Signals

1
High-Frequency Click-Evoked Otoacoustic Emissions
Shawn S. Goodman AAA AudiologyNow! 2009
2
Why High-Frequency CEOAEs?
Clinical Use
Assessment of high-frequency hearing Monitoring
cochlear status - noise-induced -
ototoxicity
Basic Research Use
Probe of cochlear mechanics - delay times
- growth rates - underlying generation
mechanisms
3
HF CEOAEs Talk Outline
Technical Issues
Transducers Calibration Extraction
Data
Physiologic / Behavioral Comparison Repeatability
Cochlear Delays Underlying generation
mechanisms
Future Directions
4
Cochlear Response to a Click
Basilar Membrane
Apex
Base
5
Cochlear Response to a Click
High Frequencies
Low Frequencies
6
Ideal Ear Canal Recording
Click Stimulus
CEOAE
Time
7
Ideal Ear Canal Recording
Analysis Window
Time
8
Problem 1 Loudspeaker Response
Real loudspeakers Finite bandwidth
Ring in time.
110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
9
Problem 1 Loudspeaker Response
Real loudspeakers Finite bandwidth
Ring in time.
110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
10
Problem 1 Loudspeaker Response
Click Stimulus overlaps HF CEOAEs
CEOAE magnitude reduced at HF due to stimulus
rolloff
11
Problem 1 Loudspeaker Response
Where to start analysis window?

• Too early -- include stimulus artifact.
• Too late -- exclude high frequencies.

12
Solutions
Theres no free lunch!
13
Solution 1 Modify Voltage Drive
Modify the voltage drive to make the output as
broad flat as possible.
110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
14
Solution 1 Modify Voltage Drive

• Reduced output levels.

110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
15
Solution 1 Modify Voltage Drive

• Reduced output levels.
• Includes unneeded LF energy.

110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
16
Solution 1 Modify Voltage Drive

• Reduced output levels.
• Includes unneeded LF energy.
• HF emphasis needed?

shape of behavioral auditory threshold curve
110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
17
Solution 1 Modify Voltage Drive
OK, get a better loudspeaker, decrease LF energy,
and increase HF energy.
110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
18
Solution 1 Modify Voltage Drive

• Good luck with that!
• Increased stimulus ringing.

110
100
0.1
1
10
0
1
2
3
Time (ms)
Frequency (kHz)
19
Solution 2 Nonlinear Extraction
Exploit stimulus linearity nonlinear OAE growth.
output
input
20
Solution 2 Nonlinear Extraction
Exploit stimulus linearity nonlinear OAE growth.
output
input


21
Solution 2 Nonlinear Extraction

• Extract full CEOAE?
• 2. Should use two loudspeakers
• to reduce stimulus distortion.

22
Solution 2 Nonlinear Extraction

• stimulus cancelled
• Early start includes high frequencies.

23
Problem 2 Calibration
2 cc Coupler Standing waves
0
-10
-20
-30
-40
4
8
1
0.5
2
16
Frequency (kHz)
24
Problem 2 Calibration
Solutions Avoid HF standing waves by using
a long tube and time windowing
Reflections from end of tube
Time
25
Problem 2 Calibration
Solutions Avoid HF standing waves by using
a long tube and time windowing
0
-10
-20
-30
-40
4
8
1
0.5
2
16
Frequency (kHz)
26
Problem 3 High-Frequency SNR
A single analysis window?
When these are the frequencies of interest
everything else is noise.
27
Problem 3 High-Frequency SNR
Multiple overlapping analysis windows improve SNR.
28
HF CEOAEs Talk Outline
Technical Issues
Transducers Calibration Extraction
Data
Physiologic / Behavioral Comparison Repeatability
Cochlear Delays Underlying generation
mechanisms
Future Directions
29
HF CEOAE Recording
Raw waveform (5 minute average 4050 repetitions)
Pressure (mPa)
Time (ms)
30
HF CEOAE Recording
Raw waveform (5 minute average 4050 repetitions)
Pressure (mPa)
Time (ms)
Clinical OAE devices discard response lt 2.5 ms,
which eliminates the high-frequency cochlear
response !
31
HF CEOAE Recording
Raw waveform (5 minute average 4050 repetitions)
A scaling cochlea has equal-duration ripples on
log-time axis.
Pressure (mPa)
0.25
0.5
2
5
20
Time (ms)
32
HF CEOAE Recording
Raw waveform (5 minute average 4050 repetitions)
Pressure (mPa)
0.25
0.5
2
5
20
Time (ms)
Clinical OAE devices discard response lt 2.5 ms,
which eliminates the high-frequency cochlear
response !
33
HF CEOAE Test Performance
446 Ears classified as normal/impaired
Perfect test
Random
34
HF CEOAE Test Performance
446 Ears classified as normal/impaired
35
HF CEOAE Repeatability
49 Ears Retested On Different Day
36
Biologic Distortion In HF CEOAEs
5 Subjects With Cochlear Implants (turned off)
37
HF CEOAEs Cochlear Delay
CEOAE after narrowband filtering
38
HF CEOAEs Cochlear Delay
Compute the Filtered Waveform Envelope
39
HF CEOAEs Cochlear Delay
Amplitude (mPa)
Compute the Filtered Waveform Envelope
40
HF CEOAEs Cochlear Delay
Amplitude (mPa)
Results show multiple peaks

• Peak delays do not change with level.
• Peak amplitudes grow at different rates.
• Suggests multiple generation mechanisms.

41
Conclusions
Technical Issues
Transducers designed for HF are needed.
Data
HF CEOAEs identify ears with HL up to 12 kHz HF
CEOAEs have good repeatability HF CEOAEs have
multiple energy peaks Suggesting multiple
generation mechanisms.
Future Directions
Explore HF CEOAEs in ototoxicity NIHL
42
Acknowledgements Douglas H. Keefe John C.
Ellison Denis F. Fitzpatrick Michael Gorga Walt
Jesteadt Michele Milanuk Boys Town National
Research Hospital, Omaha, NE NIH DC003784,
DC004662 Ian B. Mertes University of Iowa,
Iowa City, IA Goodman et al., (2009)
High-frequency click-evoked otoacoustic emissions
and behavioral thresholds in humans. JASA, 125,
1014-1032. http//www.uiowa.edu/comsci/research
/arl/publications.html
43
Thank You!