (envelope detection) and compressed before mapping to electrical pulses

—> better temporal resolution and frequency seperation —> improved speech and sound quality recognition

• ensures that electrical pulses dont overlap in time

• minimizing channel interference

• improving sound clarity

simulate information reduction in cochlear-implants for normal-hearing listeneres —> test effects on speech perception

1. Preprocessing (e.g AGC, noise reduction)

2. Filtering into frequency bands

3. Envelope extraction for amplitude modulation

4. Carrier modulation

   • Noise-vocoder: Noise modulated by extracted envelopes

   • Sinusoidal-vocoder: Sinusoidal carrier modulated by extracted envelopes

• Monopolar stimulation:

  current pulse on one electrode with skull electrode or/and implant body as reference electrode (‘ground’)

  + low thresholds

  - large current spread -> stimulate unintended areas

• Bipolar stimulation:

  current pulse between two (neighboring) electrodes

  + reduced current spread

  - higher thresholds -> limit benefit

• Tripolar stimulation:

  current pulse between three electrodes

  + balance between current spread and threshold

  - more complex

• Phase Duration: how electric charge interacts with tissue

  —> longer durations reduce threshold, but increase risk of damage

• Interphase Gap (IPG): allows tissue to recover between pulses

  —> improve safety and prevent overstimulation

= how auditory nerve fibers respond to electrical stimulation

1. input Stimulation current (I) flows in Capacitor -> charge

   voltage across Capacitor increases over time

2. Resistor allows some of charge to ‘leak’ out

   -> represents ‘leaky’ nature of membrane

3. neuron fires an action potential (spike) (after voltage of Capacitor reached predefined threshold)

4. after firing -> Capacitor discharged to reset voltage

5. process from beginning

reflect biological variability in nerve response

-> model outputs more realistic spike patterns

• low stimulation rates (250 pulses/s):

  enough time for neuron to recover btw pulses -> respond to every pulse effectively

• higher rates (1000+ pulses/s):

  neurons recovery time is insuffiecient -> fewer spikes observed -> neurons less sensitive

adaption results in reduced spike rates as stimulus continues

—> protects the auditory nerve from overstimulation

ability of auditory system to distinguish changes in pulse rates

Pulse Rate Discrimination Limens (PRDL)

measures smallest detectable difference in pulse rate

higher values repesent greater change in rate (y-axis)

above 300 pps, thresholds increase for most subjects

important for: Speech, Music and Pitch Perception

position of electrodes affects pitch perception due to place-pitch coding (specific electrodes stimulate nerve fibers corresponding do distinct frequencies)

Challenges:

• limited spectral resolution - multiple electrodes, same pitch

• non-tonotopic pitches

• variability between implants

Problem: limited number of electrodes in implants -> limit number of frequencies/pitches

Solution:

Use current steering: gain a few “extra pitches” in-between electrodes

• current is split in 2 components

  • fraction a directed to one electrode

  • fraction a -1 to neighboring electrode

• ‘center of gravity’ is shifted between electrodes, creating intermediate pitch

—> increases number of pitches —> improving auditory experience

• Threshold-level (T-level)

  minimum level of electric current, that user can hear

  ~ few hundrer microamps

• Comfortable-level (C-level)

  maximum level of electric current, comfortable for user without causing pain

  several hunderd microamps

• Electric Dynamic Range (EDR)

  difference btw T-level and C-level

  7-10 dB

Mapping Process:

acoustic sound levels are compressed and mapped to narrow electric dynamic range using Level-dependent Compression

Automatic Gain Control (AGC) dynamically adjusts input levels

summed electrical response of auditory nerve fibers to electrical stimulation from cochlear implant electrodes

Clinically:

Widespread clinical use to estimate T- and C-levels (children!)

no, with single CI

discrimination of side of sound origin (left/right) possible

Yes, with bilateral CIs

very good localization ability possible

ITDs: electrical stimulation limits accurate processing of temporal cues

ILDs: based on differences of sound intensity between ears, remain intact and more easily perceived

normal-hearing: ITDs for low-frequency sounds and ILDs for high-frequency sounds

• Good for speech:

  • high pulse rate

  • compression at each ear set for optimal loudness

• Good for localization:

  • low pulse rates (< 300 pps) to encode temporal information

  • identical compression on both ears to maintain interaural level differences

when head blocks sound from one side, improving signal-to-noise ratio for ear on other side