Transcript PowerPoint Handout - (CSD) - JMU

Results from a
Deployable StereoHearing Test System
Lincoln Gray
JMU
• I have relevant financial relationship(s) with
the products or services described, reviewed,
evaluated, or compared in this presentation.
• I receive a salary from James Madison University.
• Students, collaborators and I have received support from
•
•
•
•
•
4-VA Research Grants; 2013-2014 and 2014-2015 ($4600 and $1200)
JMU School of Engineering support for their students’ capstone projects.
JMU Roger Ruth Memorial Research Grant; 2014 to 2015 ($910)
JMU Madison Trust Fund; 2014 - present (~ $7000)
Lions of Virginia 2016 (~$1000)
• I have possibly relevant nonfinancial
relationship(s) to disclose.
• Intellectual Property Disclosures have been filed;
• We are free to distribute the system
• 5 year study: 2 School of Engineering capstone projects by
Brandon Lancaster, Brian Allen, Michael Kessler, Jonathan
Smith , Brittany Harwell, Tony Battu; Undergrad honors
thesis and Au.D. dissertation (2017) by Sofia Ganev. Many
student volunteer participants.
• Collaborated with Dr. Bradley Kesser, M.D. at UVa, and
Robert Nagel, JMU School of Engineering.
• Made possible by the late Drs. Robert Jahrsdoerfer and
Roger Ruth
• Engineers developed a small deployable system to test
sound localization and understanding speech-in-noise
• This presentation (following the Ganev dissertation)
• Show the device separates the performance of one- and
two-eared listeners in these tasks with large effect sizes
• Collect binaural processing data on patients sometime
after atresia repair
• Anatomical deformity of the outer and middle ear, due to abnormal
embryological development. Can cause maximum conductive
hearing loss (40-70 dB HL). Is surgically correctable
• 1 in 10,000-20,000 births; Unilateral > bilateral; Right ear > left ear;
Males > females.
• Surgery between the ages of 5 and 45, provides air conducted
sound to the atretic ear.
• After one month of recovery: patient returns to have the
gauze/packaging removed, an updated audiogram is completed,
revealing improved thresholds.
Audiometric Thresholds
PTA
known maximum
conductive hearing loss
that is suddenly corrected
to normal or near normal
limits
Upper
lim of
norm
or
95%
CI
5
(Wilmington et al, 1994)
• No Binaural Processing Before Surgery (unilateral listeners),
and we might expect:
• Thresholds 3 dB worse because of no binaural summation
• No head shadow effect (hard to hear on bad side)
• Poor horizontal localization
• Unilateral conductive hearing losses in children can have a
significant impact on academic performance and are more
likely to use/need intervention services (Kesser et al., 2013; PMID:
23483556)
Resource
Repeated a
grade
Any resource
Behavior
problem
Atresia
Group
(n=40)
0 (0%)
SNHL Group
(n=11)
2 (18.2%)
Bess and
Tharpe
(n=60)
21 (35%)
26 (65%)
7 (63.6%)
8 (13.3%)
5 (12.5%)
3 (27.3%)
12 (20%)
RMS Error in Degrees
Problem: significant improvement at 1 month, but not very good,
but with some indication it may improve with more time.
Horizontal Sound
Localization
7
(Wilmington et al, 1994)
Puzzlingly, because patients are normal in detecting cues
that we use to localize, but they don’t localize well – maybe
then need more time to make sense of the new cues
μs
Interaural Temporal
Difference Limens
8
Reference
Kesser et al., 2016
PMID: 26963665.
Gray et al., 2009
PMID: 19581007.
Wilmington et al. 1994
PMID: 8040103.
“Longer follow-up is an important next step; the
patients may take more time to learn new
complex tasks involving use of signals from two
ears” (Gray et al., 2009)
ATRESIA
Travel to UVA
Issue: lack of follow-up data on
improvement of binaural processing over
time
Corrective
Surgery
Follow-Up
Improved
Binaural Ability?
Improved
Audiogram
Solution: Deploy a sensitive hearing testing system for annual testing to the comfort of the
patient’s home in a UPS or FedEx package
Insurance commonly covers the costs associated with returning to the
medical center for the one month post-op follow-up appointment,
however, it will not cover any costs to return annually for analysis of
binaural hearing ability in order to collect this data
Our device (~ like a ‘picnic basket’)
• Laptop, 8 speakers, custom speaker stands,
hub-box, custom tablemat, microphone,
mouse
• Costs $2000, ships for $50-100 roundtrip; fits
in a convenient package
4
3
2
5
6
7
8
1
Glued to
Glued to
mat
Sets up on a table
Sound Localization Test
48 trials, of 250 ms broadband noise,
average 70 dB SPL (10 dB rove)
Speech in Noise Test
Speech is Coordinate Response Measure
Corpus: “Ready (call-sign), go to (color, #) now”.
at 60 dB SPL. Level of noise adaptively varied (1
up, 1 down staircase) to find threshold.
4 conditions of speech-in-noise testing
Best
Good
• Best: Speaker 8: speech
Speaker 1: noise
• Good: Speaker 5: speech
Speaker 4: noise
• Poor: Speaker 4: speech
Speaker 5: noise
• Worst: Speaker 1: speech
Speaker 8: noise
Poor
Worst
Right ear is considered the “good ear”
Preview of 2 dependent
variables:
 Difference of Best to Worse
 Difference of Good to Poor
Part 1: 50 participants
• 40 of 50 = normal or ‘bilateral’
• 10 of 50 = unilateral hearing loss (one
severely or profoundly impaired ear, 97 dB HL
average, and normal hearing in the
contralateral ear)
• Most ‘normal’ were tested again with plugged
ear
• Disposable foam earplug combined with a circumaural earmuff
covering over the earplug
• Total attenuation of this artificial conductive HL is estimated to be
approximately 56 dB SPL
• Between 18 and 65 years old
Part 1: sound localization results
• Post-Hoc
comparison:
bilaterals significantly
different than
unilaterals (p<0.001;
effect size 3.0)
RMS Error of Bilateral, Unilateral, and Artificially-Plugged Groups
RMS Error in Degrees
• Oneway ANOVA:
significant
differences between
all groups (F2,56=71;
p<0.001)
Bilateral
Unilateral
Plugged
Part 1: Speech in Noise Results
Mean Level of the Noise (dBA)
Level of ‘easy’
Difficulty
Per CRM
Condition
for Bilateral
and
Unilateral
‘hard’
Subjects
Bilateral Group
Unilateral Group
CRM Conditions
Best
Good
Poor
Worst
Part 1: Speech in Noise Results
CRM Performance Difference Between Best and Worst Conditions
significant effect of
group (F2,35=15.2,
p<.001)
 Post-hoc LSD tests
showed that all groups
were significantly
different (p=.007 or less)
 Bilaterals and
unilaterals, significantly
Effect
Size = 3
Difference Between Best and Worst
Conditions (dB)
 Oneway ANOVA:
different,
Bilateral
Plugged
Unilateral
CRM Performance Difference Between Good and Poor Conditions
significant effect of
group (F2,36=3.9,
p=.03)
 Post-hoc LSD tests
showed that all groups
were significantly
different (p=.04 or
less) except unilaterals
and plugged (p=.43)
 Bilaterals and
unilaterals,
significantly different,
Effect Size = 1.6
Difference Between Good and Poor Conditions
(dB)
 Oneway ANOVA:
Bilateral
Plugged
Unilateral
System successfully separates performance
of one- and two eared listeners. What
happens after atresia repair:
• 9 post-op atresia patients tested with our system
shipped to their homes.
• Age between 5 and 25 years old, (M = 12.9, SD = 5.8)
• Completed the study within a range of 4 months to 11
years after their (most recent) atresia repair (average of
3.8 years since surgery)
• The mean pure-tone average (PTA) for the atretic ear
prior to operation was 59.2 dB HL (SD=11.7 dB HL)
• Mean PTA for the repaired atretic ear after the most
recent surgery was 25.9 dB HL (SD=12.5 dB HL)
• 33 dB change in air-conducted audiogram
• Oneway ANOVA showed
a significant difference
between groups
(F3,64=53, p<.001)
• Post hoc tests (LSD)
showed that there was
no difference between
the atresia patients and
the ‘normal’ bilateral
listeners (p=0.129)
• Atresia patients
localized significantly
better than the
unilaterals (p=0.046,
Effect Size = 1.6)
RMS Error in Degrees
Part 2:Localization Performance in RMS
Error in Degrees for All Groups
Bilateral
Post-Op Atresia Unilateral
Plugged
Part 2: Speech-in-Noise
Performance for Groups in 4
CRM Conditions
Post-Op Atresia Group
Bilateral Group
Plugged Group
Mean Level of the Noise (dBA)
Mean Level of the Noise (dBA)
Unilateral Group
CRM Conditions
CRM Conditions
Best
Good
Poor
Worst
Best
Good
Poor
Worst
• Oneway ANOVA showed a
significant difference
between the groups,
(F3,44=11.5; p<.001)
• Post hoc tests (LSD)
showed that there was a
marginal difference
between the post-op
atresia patients and the
bilaterals (p=0.047), and a
highly significant
difference between the
post-op atresia subjects
and the unilaterals
(p=0.001)
Difference Between Best and Worst Conditions (dB)
Part 2: Speech-in-Noise Understanding
for Best and Worst CRM Conditions
Bilateral
Post-Op Atresia Unilateral
Plugged
• Oneway ANOVA showed a
significant difference
between all groups,
(F3,44=3.2; p=.033)
• Post hoc tests (LSD) showed
that there was now no
difference between the
post-op atresia subjects and
the bilaterals (p=0.775), and
still a significant difference
between the post-op
atresia subjects and the
unilaterals (p=0.05)
• The harder and most
realistic condition
Difference Between Good and Poor
Conditions (dB)
Part 2: Speech-in-Noise Understanding for
Good and Poor CRM Conditions
Bilateral
Atresia
Unilateral
Plugged
Bilateral
Post-Op
Atresia
Unilateral Plugged
Part 2: Next, we explored trends and
correlations within the atresia group
Localization and Speech-in-Noise Ability for Post-Op Atresia Subjects
High Difference
(like unilateral)
Speech-in-Noise: Difference in
Good to Poor Conditions (dB)
Localization: RMS Error in Degrees
Low Accuracy
(like unilateral)
Low Difference
(like bilateral)
High Accuracy
(like bilateral)
• Performance
trend between
these two binaural
processing tasks is
not correlated
• Suggests that
development of
binaural
processing might
not be a unitary
event
• Although
there is some
variability;
localization
improves with
chronical age
of new ear in
an expected
non-linear
trend
RMS Error in Degrees
Years Since Last Surgery and Localization Ability
Years Since Last Atresia Repair
• Speech in noise
appears to improve
with time in a nonlinear trend, as
expected
• With the exception
of a single outlier
• It should be noted
that the outlier’s PTA
was 60 dB HL before
surgery and
improved to 47 dB HL
Speech-in-Noise: Difference in Best to
Worst Conditions (dB)
Years Since Last Surgery and CRM Ability for Post-Op Atresia Subjects
Years Since Last Atresia Repair
post-op
pure tone
averages
(PTAs)
predict
speech in
noise
Difference Between Best and Worst Conditions (dB)
PTA of Post-Op Atresia Subjects and CRM Ability
r = 0.865
Post-Op Puretone Average (dB)
Conclusions:
• Atresia patients after surgery performed better than
unilateral counterparts and near their normal hearing
counterparts in two aspects of binaural processing,
confirming a benefit of the surgical repair not only for
audiometric thresholds, but also for binaural processing
• Results suggest that a learning curve may exist for
learning to use the newly repaired ear; years passed
since surgery is one predictor of success in these two
binaural tasks, yet clinical measures such as puretone
average (PTA) immediately following surgery is
predictive
• High variability suggest the need to study more patients
References
• Allen, B. D., Battu, T., Ganev, S. A., Gray, L. C., Harwell, B. N., Kesser, B. W., Kessler, M. A., Lancaster, B. C., Nagel, R.L., &
Smith, J.I. (2013). Design of a distributable stereo hearing test package. Unpublished manuscript, Department of
Engineering, James Madison University, Harrisonburg, VA.
• Berlin, C. I., Hughes, L. F., Lowe-Bell, S. S., & Berlin, H. L. (1973). Dichotic Right Ear Advantage in Children 5 to 131. Cortex,
9(4), 394-402. doi:10.1016/s0010-9452(73)80038-3
• Bolia, R. S., Nelson, W. T., Ericson, M. A., & Simpson, B. D. (2000). A speech corpus for multitalker communications
research. The Journal of the Acoustical Society of America J. Acoust. Soc. Am., 107(2), 1065. doi:10.1121/1.428288
• Breier, J., Hiscock, M., Jahrsdoerfer, R., & Gray, L. (1998). Ear advantage in dichotic listening after correction for early
congenital hearing loss. Neuropsychologia, 36(3), 209-216. PMID: 9622186.
• Gray, L., Kesser, B., and Cole, E. Detection of Speech in Noise after Correction of Congenital Unilateral Aural Atresia:
Effects of age in the emergence of binaural squelch but not in use of head-shadow. International Journal of Pediatric
Otorhinolaryngology 73: 1281–1287, (2009). PMID: 19581007.
• Harwell B, Battu T, Ganev S, Nagel R, Gray L, Kesser B. Design of a Distributable Stereo Hearing Test Package (2014).
Paper presented at: Systems and Information Engineering Design Symposium (SIEDS), 2014 IEEE, University of Virginia;
Doi: 10.1109/SIEDS.2014.6829922
• Jahrsdoerfer, R. A., Yeakley, J. W., Aguilar, E. A., Cole, R. R., & Gray, L. C. (1992). Grading System For The Selection Of
Patients With Congenital Aural Atresia. Otology & Neurotology, 13(1). PubMed PMID: 1598988.
• Kesser, B. W., Krook, K., & Gray, L. C. (2013). Impact of unilateral conductive hearing loss due to aural atresia on academic
performance in children. The Laryngoscope, 123(9), 2270-2275. PMID: 23483556
• Kesser, B. W. (2014, March 21). Aural Atresia. Retrieved March 10, 2016, from
http://emedicine.medscape.com/article/878218-overview
• Kesser, B, Cole, E. Gray, L., (2016) Emergence of Binaural Summation after Surgical Correction of Unilateral Congenital
Aural Atresia. Otology & Neurotology37: 499-503, 2016. PMID: 26963665
• Nicholas, B. D., Kaelyn, K. A., Lincoln, G. C., & Kesser, B. W. (2012). Does preoperative hearing predict postoperative
hearing in patients undergoing primary aural atresia repair. Otology & Neurotology, 33(6):1002-6. PMID: 22772017
• Shonka DC, Jr, Livingston WJ, III, Kesser BW. The Jahrsdoerfer Grading Scale in Surgery to Repair Congenital Aural Atresia.
Arch Otolaryngol Head Neck Surg.2008;134(8):873-877. doi:10.1001/archotol.134.8.873.
• Wilmington, D., Gray, L., & Jahrsdoerfer, R. (1994). Binaural processing after corrected congenital unilateral conductive
hearing loss. Hearing Research, 99-114. PMID: 8040103