Multimodal Assessment and Speech Perception Outcomes in Children with Cochlear Implants or Hearing Aids

Transcription

1 Multimodal Assessment and Speech Perception Outcomes in Children with Cochlear Implants or Hearing Aids Karen Iler Kirk, Ph.D., CCC-SLP* Shahid and Ann Carlson Khan Professor and Head Department of Speech and Hearing Science University of Illinois at Urbana-Champaign Work supported by NIDCD Grants R01 DC and P50 DC DISCLOSURE: Test materials described in this presentation are licensed to G.N. Otometrics through The University of Iowa

2 Collaborators The University of Iowa Karen Iler Kirk, PhD (PI) Lindsay Prusick, AuD Virginia Driscoll, MA Nathaniel Wisecup, BA Lauren Diamond, BA Lauren Dowdy, BA Ruth Flaherty, BA House Research Institute, Los Angeles Laurie Eisenberg, PhD (PI) Amy Martinez, MA Dianne Hammes Ganguly, MA Children s Memorial Hospital, Chicago Nancy Young, MD (PI) Susan Stentz, AuD Lisa Weber, AuD Iguehi James, MPH Washington State University Brian French, PhD (PI) Chad Gotch, MS University of Illinois Michael Novak, MD Jean Thomas, AuD Michael Hudgins, BA

3 Introduction Listeners must extract linguistic message from highly variable acoustic speech signal Variability introduced by: Talker characteristics - gender, age, dialect and speech rate Environment noise, reverberation Presentation format A-only vs. Auditory-plus-Visual Linguistic characteristics Word frequency how often words occur in language Lexical Density the number of phonemically similar words or lexical neighbors

4 Multimodal Lexical Sentence Test (MLST-C TM ) 21 lists of 8 sentences 10 talkers 3 key words per sentence Key words in each sentence drawn from the same lexical category Strong Psychometric principles Lists are reliable and equivalent within each format: V, A, AV

5 Purposes To examine performance in quiet in children with cochlear implants or hearing aids To examine performance in noise as a function of Signal-to-Noise Ratio (SNR) Group CI only CI + HA To evaluate AV enhancement as a function of SNR

6 Assessing Performance in Quiet Participants (N=68) CI Users (n = 32) HA Users (n = 27) CI+HA Users (n=9) Mean Age at Test 9.86 years 8.92 years 9.61 years Mean Age at Onset.18 years 1.07 years 0 years Mean Age Fit (current sensory aid) Hearing Loss Classification (4 frequency PTA) Receptive Vocabulary Age Gender 2.93 years 3.18 years 6.67 years Profound - 20* Unknown - 12 Mild - 9 Moderate 15 Severe - 2 Profound - 1 CI: Profound HA: Severe -8 Profound years 9.73 years 9.63 years 16 males 16 females 16 males 11 females 7 males 2 females *Pre-implant thresholds

7 Methods All MLST TM lists administered twice Every participant tested in all 3 presentation formats 1/3 Visual Only 1/3 Auditory Only 1/3 Auditory + Visual Additional auditory only testing Isolated word recognition (PBK) 2 lists of HINT-C Procedures Speech presented at 60 db SPL in quiet Verbal responses scored as percent correct Receptive Vocabulary assessed using PPVT

8 Results: Multimodal Sentence Recognition

9 Auditory-Only Speech Recognition

10 Performance in Noise: CI Participants (n=20) CI (n=14) CI+HA (n=6) Mean Age at Test 12.1 yrs 9.8 yrs Mean Age at Implantation Mean Length of Device Use 3.2 yrs 5.5 yrs 9.9 yrs 4.5 yrs Type of CI Freedom (n=5) CI24M (n=5) CI24R (n=1) Countour (n=3) Hybrid L24 (n=2) Contour (n=1) Freedom (n=2) CI512 (n=1)

11 Procedures Speech administered at 60 dba SPL Each participant tested in A and AV formats Quiet SNRs: -5, 0, +5, lists per condition (2 formats X 4 SNRS = 8 lists) Verbal responses scored as % key words correct Logistic regression computed to estimate Speech Recognition Threshold

12 Results Percent Correct Q Q A AV CI Signal-to-Noise Ratio CI+HA

13 Results: SRT Percent Correct CI Percent Correct CI + HA Signal-to-Noise Ratio Signal-to-Noise Ratio CI CI + HA A only 2.8 db -0.6 db A + V -8.5 db db A AV

14 Audiovisual Gain R a = (AV-A)/(100-A) Relative gain in accuracy in AV condition relative to A only Used by Lachs et al. (2001) to examine AV speech perception in children with CIs

15 R a by SNR R a Q Signal-to-Noise Ratio

16 Conclusions The MLST-C TM Incorporates "real-world stimulus variability Multiple talkers Different presentation formats Is a more sensitive measure of performance than traditional tests The addition of visual cues enhances speech perception Largest improvements at poorer signal-to-noise ratios Not all children show similar benefit Enhancement is larger for children with acoustic low frequency hearing Future testing to examine factors related to AV enhancement