National Acoustic Laboratories

1
Evoked cortical assessment of aided children (and
estimation of hearing thresholds in adults)
Harvey Dillon, Suzanne Purdy, Maryanne Golding,
John Seymour, Lyndal Carter, Wendy Pearce,
Richard Katsch, Mridula Sharma, Katrina Agung,
Kirsty Gardner-Berry
National Acoustic Laboratories Cooperative
Research Centre for Hearing Speech Science, The
University of Auckland, New Zealand Audiology,
Macquarie University
2
Evaluation of aided functioning in infants

• Universal new born screening

Early fitting of hearing aids
Need for an evaluation method
3
So baby, how does it sound?

• Objective hearing aid evaluation for
• young infants
• difficult-to-test people

4
Why the rush?Language ability 6 months after
implantation
5
Early intervention leads to better language
development at 6 months after fitting (n90)
Significant effect of age of fitting p 0.001
6
At 6 and 12 months after fitting (n66)
7
Why use cortical responses?
8
Why cortical responses to evaluate hearing aid
fitting in infants?

• Reliably present in awake young infants
• More likely to correlate well with perception
• Can be elicited by a range of speech phonemes
  close to desired outcomes
• Stimuli handled reasonably by hearing aids
• Can be very frequency specific if needed

9
Where do cortical responses come from?
10
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
11
The end of the road
12
Auditory cortex orientation
13
Auditory cortex orientation
14
Auditory cortex orientation
15
Hudson, 2009
16
Auditory cortex and current sources
Sussman et al (2008)
17
Cortical responses in adults with normal hearing
18
Adult
19
Adult grand mean waveforms at Cz
Speech
Tones
20
Cortical responses in infants
21
Infants
P
10
µV
5
N
0
-5
500
600
300
400
100
200
-100
0
ms
22
Maturational effects on cortical evoked response
morphology

• N8-16 per grand mean
• Cz site
• stimulus 10 click train, 2 ms ISI _at_ 65 dB SL
• rate 1.3/s

• Ponton et al (2000)

23
2 years
12 years
I
?
II
? Fewer neuro-filaments in young children,
especially in more superficial cortical layers
thought to generate N1 (Ponton, Moore
Eggermont 1999)
III
IV
V
Axonal neuro- filaments
Axonal neuro- filaments
VI
Cell bodies
Cell bodies
24
Latency versus age
25
(No Transcript)
26
Maturation with time in sound

• Ponton and Eggermont 2007

27
ConclusionAuditory system maturity

• The latency of cortical potentials indicate the
  maturity of the auditory system.
• Latency matures consistent with the time in
  sound (Ponton and Eggermont, 2007)
• . provided implantation occurs by 3.5 years of
  age (Sharma, 2002)

28
Cortical responses in infants to different speech
sounds

• Evidence of speech discrimination?

29
N1, P2 amplitude

• N1, P2 amplitude advantage at Cz in adults for
  all stimuli

30
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

31
Number of infants (N20) with significantly
different cortical responses to pairs of stimuli
m vs t m vs g t vs g
Based on MANOVA at Cz, 101 to 500 ms post-onset,
in eight bins each 50 ms
32
Grand Average n 16 infants
33
Number of subjects (out of 20) with significant
differences between responses
34
Are /tae/ /mae/ cortical responses different in
hearing impaired children?

• 9 subjects, 9 ears
• 6-12 years
• sloping, mild-severe hearing loss
• hearing aids fitted to NAL-NL1 but not optimally
• 55 had different responses

• 10 subjects (5 with poor hearing aid progress),
  14 ears
• 8 infants 6-20 months, 2 children 4 10 years
• 4 moderate, 8 severe, 2 profound ears
• 64 had different responses based on individual
  ANOVA

35
Adults
P2
P1
N1
P
Infants
36
Conclusion Cortical response shape for different
speech sounds

• Cortical response shape for different speech
  sounds, and tone bursts are similar, but not
  identical, in shape across stimuli. Cortical
  responses to sounds presented in isolation are
  better suited as indicators of detection.

• Future work Cortical response to change, as a
  measure of discrimination

37
Three speech sounds /m/ /g/ /t/
38
Why not obligatory cortical responses?

• variable shape across ages
• variable shape with auditory experience
• variable shape from person to person
• variable shape from time to time (state of
  person, especially sleepiness)
• variable shape with stimulus
• Variable shape with inter-stimulus interval

An automated method of response detection and
response differentiation
39
Automatic detection of cortical responses
40
Desirable characteristics

• No reliance on a template
• Able to use information from contributing
  portions of waveform
• Able to discount non-contributing portions of
  waveform

41
Analysis using Hotellings t2 statistic
X3
Voltage
Time

• Divide each record into 50 ms time bins
• Average data points within each time bin
• Use these averages as variables in
  Hotellings t2 analysis
• Result is probability of the
  waveform being random noise

X a1X1 a2X2 ........ a9X9 Test is
there any set of weighting coefficients for which
X ? 0?
42
Presentation of average response in series
43
Receiver Operating Characteristics Expert judges
Sensitivity
d
1 - Specificity
44
ROC 200 repetitions adults with normal
hearing to moderate loss
10 dB
0 dB
- 10 dB
45
d results - for 60 stimuli
46
d results - for 200 stimuli
47
Growth of amplitude with SLadults tonal stimuli
48
Loudness growth above threshold
Hellman Meiselman, 1990
49
Effect of response amplitude on detectability
100 epochs Adults, tonal 10, 20, 30 dB SL
50
Detectability (adults tonal stimuli)
P0.05
P0.01
P0.001
51
Significant responses normal hearing and
hearing impaired Adults tonal stimuli (n100 or
200)
52
(No Transcript)
53
Proportion with responses present - adults
54

• .. but infants move around !

55
Residual noise level

• rms noise standard deviation / vn

56
Residual noise levels (for 100 epochs)
Awake adults
5
0
But also larger responses
57
Growth of amplitude with SLadults tonal stimuli
Hearing impaired adults
Normal hearing adults
58
Detectability versus residual noise- infants -
59
Response size and residual noise
Infants speech stimuli
60
Infants Hotellings versus experts

• Normal hearing infants aged 7 to 16 months

61
Proportion with responses present (plt0.05)
normal hearing infants 100 epochs
62
Detection of speech sounds
63
Conclusions Detecting cortical response presence

• Large responses are more easily detected
• Response amplitude grows with sensation level
• Response amplitude is larger for people with
  sensorineural hearing loss that for people with
  normal hearing
• Responses are more easily detected when residual
  noise is low (lt3.4 µV for infants lt 1.5 µV for
  adults)

64
Estimating hearing thresholds in adults
65
Good agreement between CAEP and audiometric
thresholds in awake adults
Tsui, Wong Wong 2002
66
From Rickards, F. et al (1996) Cortical Evoked
Response Audiometry in
noise induced hearing loss claims. Aust.
J. Otol. 2 (3)
67
Test sequence

• Evaluation using HearLab
• Automated response detection
• 40 ms tone bursts

60
Response present P lt 0.05
Response absent P gt 0.05
85
30
100
75
45
15
110
95
80
70
55
40
25
5
90
65
50
35
20
10
0
105
68
Cortical threshold vs behavioural threshold
69
(No Transcript)
70
Davis (1965) Cortical evoked potential versus
behavioural thresholds
71
Conclusions Estimating behavioural thresholds
in hearing-impaired adults

• Cortical thresholds overestimate behavioural
  thresholds by 2.4 dB, on average
• Standard deviation of cortical behavioural
  threshold differences is 6.3 dB
• Applications hearing compensation, clients
  unable to respond

72
Cortical responses and functional performance

• Do cortical responses tell us about real-life
  auditory performance?

73
Parents Evaluation of Aural/oral performance in
Children (PEACH) Questionnaire

• Parents are asked to describe their babys
  aural/oral
• skills based on real-life experiences
  (listening in quiet
• and in noise and alertness to environmental
  sound)
• Scores are assigned based on the number of
  observed
• behaviors and how frequently these occur.
• Final overall score of 0 40 can be calculated
  (and
• reported as a percentage).

74
Results PEACH score by age
Normative data curve
75
Functional deficit vs number of cortical
responses present at
N 24 p 0.001 12 sensorineural 7 auditory
neuropathy 5 multiply disabled
76
Functional deficit versus cortical score
All aided children rs 0.60 n24 p 0.001
SN only rs0.61 n12 p 0.02 MD
only rs0.82, n5 p 0.04 AN only rs0.36
N7 p 0.22
77
Conclusions Cortical responses and functional
performance

• The greater the number of speech sounds that
  result in cortical responses in aided infants,
  the greater the functional performance in real
  life, on average

78
Practical implementation of cortical testing
HearLab

• Disclosure NAL will get a royalty for each unit
  sold.
• Thank you The HearLab development team
• Teck Loi, Barry Clinch, Isabella Tan, Dan Zhou,
  Scott Brewer, Ben Radzyn

79
Aim

• To make available to clinicians
• Stimuli - m, t, g (pure tones)
• Statistical tests (Hotellings t2)
• Age appropriate norms
• Residual noise monitoring
• Active electrodes
• Future NAL developments
• To supplement revenue available for research

80
HearLab
81
(No Transcript)
82
(No Transcript)
83
Play video

• 415
• 5.44

84
Reducing measurement variability (random
electrical signals) ? Speeding up
measurements? Increasing validity of
interpretation
85
Active electrodes
86
Capacitive Coupling 50 Hz
Passive Electrodes
87
Capacitive Coupling 50 Hz
Active Electrodes
88
Capacitive Coupling 50 Hz
Active Electrodes
Passive Electrodes
89
Signal processing options

• Noise down by x ? measurement time down by x2
• Several potential improvements under research

90
Conclusions Active on-scalp pre-amplification

• Electrical leads are less sensitive to
• capacitive coupling electrical interference
  (because of low amplifier output impedance)
• Inductive coupling electrical interference
  (because of gain within pre-amplifier)

91
Clinical applications and implications
92
Clinical applications of corticals

• For finding thresholds (when awake) only so far
  tested in adults
• Determining whether speech sounds are audible
• aided or unaided
• for patients who cant respond reliably by
  behavioral testing e.g., infants, multiply
  disabled people.

93
Cortical potential and new-born screening
94
Clinical implications of corticals

• Significant response is obtained to speech at 65
  dB SPL
• No significant response is obtained to speech at
  65 dB SPL or to speech at 75 dB SPL

95
Noisy results - chewing
96
Clinical implications of corticals (cont)
No /t/ response
Mixed results (and noise is low)
No /g/ response
No /m/ response
Mixed results (and noise is high)
97
Hearing loss at birth .. for parents
Parental denial
Working towards a solution
Pessimism and hopelessness
98
Some recent cases

• Hsiuwen Chang

99
No /m/ response
100
Case 1
No cortical responses, and the results helped the
parents accept the need for cochlear implants

• Age at aided cortical testing
• Visit 1
• 6 weeks old (Initial hearing aid fitting day)
• Visit 2
• 3 months old

101
(No Transcript)
102
Visit 1
P lt 0.05 ? . No
103
Visit 2
104

• The infant received bilateral cochlear
  implantations at 5 months of age.
• Email from the babys parents
• Thank you so much for the information you gave
  us on the previous testing as it helped us with
  our decision to proceed with the implants.

105
Case 2
No cortical responses, even after hearing aids
have been increased in gain for the third time
and the parents dont want a cochlear implant for
their baby

• Age at aided cortical testing
• Visit 1
• 13 weeks old
• Visit 2
• 17 weeks old
• Visit 3
• 21 weeks old
• Hearing aid fitting at 8 weeks of age

106
(No Transcript)
107
Visit 1
108
Visit 2
Right aided
Left aided
109
Visit 3
110

• The parents still believe that their baby boy can
  be oral by using hearing aids.
• They are hoping to see that their baby can
  benefit from a more powerful hearing aids.
• The babys hearing aids were changed from Siemens
  Explorer 500 P to Phonak Una SP after Visit 3.

111
Case 3
Corticals provided reassurance about the baby
hearing well

• Age at aided cortical testing
• Visit 1
• 4.5 months old
• Visit 2
• 5.5 months old
• Hearing aid fitting at 5 weeks of age

112
(No Transcript)
113
Visit 1
114
Visit 2
115
Email from mother We feel very relieved, as
our faith has been restored in the hearing aids
as a result of what we discovered from the
results.
116
Case 4
Too few significant cortical responses, and the
aid gain was increased, resulting in more
cortical responses

• Age at aided cortical testing
• Visit 1
• 8 months old
• Visit 2
• 9 months old
• Hearing aids have been increased in gain two
  weeks before the second visit.
• Hearing aid fitting at 9 weeks of age

117
(No Transcript)
118
Visit 1
Visit 2
119
Case 5
A case where cortical testing was not possible

• Age at testing 4.5 years
• Multiple disabilities
• A reliable behavioural audiogram has not yet been
  obtained.

120

• She was wiggling all the time.

121

• This is the quietest state she could be, but only
  lasted for a few seconds.

122
Case 6 Auditory neuropathy, prior to behavioural
testing

• Some background .

123
Application for auditory neuropathy spectrum
disorder (AN)

• 15 of babies found to have hearing loss at birth
  in NSW have AN
• Management unclear (no device, hearing aid or
  cochlear implant)
• Rance showed close relationship between cortical
  response in older children and benefit from
  hearing aids
• Cortical responses more indicative than ABR of
  behavioural thresholds

124
FREQUENCY (Hz)
ABR 28/8/03 - NR
250 500 750 1000 1500 2000
3000 4000 6000 8000
0 10 20 30 40 50 60 70 80 90 100 110
HEARING LEVELS IN DECIBELS
ABR NR
ABR NR

Case Study 2
125
FREQUENCY (Hz)
ABR 28/8/03 - NR CAEP 14/10/03
250 500 750 1000 1500 2000
3000 4000 6000 8000
0 10 20 30 40 50 60 70 80 90 100 110
HEARING LEVELS IN DECIBELS
m g
t
ABR NR
ABR NR



Case Study 2
126
FREQUENCY (Hz)
ABR 28/8/03 - NR 15/3/04 - NR CAEP
14/10/03 30/3/04 ECochG 15/3/04
250 500 750 1000 1500 2000
3000 4000 6000 8000
0 10 20 30 40 50 60 70 80 90 100 110
HEARING LEVELS IN DECIBELS
m g
t
ABR NR
ABR NR
ECochG
ECochG
ECochG
ECochG
Case Study 2
127
FREQUENCY (Hz)
ABR 28/8/03 - NR 15/3/04 - NR CAEP
14/10/03 30/3/04 ECochG 15/3/04 VROA 29/4/04
250 500 750 1000 1500 2000
3000 4000 6000 8000
0 10 20 30 40 50 60 70 80 90 100 110
HEARING LEVELS IN DECIBELS
m g
t
ABR NR
ABR NR
ECochG
ECochG
ECochG
ECochG
Case Study 2
128
Corticals for more advanced measurements

• To summary

129
/Ah/ 2 second duration
0msG.avg
5msG.avg

10msG.avg
20msG.avg
50msG.avg
5.0
2.5
Offset
0.0
Onset
µV
Gap
-2.5
-5.0
-7.5
-350.0
150.0
650.0
1150.0
1650.0
2150.0
2650.0
ms
Kirsty Gardner-Berry
130
Indicator of binaural functioning
N1 and P2 cortical amplitudes for /a/ in noise
are enhanced when 700 ?s inter-ear delay is
introduced to noise in normal listeners (N8,
19-32 years)
131
Size of the electrophysiological unmasking
effect is correlated with behavioral MLD (speech
detection threshold) at -10 dB SNR
r-.76, p.028
132
Summary

• Cortical responses
• Evaluate the audibility of speech sounds
• Indicate the maturity of the auditory system
• Estimate hearing thresholds when the patient is
  awake (adults)
• Automatic detection as good as experts
• Residual noise size critical

133

• Thanks for listening
• www.nal.gov.au