J Am Acad Audiol 24:927–940 (2013)

Electromagnetic versus Electrical Coupling of Personal Frequency Modulation (FM) Receivers to Cochlear Implant Sound Processors DOI: 10.3766/jaaa.24.10.5 Erin C. Schafer* Denise Romine* Elizabeth Musgrave* Sadaf Momin* Christy Huynh*

Abstract Background: Previous research has suggested that electrically coupled frequency modulation (FM) systems substantially improved speech-recognition performance in noise in individuals with cochlear implants (CIs). However, there is limited evidence to support the use of electromagnetically coupled (neck loop) FM receivers with contemporary CI sound processors containing telecoils. Purpose: The primary goal of this study was to compare speech-recognition performance in noise and subjective ratings of adolescents and adults using one of three contemporary CI sound processors coupled to electromagnetically and electrically coupled FM receivers from Oticon. Research Design: A repeated-measures design was used to compare speech-recognition performance in noise and subjective ratings without and with the FM systems across three test sessions (Experiment 1) and to compare performance at different FM-gain settings (Experiment 2). Descriptive statistics were used in Experiment 3 to describe output differences measured through a CI sound processor. Study Sample: Experiment 1 included nine adolescents or adults with unilateral or bilateral Advanced Bionics Harmony (n 5 3), Cochlear Nucleus 5 (n 5 3), and MED-EL OPUS 2 (n 5 3) CI sound processors. In Experiment 2, seven of the original nine participants were tested. In Experiment 3, electroacoustic output was measured from a Nucleus 5 sound processor when coupled to the electromagnetically coupled Oticon Arc neck loop and electrically coupled Oticon R2. Data Collection and Analysis: In Experiment 1, participants completed a field trial with each FM receiver and three test sessions that included speech-recognition performance in noise and a subjective rating scale. In Experiment 2, participants were tested in three receiver-gain conditions. Results in both experiments were analyzed using repeated-measures analysis of variance. Experiment 3 involved electroacoustictest measures to determine the monitor-earphone output of the CI alone and CI coupled to the two FM receivers. Results: The results in Experiment 1 suggested that both FM receivers provided significantly better speech-recognition performance in noise than the CI alone; however, the electromagnetically coupled receiver provided significantly better speech-recognition performance in noise and better ratings in some situations than the electrically coupled receiver when set to the same gain. In Experiment 2, the primary analysis suggested significantly better speech-recognition performance in noise for the neck-loop versus electrically coupled receiver, but a second analysis, using the best performance across gain settings for each device, revealed no significant differences between the two FM receivers. Experiment 3 revealed

*Department of Speech and Hearing Sciences, University of North Texas Erin C. Schafer, 1155 Union Circle #305010, Denton, TX 76203-5017; Phone: 940-369-7433; Fax: 940-565-4058; E-mail: [email protected] The data in this study were presented in a podium presentation at the October 2011 Oticon Pediatrics Conference in San Antonio, TX, and in research posters at AudiologyNOW! 2012 in Boston, MA, and the November 2011 Texas Academy of Audiology conference in Richardson, TX. Funding for participants and the FM-system equipment was provided by an Oticon Pediatrics Research Initiative grant to the first author. The funds were used to compensate participants for their time, efforts, and mileage to the test center. The authors of this manuscript received no monetary compensation related to the study.

927

Journal of the American Academy of Audiology/Volume 24, Number 10, 2013

monitor-earphone output differences in the Nucleus 5 sound processor for the two FM receivers when set to the 18 setting used in Experiment 1 but equal output when the electrically coupled device was set to a 116 gain setting and the electromagnetically coupled device was set to the 18 gain setting. Conclusions: Individuals with contemporary sound processors may show more favorable speech-recognition performance in noise electromagnetically coupled FM systems (i.e., Oticon Arc), which is most likely related to the input processing and signal processing pathway within the CI sound processor for direct input versus telecoil input. Further research is warranted to replicate these findings with a larger sample size and to develop and validate a more objective approach to fitting FM systems to CI sound processors. Key Words: Cochlear implant, FM system, neck loop Abbreviations: ADRO 5 adaptive dynamic range optimization; ASC 5 autosensitivity control; BKB-SIN 5 Bamford-Kowal-Bench Speech in Noise; CI 5 cochlear implant; FM 5 frequency modulation; SNR 5 signalto-noise ratio

T

he ability to hear and understand speech in noisy environments remains a persistent problem for children and adults with cochlear implants (CIs). These decrements in speech perception are seen both behaviorally and subjectively. For instance, in previous research, speech-recognition performance in noise compared to that in quiet decreases by an average of 35 or 45% for children and adults with CIs, respectively (Schafer and Thibodeau, 2003, 2004; Firszt et al, 2004). Further, CI users report that understanding speech in background noise is one of their most difficult listening situations (Schafer and Thibodeau, 2004; Noble et al, 2008). As children and adults consistently experience noisy environments in their daily lives, at school, work, and home, these findings are cause for concern. Previous research has suggested that use of personal frequency modulation (FM) systems, which enhance the signal-to-noise ratio (SNR) to the CI, provide significantly better speech-recognition performance in noise over the CI alone and over other types of FM systems (Schafer and Kleineck, 2009). Specifically, in comparison to the CI alone, the personal, electrically coupled FM systems improve speech-recognition performance in noise in children by an average of 38% and in adults by an average of 44% (Schafer and Kleineck, 2009; Wolfe et al, 2009). Overall, these results indicate that FM systems significantly improve speech-recognition performance in noise, but more importantly, the FM system allows users to perform at a level that is often comparable to their performance in a quiet condition. These substantial improvements in speech-recognition performance occur because the primary talker’s signal is transmitted directly to the listener’s ear via FM radio waves. The FM transmitter, which is worn by the primary talker, sends the speech from the primary talker on a specific carrier frequency to an FM receiver, which connects to a CI sound processor with cables, specialized earhooks, adaptors, or battery covers. The FM receiver detects the incoming FM carrier frequency and demodulates the signal to decode the embedded signal of interest. The sound processor then determines

928

an audio-mixing ratio to control the emphasis of input from the FM system relative to the input from the sound processor microphone. The audio-mixing ratio may provide equal emphasis from each signal, as seen with the 50/50 and 1:1 mixing ratios used by Advanced Bionics and Cochlear Corporation, respectively, or may emphasize the FM signal by attenuating the signal from the sound-processor microphone by approximately 10 dB, as seen in 30/70 or 3:1 mixing ratios. The mixing ratios are either programmed into the CI sound processor by the audiologist or accessible to the user on the remote control of some devices (e.g., Nucleus 5 Remote Assistant). As cited above, there are several studies to support the benefits of electrically coupled FM receivers. Modern CI sound processors, however, may also couple to an FM system electromagnetically using a neck-loop FM receiver. Similar to an electrically-coupled receiver, a neck-loop FM receiver demodulates the signal from the FM transmitter. At this point, an electromagnetic signal is sent through a neck loop that is worn around the user’s neck, which is detected by the built-in telecoil of the CI sound processor. The electromagnetic signal is then converted to an electrical signal by the CI sound processor; similar to the electrically coupled receiver, the processor determines a specific audio-mixing ratio for the signal from the FM receiver as well as the signal from the CI sound-processor microphone. Transmitting the FM signal via the telecoil of CI sound processors has several advantages over electrically coupled receivers. First, a neck-loop receiver allows for diotic FM input into two separate devices (i.e., those using bilateral CIs or bimodal stimulation, which is one CI and a telecoil-enabled hearing aid on the nonimplant ear). Therefore, this binaural arrangement reduces the cost of an FM system in comparison to one electrically coupled FM receiver for each device. Second, the user does not require any special adaptors, earhooks, or battery covers to couple a neck-loop FM receiver to a CI sound processor. Finally, most neckloop FM receivers provide a visible volume control for

FM Systems and Cochlear Implants/Schafer et al

the user. Changes in the FM-receiver volume will not influence the preprogrammed audio-mixing ratio for the CI, but it will increase the strength of the signal from the FM receiver as well as the perceptual loudness of the FM system to the user. The only issue with using this type of receiver is the limited evidence to support its use with CIs and the fact that neck-loop FM receivers are not currently recommended for CI users in guidelines published by the American Academy of Audiology (Academy, 2008). Since the publication of the Academy (2008) guidelines, two recent studies have examined the potential benefit of neck-loop FM systems (Schafer et al, 2013; Wolfe et al, 2013). In the first study, Schafer et al (2013) examined performance with a neck-loop system relative to the CI alone. The 14 participants in this study, who were using Cochlear Freedom sound processors, showed significant improvements in speechrecognition performance in noise by 12 dB and in subjective ratings when listening at home, work/school, or in social environments when using a neck-loop FM receiver (Oticon Arc receiver with T20 transmitter) relative to the CI alone. In the second study, Wolfe et al (2013) compared performance between individuals using Freedom or Nucleus 5 sound processors while they used neck-loop and electrically coupled FM systems. Results of this study suggested that adults using Cochlear Freedom sound processors had significantly poorer speech-recognition performance in noise with a neck-loop FM receiver (Phonak MyLink1) in comparison to performance with an electrically coupled receiver (Phonak MicroLink Freedom) by an average of 25% at a 65A noise level and by 39% at a 75 dBA noise level (sentences presented at 65 dBA at the listener’s head and 85 dBA at location of transmitter microphone). However, when these same adults upgraded to the Nucleus 5 sound processor (CP810, Build C) with the most current front-end signal processing, there was no significant difference (i.e., 11% difference at 65 dBA noise level and 19% difference at 75 dBA noise level) in speech-recognition performance in noise between the neck-loop FM receiver (MyLink1) and the electrically coupled FM receiver (Phonak MicroMLxS). In the Wolfe et al (2013) study, the authors attributed the performance differences between the Freedom and Nucleus 5 processors to the location of a specific type of front-end processing, known as the autosensitivity control (ASC), within the signal processing pathways of the two processors. The ASC reduced the sensitivity of the processor microphone in noisy situations, which provides an emphasis for signals arriving from the FM system. The latest firmware for the Nucleus 5 uses a two-stage ASC processing approach, whereas the Freedom only has a one-stage ASC approach. In the one-stage ASC approach used in the Freedom, ASC was applied before the audio mixing of FM and processor inputs for electrically coupled receivers but after the audio mixing

for telecoil inputs (i.e., for neck loop, ASC is only applied to the mixed FM and processor signal, which will not result in emphasis of the FM signal over the processor signal). As a result, performance is better with the electrically coupled receivers in comparison to the neck-loop receivers. In the two-stage ASC approach used in the Nucleus 5, the first stage of ASC aims to reduce the sensitivity of the microphone, in noisy environments, prior to mixing the FM and processor signals. The second stage of ASC aims to improve the SNR of the combined signal from the FM and processor microphone. Overall, the Schafer et al (2013) study suggested that there are significant benefits from neck-loop FM systems for users of Freedom processors; however, findings in the Wolfe et al (2013) study suggested that electrically coupled FM receivers may be even more beneficial than neck-loop FM receivers. In addition, the latter study showed that performance may vary across different types of CI sound processors. The two previous studies are limited to Cochlear CI sound processors, and the comparison study by Wolfe et al (2013) only included FM systems from one manufacturer. Further research is necessary to compare performance and subjective preferences between electrically and electromagnetically coupled FM receivers for users of various CI sound processors. As a result, Experiment 1 of the current study aimed to compare speech-recognition performance in noise and subjective ratings for adolescents and adults using one of three contemporary CI sound processors coupled to electromagnetically and electrically coupled FM receivers. The investigators hypothesized that (1) the use of FM systems electromagnetically and electrically coupled to CI sound processors would provide significant improvements in speech-recognition performance in noise and subjective ratings relative to the CI alone and (2) no significant differences in speech-recognition performance in noise or subjective ratings would be seen when comparing the two FM-system types. Following Experiment 1, two additional experiments were conducted to further investigate the results of the initial experiment. EXPERIMENT 1 METHODS Design A repeated-measures design was used for Experiment 1 for data collected in three test sessions. In each session, participants completed speech-recognition-test conditions in noise without and with two types of FM systems and a subjective questionnaire employing a scale rating about each device condition, which is described in detail below and shown in the Appendix. Participants In Experiment 1, the participants included nine adolescents or adults with unilateral or bilateral Advanced

929

Journal of the American Academy of Audiology/Volume 24, Number 10, 2013

Bionics Harmony (n 5 3), Cochlear Nucleus 5 (n 5 3), and MED-EL OPUS 2 (n 5 3) CI sound processors (Table 1). All of the CI sound processors had microphones that were built into the ear-level processors, and participants using the Nucleus 5 sound processor utilized directional microphones, which were activated by selecting the noise program during speech-recognition testing in noise. None of the participants were using a bimodal listening arrangement. Users of these three sound processors were recruited for inclusion in this study to represent the three most current processors on the market in the United States. The participants with Nucleus 5 processors were using either the Build B or Build C devices; however, coupling of the FM receiver to the processor did not differ between these two builds. Adequate power was expected for this study given the use of a within-subjects design and because smaller sample sizes have been used in previous FMsystem investigations with multiple test conditions (Schafer and Thibodeau, 2004; Schafer et al, 2009).

transmitter) and noise, originating mostly from other children, occurring behind the listener. When the FM system was in use during testing, the FM-transmitter microphone was suspended in front of the single-coned loudspeaker at a distance of 15.2 cm (6”). This loudspeaker and equipment arrangement has been used in multiple peer-reviewed studies and was able to measure the expected significant benefit from FM systems (e.g., Schafer and Thibodeau, 2003, 2004, 2006; Schafer et al, 2009). Stimuli were calibrated using the calibration track on the CD and a sound level meter (LarsonDavis 824). During the four-week trials, each participant was loaned an Oticon Amigo T30 FM transmitter and either an Oticon Amigo Arc neck-loop FM receiver or an Oticon Amigo R2 receiver (Fig. 1). In Experiment 1, both receivers were set to the manufacturer-recommended 18 gain setting. Participants with bilateral CIs were loaned two R2 receivers to use during testing and during the trial period with the R2.

Test Room and Equipment

Speech-Recognition Stimuli and Subjective Scale

Speech-recognition testing was conducted in a classroom that was 4 m wide by 4.9 m long (13’2” 3 16’1”) with a volume of 52.9 mm3 in order to simulate listening in a real environment. In this classroom, the unoccupied noise level, averaged across eight locations, was 46.9 dBA, and the average reverberation time was 0.33 sec across octave frequencies between 500 and 4,000 Hz as measured with a Larson Davis System 824 sound level meter. Participants were seated in the middle of the classroom during testing. Speech and noise stimuli were presented from a CD player with two detachable, single-coned loudspeakers (Sony CD Radio Cassette-Corder), which were located at 0 (speech) and 180 (noise) degrees azimuth relative to the listener’s head. These head-level loudspeakers were each 1 m (3.2’) from the participant in order to simulate preferential seating in a small classroom with the teacher located at the front of the room (i.e., the

The split-track compact disc (CD) of the BamfordKowal-Bench Speech in Noise (BKB-SIN) test (Etymotic Research, 2005) was used in Experiments 1 and 2 to estimate speech-in-noise thresholds at the 50% correct level (i.e., SNR-50). One randomly selected list pair, consisting of 16 to 20 sentences, was used for each test condition, and all stimuli and intensity changes were prerecorded on the CD. In the CI-alone condition, the sentences were presented at a constant intensity of 60 dBA while the noise decreased in 3 dB steps with SNRs beginning at 121 dB and ending at 26 dB. For the FM-system conditions, acoustic editing software (Syntrillium Software, 2003) was used by the investigators to create a modified version of the split track BKBSIN in order to avoid ceiling effects. These ceiling effects were expected for this study according to previous pilot data and investigations with FM systems (Schafer et al, 2009). On the modified tracks of the BKB-SIN, the

Table 1. Participant Demographic and Cochlear Implant (CI) Information Subject

Age

1 2 3 4 5 6 7 8 9 Average

13 65 13 17 23 58 62 59 53 45.4

930

Manufacturer: CI Processor (Ear)

Duration 1st/2nd CI (yr;mo)

Advanced Bionics: Harmony (L) Advanced Bionics: Harmony (R/L) Advanced Bionics: Harmony (R) Cochlear: Nucleus 5, Build C (R/L) Cochlear: Nucleus 5, Build B (L) Cochlear: Nucleus 5, Build C (L) MED-EL: OPUS 2 (R) MED-EL: OPUS 2 (R/L) MED-EL: OPUS 2 (L)

10;0 4;11/4;6 9;10 4;2/2;4 1;0 1;0 3;2 1;0/1;0 7;0 4;7/2;6

Prelingual/Postlingual

Receiving Aural (Re)habilitation

Prelingual Postlingual Prelingual Prelingual Postlingual Postlingual Postlingual Postlingual Postlingual

Yes No Yes Yes No No No No No

FM Systems and Cochlear Implants/Schafer et al

Figure 1. Oticon frequency modulation (FM) systems in study: (A) T30 transmitter, (B) Arc electromagnetically coupled receiver, and (C) R2 electrically coupled receiver.

intensity of the sentences was decreased by 9 dB resulting in a sentence-presentation level of 51 dBA. The noise in the FM-system conditions decreased in 3 dB steps with SNRs beginning at 110 dB and ending at 215 dB. These stimuli have been shown to be sensitive enough to detect differences between FM-system conditions in previous investigations (Schafer et al, 2013; Schafer et al, 2009). Participants rated the sound quality and perceptions about their CI alone and CI coupled to each the FM receivers with an examiner-developed questionnaire, the Auditory Performance and Satisfaction Scale (shown in the Appendix). The scale consists of a sevenpoint Likert scale and general questions. The 20 statements are related to hearing at home, work, or school, and in social situations; 13 statements related to satisfaction with devices; and several general and open-ended questions. The FM ratings were provided after a trial period with each FM receiver. This scale has been shown to be sensitive in previous investigations and was shown to be sensitive to the benefits of FM systems in everyday listening environments (Schafer et al, 2013). Procedures In Experiment 1, participants completed test sessions before and after a four-week trial with each type of FM receiver. At the first session, participants completed the Auditory Performance and Satisfaction Scale regarding their perceived performance with their CI alone, and their speech-recognition performance in noise was assessed in randomized no-FM and FM-system conditions with both receivers at the manufacturer-recommended 18 receiver gain. Following testing, participants used the FM system with a randomly selected receiver (Arc or R2) several hours a day during a four-week trial. Specifically, participants were asked to use the FM system during conversations, meetings, instruction at school, or religious services

by having the primary talker wear the transmitter with lapel microphone and by coupling the receiver to their CIs. In addition, participants were asked to use the audio cable that they were provided to couple the FM transmitter to their cell phones, televisions, and other audio devices (e.g., iPod, MP3 players) or to place the transmitter microphone near the external audio source (e.g., loudspeaker of the television). After the trial, users returned to the laboratory to complete speech-recognition testing in noise in no-FM and FM conditions with both receivers set to the 18 gain setting. The Auditory Performance and Satisfaction Scale was completed in reference to the FM system used during the trial. Participants were allowed to see their ratings from Session 1 to avoid issues with any false memories of the CI-alone ratings (Schafer et al, 2013). Participants then completed a second fourweek trial with the other FM receiver, returning at the end of the trial to complete a third test session. In the third session, speech-recognition testing and the Auditory Performance and Satisfaction Scale were once again completed with the second FM system. Speech-recognition performance in noise in the no-FM and both FM-system conditions was repeated in all three test sessions to examine the reliability and repeatability of the measurements from one test session to the next, given the 8 wk time frame for the study. EXPERIMENT 1 RESULTS Average Speech-Recognition Performance in Noise The average speech-in-noise thresholds in dB SNR across the three test sessions are shown in Figure 2. A repeated-measures analysis of variance (RM ANOVA) was conducted on the average speech-recognition results with the independent variables of device condition (no FM, Arc, R2) and test session (1, 2, 3). This analysis

931

Journal of the American Academy of Audiology/Volume 24, Number 10, 2013

found between test sessions. However, as shown by the standard deviations in Figure 3, the within-subject variability was large. These standard deviations represent the within-subject variability across the three test sessions, which would be small if each participant’s performance was similar across the three sessions. In particular, Participant 7 showed standard deviations ranging from 3 to 7 dB across the three conditions. Other participants (1, 6, 9) also showed large standard deviations ranging from 3 to 7 dB in the no-FM condition. Subjective Scale

Figure 2. Average signal-to-noise ratios in dB where participants obtained 50% of key words correct (dB SNR-50) in each device condition across the three test sessions in Experiment 1. Vertical lines represent one standard deviation.

The average results of the subjective scale are provided in Table 2. In over half of statements 1–20, one or both FM receivers were rated lower (better) than the CI alone. When comparing average ratings between the Arc and R2, over half of the participants rated the

revealed a significant main effect of device condition (F[2, 81] 5 29.1, p 5 .000005), no significant main effect of test session (F[2, 81] 5 1.1, p 5 .34), and no interaction effect between device condition and test session (F[4, 81] 5 .64, p 5 .64). Post hoc testing with the Tukey-Kramer multiple comparisons test suggested significant differences (p , .05) across all three device conditions, with best performance seen with the Arc, followed by the R2 and the CI alone. To ensure that the results were not influenced by a potential acclimatization effect with the FM systems, the data were reanalyzed with only the best (i.e., lowest SNR) from each test condition. According to this one-way RM ANOVA, the significant main effect of condition remained (F[2, 27] 5 23.7, p 5 .00002). The associated post hoc analysis showed significantly better performance in the two FM conditions relative to the no-FM condition (p , .05) and significantly better performance with the Arc over the R2 receiver (p , .05). Individual Speech-Recognition Performance in Noise Although there was not enough power or a sufficient number of subjects to support manufacturer as a factor in the RM ANOVA, each participant’s data were plotted according to manufacturer to examine potential trends. The speech-in-noise thresholds of each participant were averaged across the three test sessions and are shown in Figure 3. According to a visual analysis, it appears that most of the participants showed their best performance with the Arc over the CI alone, which is consistent with the post hoc results in the analysis of the average results. Specifically, seven participants performed best with the Arc, and two participants performed equally well with the Arc and R2 receivers. As described in the previous paragraph, there were no significant differences

932

Figure 3. Individual signal-to-noise ratios in dB where participants obtained 50% of key words correct (dB SNR-50) averaged across the three test sessions in Experiment 1 and grouped by cochlear implant manufacturer, Advanced Bionics (A), Cochlear (B), and MED-EL (C). Vertical lines represent one standard deviation.

FM Systems and Cochlear Implants/Schafer et al

Table 2. Participant Ratings in Various Listening Situations and Device Satisfaction on the Auditory Performance and Satisfaction Scale CI Alone Average (SD)

Arc Average (SD)

R2 Average (SD)

F Value

Hearing at Home

1.9 (1.5)

1.1 (1.2)

1.7 (1.6)

5.8

Hearing at Work/School

2.5 (1.4)

1.3 (1.0)

1.5 (1.2)

20.0

Social Situations

2.5 (1.5)

1.6 (1.0)

2.0 (1.4)

5.8

Overall Satisfaction: CI and FM Receiver Satisfaction: FM Transmitter

0.9 (1.3)

1.1 (1.4)

1.2 (1.3)

NA

1.1 (1.2)

1.4 (1.1)

Scale Category

.52 1.3

Probability Level .003*

Post Hoc Results

.60

Arc significantly better than CI alone and R2 Arc and R2 significantly better than CI alone Arc significantly better than CI alone NA

.26

NA

.000000* .004*

Note: F value and probability level determined with general linear modeling; post hoc comparisons conducted with Tukey-Kramer multiple comparisons test. CI 5 cochlear implant; FM 5 frequency modulation system. *Significant results.

Arc receiver as the most helpful. To determine any significant average differences for statements 1–20 at home, work/school, and in social situations, a general linear model was used, which accounted for the occasional missing response from a participant on one or two questions that did not apply to his or her life. According to the analysis, there was a significant main effect of test condition (statistics provided in Table 2). Post hoc analyses were done with the Tukey-Kramer multiple comparisons test. According to the post hoc analyses of the home ratings, the Arc provided significantly greater benefit ( p , .05) than the CI alone or the R2 receiver(s), with no significant differences between the latter two conditions ( p . .05). In the work/school condition, both of the FM receivers resulted in significantly better ratings ( p , .05) than those for the CI alone with no significant differences between the two receivers ( p . .05). For the social ratings, the Arc resulted in significantly better experiences ( p , .05) than the CI alone, but the ratings between the two FM receivers did not differ significantly ( p . .05). Average satisfaction ratings with each FM receiver (statements 21–28) and with the single FM transmitter used with the two FM receivers (statements 29–33) are also provided in Table 2. The average ratings for the CI alone as well as the two FM receivers were favorable across all three devices and, for the most part, were rated below 2 (satisfied or very satisfied). General linear modeling was conducted on the average ratings from each condition. According to this analysis, ratings were equally favorable among the CI alone, the Arc, and the R2 because no significant differences were detected in the analysis. The analyses for statements 29 through 33 also suggested favorable ratings about the FM transmitter, with no significant differences between ratings provided after the trial periods with the Arc and R2. The general questions at the end of the questionnaire revealed that the Arc was used by the participants an average of 3.6 hr per day (SD 5 3.6) while the R2

was used an average of 3.3 hr per day (SD 5 1.7). Most participants did not answer the question about how much they would pay for the devices because they were unsure of their answers or they would not purchase the device. However, three participants stated that they would pay an average of $483 for the Arc and $1383 for the R2. When asked to list the situations where the receivers were most helpful and what they liked most about the system, the participants provided the answers shown in Table 3. What they liked least about the system is provided in Table 4. The general questions about the transmitter yielded favorable comments including that it was easy to use, had good sound and range, and worked well with the TV and computer. However, one participant stated that the loudness from the transmitter was not high enough (which was more likely related to one or both receivers), and another participant did not believe that the transmitter improved the sound clarity. EXPERIMENT 2 RATIONALE AND METHODS

E

xperiment 2 was conducted to examine the potential influence of FM-gain settings on the speechrecognition results in Experiment 1. A repeated-measures design was also used for Experiment 2 where participants were tested at different FM-gain settings. Participants. Seven of the original nine participants, who were interested in coming back in for additional testing, were included in Experiment 2 (Participants 1, 2, 3, 4, 5, 7, 9). Test Room and Equipment. The same test room and equipment were used in Experiment 2 with the exception of the gain settings on the receivers. In this experiment, participants were tested with 18, 112, and 116 gain settings. These receiver-gain setting changes were made and confirmed using the T30 transmitter. Speech-Recognition Stimuli. The same modified BKBSIN stimuli that were used in Experiment 1 were used for this experiment.

933

Journal of the American Academy of Audiology/Volume 24, Number 10, 2013

Table 3. General Comments from Participants regarding Environments Where the Systems Were Most Helpful and What They Liked Most about the FM Receivers Participant 1 2 3 4 5 6 7 8 9

Arc

R2

Family gathering, big group and loud, clear, easy to use Church, computer A-V class, easy to use, clarity At school when in group or teacher at distance, easy to use, comfortable, battery life, volume control, clear Watching movies, hearing people at a distance, good volume Watching TV, in car, direct sound from speaker and person speaking Outside when walking, computer, clear, cuts out static, TV Dinner table, noisy bar, clarity, range TV, easy to use, quality of technology TV, church, size, clear

Quiet listening situations, not large group and clear Shopping with wife, TV, sound quality Car, one-on-one situation, not too visible, synced easily

Procedure. In Experiment 2, seven of the original nine subjects returned speech-recognition testing in noise with the Arc and R2 receivers set to three receiver-gain settings (18, 112, 116) tested in randomized order. No field trial was included in this experiment. Experiment 2 The significant difference in performance (speech recognition; ratings) between the Arc (neck loop) and the R2 (direct-connect) receivers was unexpected by the investigators given that the receivers were both manufactured by Oticon and were set to the same manufacturer-recommended gain setting (18). As a result, the second experiment was conducted with seven of the nine original participants to determine whether higher FM-receiver gain settings, particularly on the R2, would result in more equivalent speech-recognition performance between the Arc and R2 receivers. Average results of this experiment are shown in Figure 4. A two-way RM ANOVA was used to examine the independent variables of device condition (Arc, R2) and gain setting (18, 112, 116). Results suggested a significant main effect of FM system (F[1, 42] 5 8.0, p 5 .03), with the Arc condition showing superior performance. No main effect of gain setting (F[2, 42] 5 .54, p 5 .54) or interaction effect between device condition and gain setting (F[2, 42] 5 1.4, p 5 .24) was detected.

N/A Not helpful Conversation, TV, computer, sound quality, easy to use Was not helpful at all Size TV, no more captions, clear sound in quiet

Because there was substantial variability in the average data shown in Figure 4, the investigators determined each participant’s best (i.e., lowest SNR) performance in the two FM-receiver conditions through visual inspection of the data. According to this visual inspection, the best performance for the Arc and R2 often resulted at different gain settings, within and across individuals. As a result, a second RM ANOVA was conducted with the best gain setting for each listener and condition to determine if significant differences would be detected between the two FM receivers. This analysis showed no significant main effect (F[1, 14] 5 3.4, p 5 .12) of FM receiver. EXPERIMENT 3 METHODS

G

iven the unexpected findings in Experiments 1 and 2, the goal of Experiment 3 was to determine the feasibility of conducting objective, electroacoustic measures with one of the CI sound processors in this study while coupled to the two FM systems. Thibodeau et al (2005) proposed a method to measure the microphone output of previous-generation Cochlear sound processors that were electrically coupled to FM receivers through the use of Cochlear monitor earphones, which are normally used to listen to and verify the function of the CI microphone. In addition to hearing output from the CI processor, these monitor earphones also allow a normal-hearing person to hear the output from the FM

Table 4. General Comments from Participants regarding What Participants Liked Least about the FM Receivers Participant 1 2 3 4 5 6 7 8 9

934

Arc

R2

Visibility to others, sometimes too loud Nothing Nothing Bulky, intermittent, static Constant humming, distance limitation Too many pieces for cell phone Dampness causes problems, battery life N/A Finding the right channel, needs connection capability with office phone

Cannot hear other people talking Hard to get hooked up, feel tied down No volume control, some static N/A Drains battery, not significantly louder Mute on receiver Size, range, lack of volume, cut out, hissing Discrimination of microphone Light located on receiver

FM Systems and Cochlear Implants/Schafer et al

Figure 4. Average signal-to-noise ratios in dB where participants obtained 50% of key words correct (dB SNR-50) in each device condition and across three receiver-gain settings in Experiment 2. Vertical lines represent one standard deviation.

system. Using a standard hearing-aid test box and a HA-1 coupler, Thibodeau et al measured and compared the output from Cochlear monitor earphones with inputs to (1) the CI sound processor alone and (2) the CI sound processor electrically coupled to an FM system. The measurements were conducted to determine effects of gain settings on the FM receiver and sensitivity settings on the CI sound processor with the goal of achieving a 10 dB FM advantage. Since this study, the Academy published guidelines for fitting FM systems and suggested that, in hearing aids, the FM gain should be set such that equal output (i.e., transparency) is measured from the hearing aid and FM system. Transparency is achieved when equal inputs of 65 dB SPL to the CI and FM microphones result in equivalent outputs across the frequencies of 750, 1000, and 2000 Hz. Despite the fact that FM systems are expected to provide higher outputs to the hearing aid or CI given the placement of the FM transmitter microphone (3–6 in from talker’s mouth) relative to the hearing aid or CI microphone (3–6 in), equal inputs are used for the test procedure because it allows for normalization of settings between the FM and hearing aid/CI; this procedure avoids issues with differing compression characteristics across devices. However, the actual SNR received by the listener is expected to be better than what is measured electroacoustically because typical inputs at the FM transmitter microphone are approximately 85 dB SPL. The current Academy (2008) guidelines recommend determining the actual benefit obtained with CIs coupled to FM systems through behavioral speech-recognition testing. It is possible that the electroacoustic approach of Thibodeau et al combined with the Academy guidelines could be used to facilitate a more objective approach to fitting FM systems on patients with CIs. In addition, the ability to measure and compare the output from

the Arc and R2 receivers coupled to a CI may further explain the findings in Experiments 1 and 2. To determine the feasibility of the aforementioned electroacoustic measurements with a contemporary CI sound processor, the examiners attempted the measures with the following equipment: Audioscan Verifit, HA-1 coupler, Cochlear Nucleus 5 (Build C) sound processor, Nucleus 5 Monitor Earphones, Oticon T30 transmitter, and Arc and R2 receivers set to the 18 gain setting. Using a 65 dB SPL standard signal on the test box, the outputs from the monitor earphones (in dB SPL) were measured with input provided to the CI alone, CI coupled to Arc, and CI coupled to R2. In the CI-alone condition, the sound processor was placed in the text box. When the CI was coupled to the Arc, the examiner wore the Arc neck loop and the Nucleus 5 on her ear, and the active FM transmitter microphone was placed inside the hearing aid test box. When the CI was coupled to the R2, the Nucleus 5 was placed in a sound-attenuating chamber (no input), and the FM transmitter microphone was, once again, placed inside the hearing aid test box. Outputs of each measurement will be provided in the results section. EXPERIMENT 3 RESULTS

T

he electroacoustic outputs in SPL of the CI alone and CI coupled to the two different FM receivers were plotted as a function of frequency in Figure 5. At a 18 setting, the output from the R2 receiver was approximately 10 dB lower than the output from the CI alone or the output from the CI coupled to the Arc. In contrast, when the gain of the R2 was increased to a 116 setting, the output was transparent or equivalent (6 3 dB difference in output averaged across the three frequencies) to the CI alone and CI coupled to Arc. DISCUSSION

T

he following sections will provide an overview of the experiments as well as a comparison of the results to previous studies on performance of individuals with CIs. Following these sections, limitations of the study and areas of future research will be addressed. Experiment 1 Until users of CIs experience substantial improvements in speech-recognition performance in noise from improved CI technology, FM-system research for this population is critical given its potential to enhance performance in real-world environments. Much like previous investigations (Wolfe et al, 2013; Schafer et al, 2013), the results of the present study support the use of electromagnetically and electrically coupled FM systems for significantly improving speech-recognition performance in noise and communication abilities in

935

Journal of the American Academy of Audiology/Volume 24, Number 10, 2013

Figure 5. Output in dB SPL at 750, 1000, and 2000 Hz from a Nucleus 5 cochlear implant (CI) sound processor and from the processor coupled to the Arc and R2 receivers at varying gain settings.

everyday environments when compared to performance with the CI alone. The participants in the present study achieved equivalent improvements when compared to participants with Cochlear Freedom CI sound processors in a previous study that evaluated speech-recognition performance in noise with a neck-loop receiver (Schafer et al, 2013). Specifically, in the previous study, participants showed an average improvement (i.e., Arc minus no FM) in speech-recognition performance in noise of approximately 11.4 dB (SD 5 4.4). In the present study participants improved by an average of approximately 11.5 dB (SD 5 6.4) across the three test sessions when using the Arc neck loop. When examining the data with the electrically coupled FM receiver, the R2, results were slightly poorer than results from a previous investigation that included participants with previous generation Advanced Bionics processors (Schafer et al, 2009). In the previous Schafer et al study, the average improvement from the electrically coupled FM receiver (at the highest programmable gain setting) was 16 dB. Participants with previous generation Cochlear processors were also included in the Schafer et al (2009) study, but their performance was most likely poor because certain input-processing parameters (autosensitivity [ASC] and advanced dynamic range optimization [ADRO]) were not utilized during this study (Schafer et al, 2009). A subsequent study determined that the activation of ASC and ADRO was necessary in order to obtain maximal benefit from electrically coupled receivers (Wolfe and Schafer, 2008). In comparison to the 16 dB improvement in the previous study, participants in the present study only improved by 6.7 dB when using the R2 compared to the CI alone (SD 5 5.6). The major methodological difference between the previous and present study is the inclusion of newer model versus older model CI sound processors. Newer sound processors, such as the Freedom and Nucleus 5, likely contain more inputprocessing parameters that attempt to improve the signal

936

before it is sent to the internal portion of the CI. It is possible in some cases (e.g., the one-stage ASC in the Freedom; Wolfe et al, 2013) that this input processing negatively affects the input from one type of FM system (neck loop) relative to another type (electrically coupled). On average, the subjective Auditory Performance and Satisfaction Scale revealed positive ratings for the CI alone as well as the two FM receivers. At home, the Arc resulted in significantly better ratings than the CI alone or the CI coupled to the R2 receiver. In work and school situations, the participants rated the two FM receivers similarly and significantly better than the CI alone. In social situations, the Arc was rated significantly higher than the CI alone but not different from the R2. Although the questionnaire yielded varied results, the Arc appeared to have a slight advantage over the R2 and certainly over the CI alone. These results correspond to the significant difference in speech-recognition performance in noise between the two receivers. In addition, the subjective ratings suggested a slight advantage for the Arc at home and in social situations or for both FM receivers at work and school as compared to the CI alone. Satisfaction ratings for the CI alone and two FM receivers were highly positive and did not yield any significant differences. Two findings in this study were unexpected. First, the within-subject variability (SD) in speech-recognition performance in noise, shown in Figure 3, was particularly large across the three listening conditions. In the BKBSIN test manual, test-retest for adults with CIs at the 95% confidence interval for one list is expected to be approximately 4.4 dB. In this study, however, many of the standard deviations from the participants were greater than the expected 4.4 dB. It is important to remember, however, that this study used spatially separated speech and noise (S0/N180) while the normative data were collected with spatially coincident speech and noise (S0/N0). Therefore, it may be possible that some participants received more benefit from the spatial separation of the sound sources than others. The exact cause of this variability was unknown; however, several factors could have contributed including etiology, duration of implant use, use of single list pairs, and typical variability in noise. Etiology of hearing loss may have caused variability in two participants who were diagnosed with Me´nie`re’s disease, which is known to result in performance fluctuations when vestibular symptoms arise (Lustig et al, 2003). The investigators noted that one participant had to reschedule a test session due to severe vestibular disturbance. Duration of implant use may also have contributed because, as shown in Table 1, three participants had only been using their implants for durations of 1 yr. According to previous research, performance in adults may increase until approximately 2 yr of implant use (Oh et al, 2003). Therefore, some variability could have been due to improvements in speech-recognition performance in noise over the time period of the study. Third, the

FM Systems and Cochlear Implants/Schafer et al

use of only one list pair from the BKB-SIN test may have resulted in variability. It is possible that more stable results would have been measured across the three test sessions if two or three list pairs were used per condition rather than the one list used per condition. Finally, it is possible that adults with CIs have variable scores in noise when they are tested several weeks apart. To our knowledge, there is no published research to specifically examine the test-retest variability across a 2 mo period. However, similar levels of variability were found on one of our previous investigations using single lists of the BKB-SIN (Schafer et al, 2013). In this previous study, the within-subject standard deviations across two noFM conditions ranged from 0.71 to 7.7 dB, and two FM-system conditions ranged from 0.0 to 8.5 dB. Testing a participant in noise at his or her threshold is a challenging task, and it is possible that this type of testing produces more variable results than fixed-intensity testing. Fixed-intensity testing is conducted at suprathreshold levels and yields a percent correct score. Previous research has shown that fixed-intensity testing may be more sensitive to detecting binaural benefits (e.g., binaural summation, squelch, and the head-shadow effect) (Schafer et al, 2011); therefore, the same findings may apply to speech-recognition testing in noise with FM systems. Further research is necessary to evaluate this hypothesis. The second unexpected finding was related to the significantly better performance for the electromagnetically coupled Arc receiver compared to that for the electrically coupled R2 receiver. In theory, a direct electrical connection should provide a more direct signal than the neckloop FM receiver, which might be affected by telecoil strength or the orientation of the telecoil in the CI sound processor. Following these findings, Experiment 2 was added to the investigation to determine if performance with the R2 relative to the Arc could be improved by increasing the gain of the receiver. Experiment 2 According to the statistical analysis, the Arc receiver continued to provide superior performance over the R2 receiver across three different receiver-gain settings; however, there were not significant differences across the 18, 112, and 116 gain settings. Examination of individual data across manufacturers did suggest a trend in six of seven participants. These six participants had fairly stable performance across the gain settings with the Arc receiver, and all six had improved speech-recognition performance in noise as the gain levels were increased in the R2 receiver. The remaining participant (4) showed worsening performance as the gain was increased in both the Arc and the R2. Because it appeared that different gain settings for the Arc and R2 resulted in the best performance, a second analysis was conducted. The results of this analysis showed that use of the single data point (at any

gain setting) for each participant resulted in no significant difference between the Arc and R2. These findings are critical because they address the performance differences in Experiment 1. In the first experiment, both receivers were set to the 18 setting; however, as shown in Experiment 2, the 18 setting is not adequate for some participants, particularly when using the electrically coupled R2 receiver. Based on results in Experiments 1 and 2, it is clear that different FM-receiver-gain settings may be necessary across contemporary CI sound processors coupled to FM systems. The FM-system manufacturers often determine their default receiver-gain settings by what works best for their own hearing instruments. However, hearing aids and CIs likely do not have the same input sensitivities. Therefore, it is not surprising that variations in performance occurred between the two types of FM systems and across the processors from the three CI manufacturers. These results also highlight the importance of behavioral testing to determine the most appropriate gain setting for a particular patient. Because subjective measures are not always feasible with young children or difficult-to-test populations, and as shown in this study, subjective ratings may not reveal the significant differences found in speech-recognitionin-noise measures, an objective approach to determining an appropriate FM-gain setting is warranted. Experiment 3 In theory, if an FM receiver is transparent, it would provide the same output as the CI alone given the same input. However, at the 18 setting used for the study, this was not the case for the R2 (Fig. 5). Assuming that the electroacoustic measurements are valid, the output differences explain the performance differences measured in Experiment 1 when using the two FM receivers at the 18 gain setting, at least for the Nucleus 5 users. Based on previous research, better performance would be expected for a receiver that provides substantially greater output than another (Schafer et al, 2009). As a result, the investigators hypothesize that more equivalent speech-recognition performance would be obtained for the Arc and R2 if the Nucleus 5 was coupled to the Arc set to a 18 gain and the R2 set to a 116 gain. For the most part, this was shown in Experiment 2 because, in the second analysis, five of seven participants did best with a 18 gain setting on the Arc while five of seven participants did best at the 112 or 116 setting on the R2. Furthermore, when the best performance was used in an analysis to compare the two receivers, no significant differences were found. Despite these findings, a follow-up study with a larger group of sound processors is needed to further examine the proposed electroacoustic-test protocol. Study Limitations There were several limitations to the present study. First, there was a small sample size. Although repeated-

937

Journal of the American Academy of Audiology/Volume 24, Number 10, 2013

measures designs allow for adequate power with smaller sample sizes, a larger sample would be helpful in generalizing these results to the population of individuals using CIs. Second, the variability in the speech-recognition-innoise performance measured across the sample may also have been reduced if the participants were more homogeneous in terms of the ear of implantation, unilateral and bilateral implantation, type of sound processor, duration of deafness, duration of CI use, and age at severe to profound hearing impairment. Again, the use of a within-subjects design avoided confounds due to participant differences, but as a whole, the group results were variable. Third, only two FM systems were used, and the systems were from the same manufacturer. Different performance and electroacoustic outputs may be found with electrically and electromagnetically coupled FM systems from other manufacturers. A follow-up study is currently underway that examines a wider range of FM-device manufacturers. CONCLUSION

T

he results of Experiment 1 suggest that both FM receivers provided significantly better speechrecognition performance in noise than the CI alone; however, the electromagnetically coupled receiver provided significantly better speech-recognition performance in noise than the electrically coupled receiver. Subjective findings supported the behavioral data, with a slight advantage of the Arc receiver relative to the R2 receiver or CI alone. In Experiment 2, increases to the gain on the FM receivers did not resolve differences between the two types of FM receivers. However, a second analysis of the best performance across the three FM-gain settings yielded no significant differences between the receivers. These findings suggest that different FM-gain settings may be required on different FM receivers, even from the same manufacturer, in order to achieve optimal performance. In Experiment 3, electroacoustic-test measures were attempted to examine output differences between the two types of FM receivers when both types were compared at the same gain setting. These initial measures suggested that the output from the electromagnetically and electrically coupled FM receivers, which were set to the same gain level of 18, did not produce the expected equal output in the Nucleus 5 processor. Instead, the gain on the electrically coupled receiver had to be increased by 8 gain units to match the output of the electromagnetically coupled receiver. In conclusion, in order to provide a more objective approach to fitting FM systems to CIs, further research is warranted to determine the feasibility and validity of electroacoustic-test measures on these devices.

Acknowledgments. Appreciation is expressed to Danielle Bryant for her assistance in preparing and editing this manuscript.

938

REFERENCES American Academy of Audiology (Academy). (2008) Remote Microphone Hearing Assistance Technologies for Children and Youth from Birth to 21 Years. www.audiology.org/resources/documentlibrary/ Documents/HAT_Guidelines_Supplement_A.pdf. Etymotic Research. (2005) Bamford-Kowal-Bench Speech in Noise Test. Elk Grove Village, IL: Etymotic Research. Firszt JB, Holden LK, Skinner MW, et al. (2004) Recognition of speech presented at soft to loud levels by adult cochlear implant recipients of three cochlear implant systems. Ear Hear 25(4):375–387. Lustig LR, Yeagle J, Niparko JK, Minor LB. (2003) Cochlear implantation in patients with bilateral Me´nie`re’s syndrome. Otol Neurotol 24(3):397–403. Noble W, Tyler R, Dunn C, Bhullar N. (2008) Hearing handicap ratings among different profiles of adult cochlear implant users. Ear Hear 29(1):112–120. Oh SH, Kim CS, Kang EJ, et al. (2003) Speech perception after cochlear implantation over a 4-year time period. Acta Otolaryngol 123(2):148–153. Schafer EC, Amlani AM, Paiva D, Nozari L, Verret S. (2011) A meta-analysis to compare speech recognition in noise with bilateral cochlear implants and bimodal stimulation. Int J Audiol 50 (12):871–880. Schafer EC, Huynh C, Romine D, Jiminez R. (2013) Speech recognition and subjective perceptions of neck-loop FM receivers with cochlear implants. Am J Audiol 22(1):53–64. Schafer EC, Kleineck MP. (2009) Improvements in speech-recognition performance using cochlear implants and three types of FM systems: a meta-analytic approach. J Educ Audiol 15:4–14. Schafer EC, Thibodeau LM. (2003) Speech recognition performance of children using cochlear implants and FM systems. J Educ Audiol 11:15–26. Schafer EC, Thibodeau LM. (2004) Speech recognition abilities of adults using cochlear implants with FM systems. J Am Acad Audiol 15(10):678–691. Schafer EC, Thibodeau LM. (2006) Speech recognition in noise in children with cochlear implants while listening in bilateral, bimodal, and FM system arrangements. Am J Audiol 15(2):114–126. Schafer EC, Wolfe J, Lawless T, Stout B. (2009) Effects of FMreceiver gain on speech-recognition performance of adults with cochlear implants. Int J Audiol 48(4):196–203. Thibodeau L, Schafer E, Cox S. (2005) Electroacoustic evaluation of FM systems with Nucleus speech processors. Poster presented at the 10th Symposium on Cochlear Implants in Children, Dallas, TX. Wolfe J, Schafer EC. (2008) Optimizing the benefit of sound processors coupled to personal FM systems. J Am Acad Audiol 19(8): 585–594. Wolfe J, Schafer EC, Heldner B, Mu¨lder H, Ward E, Vincent B. (2009) Evaluation of speech recognition in noise with cochlear implants and dynamic FM. J Am Acad Audiol 20(7):409–421. Wolfe J, Schafer EC, Parkinson A, et al. (2013) Effects of input processing and type of personal frequency modulation system on speech-recognition performance of adults with cochlear implants. Ear Hear 34(1):52–62.

FM Systems and Cochlear Implants/Schafer et al

Appendix A

939

Journal of the American Academy of Audiology/Volume 24, Number 10, 2013

Appendix A continued

940

Copyright of Journal of the American Academy of Audiology is the property of American Academy of Audiology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.