The next 100 years

1
The next 100 years

• Harvey Dillon
• NAL,
• Cooperative Research Centre for Cochlear Implants
  and Hearing Aids
• Australia

2
This talk can be obtained from www.NAL.gov.au

• (After June 16)

3
Areas of current and future improvement

• Improved directional performance
• Wireless linked binaural processing
• Improved automatic sensing
• Resynthesis hearing aids
• Patient involvement in fine-tuning
• Trainable hearing aids
• Overcoming auditory deprivation
• Electrophysiological fine-tuning (infants)
• Implanted devices
• Bimodal devices
• Integration with communications devices
• DSP capabilities
• Receiver location
• Batteries
• Hair cell regrowth

Dillon, NAL
4
Improved directional performance
Dillon, NAL
5
Improved directional performance
90
Super
60
120
150
30
-10
-20
-30
180
0
Hyper
330
210
300
240
270
Figure-8
Dillon, NAL
6
Adaptive directional microphone
Front
?
-

?
T
Output
Figure 7.4 A simple adaptive directional
microphone with steerable nulls.
Source Dillon (2001) Hearing Aids
Dillon, NAL
7
Multi-band adaptive directional microphone
Front
?
-

?
Filter
Filter
T
Dillon, NAL
/ binaural
8
Binaural signal processing hearing aids

• Jorge Mejia, Michael Fisher

Dillon, NAL
9
Binaural processing hearing aids

• Different signal to noise ratios and phase
  relationships on different sides of the head
• People with normal hearing use these differences
  to suppress noise
• People with hearing loss need better SNR, but may
  have reduced ability to use binaural cues

Dillon, NAL
10
SNR needed increases with hearing loss
Killion
Dillon, NAL
11
Binaural processing hearing aids

• Different signal to noise ratios and phase
  relationships on different sides of the head
• People with normal hearing use these differences
  to suppress noise
• People with hearing loss need better SNR, but may
  have reduced ability to use binaural cues
• Hearing aids could combine outputs from
  microphones on both sides of head
• Noise reduction increases with number of
  separated microphones

Dillon, NAL
12
Separating signals and noise
Dillon, NAL
13
Physical arrangement simulated
Dillon, NAL
14
Blind source separation matrix(Inverse based on
initial 100 ms of signal)
Jorge Mejia
Dillon, NAL
15
Need wireless-linked hearing aids

• Already available in BTE and ITE form but only
  used as CROS and BICROS aids
• Functions as a weakly directional microphone array

Transmission distance 4.5 - 8 (12 20
cm) Transmission frequency of 1.8 MHz reduces
interference Transmission power 0.8 mW
Dillon, NAL
/ environment sensing
16
Improved environment sensing

• Gitte Keidser

Dillon, NAL
17
Environment sensing hearing aids
?
Dillon, NAL
18
Improved environment sensing

• Use information about the spatiality of sound, as
  well as its spectral and temporal properties
• Auditory Scene Analysis

Dillon, NAL
19
Environment sensing hearing aids

• Automatic variation of response between programs
• Microphone directionality
• Low frequency gain
• Degree of noise suppression
• Wind noise reduction (low cut, omni)
• Automatic telecoil selection

Dillon, NAL
20
Environment sensing hearing aids

• Implications for fitting now have to fit, or
  auto-fit, multiple programs!
• Patient involvement in fitting
• simulated listening situations
• computer-based fine-tuning
• . Clinical time!

Dillon, NAL
/ patient fine tuning
21
Patient fine-tuning

• Wouter Dreschler, Gitte Keidser, Liz Convery

Dillon, NAL
22
Patient fine-tuning

• Two loudspeakers and a TV-screen
• Binaural presentation
• With or without hearing aids

Dillon, NAL
23
Controller types
Dillon, NAL
24
Dillon, NAL
25
Preferred gain relative to NAL-RP prescription
Reasonable similarities between the overall
results for different controllers
Consistent differences across listening
environments
Dillon, NAL
26
The starting baseline affects the apparent
preference!
Flatter starting response
Steeper starting response
Dillon, NAL

• The effect occurs almost equally for all videos

27
The starting baseline affects the apparent
preference!
Dillon, NAL

• The effect occurs almost equally for all
  controllers

28
But not all preferred results are equally good!
Dillon, NAL
/ trainable aid
29
A Trainable Hearing Aid Justin Zakis,
Harvey Dillon, Gitte Keidser, Liz Convery, Hugh
McDermott
Dillon, NAL
30
Existing hearing aids

• Fitted in a quiet clinic, away from real-life
  environments encountered in everyday life.
• Fitted with a prescription based on averages
• No customization possible in real-life
  environments
• User always returns to clinic if aid
  unsatisfactory
• User returns hearing aid if too unsatisfactory

Dillon, NAL
31
Aid user adjusts settings...
Trainable Aid
Dillon, NAL
32
Aid user adjusts settings...
Trainable Aid
Dillon, NAL
33
...then presses voting button
Trainable Aid
Dillon, NAL
34
Process repeats for other sounds
Trainable Aid
Dillon, NAL
35
After training, preferred settingsare
automatically applied...
Trainable Aid
Dillon, NAL
36
After training, preferred settingsare
automatically applied...
Trainable Aid
Dillon, NAL
37
After training, preferred settingsare
automatically applied...
Trainable Aid
Dillon, NAL
38
After training, preferred settingsare
automatically applied...
Trainable Aid
Dillon, NAL
39
Prototype trainable aid

• Stereo Hearing Aid Research Processor (SHARP)
• Fully-programmable digital hearing aid

Power/mode switch
Gain control
Voting button
Dillon, NAL
40
First and Second Comparison Trial Results
p lt 0.05 p lt 0.01 Only S201 had a
statistically significant difference in scores
between comparison trials (p 0.0176)
Dillon, NAL
41
Advantages for Clients

• Can customise the sound of aid in environments
  they encounter, which cannot be reproduced in a
  clinic.
• More parameters can be trained, simultaneously
  or sequentially, than can be practically set in
  the clinic
• Can retrain the aid at any time if preferences
  or environments change
• Fewer return visits to the clinic
• Better hearing in more environments
• Greater confidence to use hearing aid more often
• Sense of ownership and personalisation over
  fitting process and the aid

Dillon, NAL
/ resynthesis
42
Resynthesis hearing aids
Dillon, NAL
43
Resynthesis hearing aids

• Recognize speech -gt Synthesize clear speech
• Problems with recognition
• Accuracy
• Time synchronization
• Sloppy speakers!
• Maybe a what did he say button
• Maybe even a translation.

Dillon, NAL
/ auditory deprivation
44
Overcoming auditory deprivation
Dillon, NAL
45
The end of the road
Dillon, NAL
46
Corpus callosum
Cortex Complex detection
Perception
Medial geniculate nucleus
Thalamus Auditory and visual map integrated,
relayed to cortex
SC Visual spatial map
IC Form full spatial map, Parallel processing
paths join, History dependent
Lateral lemnisci
VNLL Fed by contralateral CN
Sorting, comparing and categorizing
MSO Detect interaural time LSO Detect
interaural level
AVCN Frequency analysis, PVCN Timing well
preserved DCN Inhibitory circuits, pinna cue
detection? Parallel processing Needs to be fed to
develop maintain
Dillon, NAL
47
But remember..

• One side only shown
• Parallel processing
• Afferent (up) and efferent (down) connections

Dillon, NAL
48
Effect of past conductive loss on MLD
Aithal, Aithal Yonovitz (2004)
1
Dillon, NAL
49
Auditory processing disorders

• Inefficient processing of sounds (e.g. to
  understand speech when spatially diverse talkers
  present)
• With or without presence of loss of sensitivity
  to pure tones
• Impacts on education
• Training therapies available - evidence positive
  but controversial
• Wireless systems effective
• class-room sound-field systems
• personal sound-field systems
• ear-level wireless systems

Dillon, NAL
/ electrophysiology
50
Electrophysiological fine-tuning of hearing aids

• Suzanne Purdy, Richard Katsch, Mridula Sharma,
  Lydia Storey, Teresa Ching, Wendy Pearce, Mandy
  Hill, Maryanne Golding

Dillon, NAL
51
Motivation for investigating aided auditory
evoked potentials

• Objective hearing aid validation techniques
  needed for young infants and difficult-to-test
  children

Dillon, NAL
52
Medial geniculate nucleus
Steady State Response
Pa (25 ms) is thalamus and A1. Pb (50 ms) is same
as P1 in cortical
Wave V is from IC and possibly LL
Dillon, NAL
53
Initial Research Question

• Do different stimuli lead to different cortical
  responses?

Dillon, NAL
54
Grand average infant cortical responses recorded
at Cz
Dillon, NAL
55
Multivariate Analysis of Variance
Voltage
Time

• Divide each record into 50 ms time bins
• Average data points within each time bin
• Use these averages as variables in MANOVA
  analysis
• MANOVA finds the combination of variables that
  best distinguishes two or more stimuli
• Result is probability of two stimuli coming from
  different distributions

Dillon, NAL
56
Differentiation of responses for individuals
Dillon, NAL
57
Additional research questions

• Aided CAEP in hearing-impaired infants and
  children
• recordable in infants and children with
  moderate-profound loss?
• show appropriate changes for different stimuli,
  as hearing aid settings are altered?

Dillon, NAL
58
of ears tested with aided cortical response to
65 dB SPL speech stimulus
13 18 8 8 10 7 13
20 7
Number per category
Dillon, NAL
59
Enhanced cortical P1 response to /gae/ with
increased hearing aid gain where initially there
was no response when hearing aid was set
conservatively relative to the measured
tone-burst ABR thresholds
Dillon, NAL
60
Cortical auditory evoked responses

• can be reliably recorded in young infants
• are sensitive to changes in stimulus
  characteristics within individual infants
• are recordable in infants and children with
  moderate-profound hearing loss
• Usually sensitive to changes in hearing aid
  characteristics

Dillon, NAL
61
Main remaining goal

• Is the hearing aid adjustment that gives the best
  set of cortical responses the same adjustment
  that best helps the child in real life?

Dillon, NAL
/ implanted devices
62
Implanted devices
Dillon, NAL
63
Bone-anchored hearing aid
Figure 16.6 Bone-anchored hearing aid, showing
its attachment through the skin to the bone.
Amended by permission from Entific Medical
Systems.
Dillon, NAL
Source Dillon (2001) Hearing Aids
64
Through-the-ear aid
Dillon, NAL
65
Implantable hearing aids
Dillon, NAL
66
Implantable hearing aids
Output vibrator
Microphone
Magnet coil
Magnet
Coil
Dillon, NAL
67
Fully implantable aids and the future

• Current limitations
• Quiet implanted microphone
• Battery renewal
• Advantages
• No occlusion
• No feedback?
• Convenience
• Other?

Dillon, NAL
68
Choose your style

• BTE Behind-the-ear
• ITE In-the-ear
• ITC In-the-canal
• CIC Completely in-the-canal
• PAC Post-auricular canal
• TTE Through-the-ear
• PI Partly implanted
• FI Fully implanted

Dillon, NAL
Jump to communication devices
/ bimodal devices
69
Bimodal devices

• Teresa Ching, Emma van Wanrooy, Mandy Hill

Dillon, NAL
70
Hybrid implants and hearing aids

• For people with severe hearing loss
• Hearing aids - most effective for low-frequency
  cues and pitch / timing
• Cochlear implants - most effective for high
  frequency cues and spectral shape
• Implanting does not destroy potential for
  conventional acoustic aiding

Dillon, NAL
71
Combining hearing aids and implants

• 1. Hearing aids and implants in opposite ears
• Now recommended practice
• Fine-tuning of hearing aid beneficial
• 2. Hearing aid and implant in same ear
• Separate devices
• Short electrode used so far
• 3. Combined hearing aid and implant

Dillon, NAL
/ communication devices
72
Integration with communication devices
Dillon, NAL
73
Things in your ears

• Now mobile phones, personal digital assistants,
  solid-state music or entertainment devices, and
  of course hearing aids.
• In the future personal navigation aids,
  ultra-localised communication systems (e.g. in
  museums), and local area (human communication)
  wireless networks, all voice controlled, of
  course

Dillon, NAL
74
Integration with communication devices

• Headset device as a hearing aid, with Bluetooth
  communication to
• mobile phone
• personal digital assistant

Dillon, NAL
75
Integration with communication devices

• Telephone with bone-
• conduction output
• Communication device look-alike
• hearing aids
• Mobile phones with multi-band compression
  processing, programmable for individual
• Headset device as a hearing aid, with
  Bluetooth communication to
• mobile phone
• personal digital assistant

Dillon, NAL
/ shrinking technology
76
The shrinking hearing aid
Dillon, NAL
77
1. Shrinking amplifiers
Dillon, NAL
78
Early vacuum-tube hearing aid
Radioear (1925)
Dillon, NAL
79
Rapid size reduction
Dillon, NAL
80
Ca. 1950 - 1960, Blaupunkt Advertisment
Dillon, NAL
From Stefan Launer
81
Transistor era
Dillon, NAL
82
Power Dissipation Trends in DSP
Source Texas Instruments
Dillon, NAL
83
SIA Roadmap SummaryFeature Size
Trouble!
Dillon, NAL
84
SIA Roadmap SummaryFeature Size
Dillon, NAL
85
SIA Roadmap SummaryInternal Clock (high
performance)
Dillon, NAL
86
SIA Roadmap SummaryFeature Size
Dillon, NAL
87
SIA Roadmap SummaryLogic Transistors
Dillon, NAL
88
SIA Roadmap SummaryLogic Transistors
Dillon, NAL
www.aeiveos.com/bradbury/petaflops/siardmap.html
89
SIA Roadmap SummaryDRAM Size
Dillon, NAL
90
SIA Roadmap SummaryDRAM Size
Dillon, NAL
91
SIA Roadmap SummaryVoltage
Dillon, NAL
92
The march of DSP Implications

• Hearing aid chips all become general purpose DSP.
• Companies that sell hearing aid hardware, and
  companies that sell signal processing software.
• always need an upgrade!

Dillon, NAL
/ shrinking packages
Jump to prescriptions
Jump to hair cell
93
2. Shrinking packages
Dillon, NAL
94
Occlusion effect generation
Dillon, NAL
95
Occlusion effect generation and low receiver
current
A
Need soft material!
Dillon, NAL
96
3. Shrinking batteries
Dillon, NAL
97
Implications for batteries

• Small amplifier current
• Small receiver current
• Small battery
• Rechargeable
• Auto switch-on and switch-off

Dillon, NAL
98

• Waterproof hearing aids, please

Dillon, NAL
99
Laser construction of earmolds and shells

• 1. Impression scanned at factory (now)
• Computer-assisted construction (reduced cost)
• Minimum size
• Retention
• 2. Impression scanned at hearing centre
• As above, plus
• Reduced turn-around time
• 3. Ear canal scanned
• Reduced clinical time

Dillon, NAL
100
Better prescriptions

• Teresa Ching, Richard Katsch, Frances Lockhart

Dillon, NAL
101
Better prescriptions .. NAL-NL2

• Large study (80 subjects) just finished
• Examining relations between dead regions, tuning,
  effective audibility
• New loudness targets (reduced at high levels)
• Will lead to NAL-NL2
• definitely threshold based
• perhaps supra-threshold additional test
• NAL-NL21 by 2104 !

Dillon, NAL
102
100 years of hearing aids ..?

• Much less than 100 years to go?

Dillon, NAL
103
Growing hair cells (in chickens and mice)
Li, Roblin Heller 2003
Dillon, NAL
104
Thats all
105
This talk can be obtained from www.NAL.gov.au

• (After June 16)