Clinical outcomes with the Kanso™ off-the-ear cochlear implant sound processor

Abstract

Objective: To investigate clinical outcomes and subjective ratings of the Kanso™ off-the-ear (OTE) cochlear implant sound processor. Design: Prospective, within-subject design investigating outcomes with a range of single and dual-microphone programmes for Kanso compared to conventional behind-the-ear (BTE) sound processors. Study sample: Twenty post-lingually hearing-impaired cochlear implant recipients who were experienced Nucleus® 5 or Nucleus® 6 BTE users. Results: No significant difference in performance was found for words in quiet or sentences in co-located noise between the Kanso and Nucleus 6 devices. For the moderately directional Standard programme, no significant difference was found for sentences in spatially separated noise between the Kanso and Nucleus 6 devices, but a performance decrement between 1.4 and 2.0 dB was found in highly directional and adaptive directional programmes. The default Kanso programme, SCAN, provided improvements of 6.9 dB over a single-microphone programme and 2.3 dB over the Standard programme in spatially separated noise. Participants rated Kanso significantly better than their own BTE processor on measures of comfort, look and feel, ease of use, music and overall hearing performance. Conclusion: Dual-microphone directional processing provides significant benefit over a single microphone for OTE processors. This study demonstrates clinical outcomes and acceptance of the Kanso OTE sound processor.

Key Words:

• Kanso
• cochlear implant
• clinical evaluation
• speech perception
• Nucleus 6
• off-the-ear

Introduction

The introduction of microprocessors and the transition from analogue to digital processing has resulted in the miniaturisation of modern day hearing devices, leading to less visible and vastly superior devices that offer flexible wearing options. These benefits have been seen in a range of hearing devices, including hearing aids, bone conduction devices and cochlear implants.

Hearing aids

Hearing aid design has successfully leveraged advanced technologies such as miniaturisation, digital processing and directionality which are less dependent on hardware alone to achieve small and cosmetically appealing devices. These come in a range of sizes and wearing locations including those worn behind the ear (BTE), in the ear or in the canal, and offer a range of output levels, localisation capabilities, and user driven preferences (Kochkin, 2010; Van den Bogaert et al, 2011). Of the range of technologies available in modern hearing aids, the only hearing aid feature consistently proven to improve speech understanding in noise is directional microphones (Bentler, 2012). The speech understanding benefits in noise with dual-microphone directionality have been especially well documented for BTE hearing aids, ranging from approximately 10% to over 30% depending on the speech test, room reverberation, and spatial noise configurations (Valente et al, 1995; Valente, 1999).

Bone conduction implants

Bone conduction devices are a surgically implanted system for treating hearing loss via direct bone conduction and provide significantly improved speech understanding in noise for those with conductive and mixed hearing losses (Lin et al, 2006; Stenfelt, 2005). The titanium implant is surgically implanted and is located approximately 50-70 mm behind the ear canal opening and approximately in line with the top of the pinna (Flynn, 2011), to which an external sound processor (transducer) is attached either magnetically or via an abutment. The microphone location on the bone conduction device processor is therefore typically in line with the top of the pinna and approximately 60 mm behind the ear canal opening compared to the microphone location of a conventional BTE hearing aid (Flynn, 2011; Stenfelt, 2005). While this location has positive advantages for surgical placement, vibratory sensitivity, and aesthetics, the microphone position results in higher sensitivity for sounds to the rear of the listener compared to those coming from in front, due largely to head shadow effects. To offset potential negative impacts on speech understanding, early bone conduction sound processors quickly implemented dual-microphone technologies capable of attenuating sounds from the rear and side of listeners while retaining sounds from the front (Flynn et al, 2011). Dual-microphone studies with bone-anchored devices have shown significant speech understanding improvements compared to a single-microphone processor (Flynn et al, 2011; Kompis et al, 2007; Oeding et al, 2010), but not in all cases (Oeding & Valente, 2013). Comparing omni-directional microphones to directional microphones, Kompis et al (2007) reported a 1.9 dB improvement testing with speech from the front and noise from the rear for participants with bilateral hearing loss. For single-sided deafened participants (occluding the better ear) Oeding et al (2010) found a 4.2 dB improvement for directional microphones compared to omni-directional microphones. For a conductive/mixed hearing loss group tested by Flynn et al (2011), a 2.1 dB improvement was reported. The smaller improvement demonstrated by Kompis et al (2007) could be due to limitations of the directional microphone setting because study participants were prevented from compensating for decreased microphone loudness by increasing their volume setting (Kompis et al, 2007). Other signal processing technologies available with bone conduction devices, such as frequency specific gain compression are thought to provide additional benefit in a posterior pinna location. A study by Pfiffner et al (2011) reported that the large benefit shown from upgrading users from the Baha Divino® device to the Baha® 3 (BP100) processor (both of which have directional microphones), was likely due to additional signal processing capabilities in the newer model device.

These early bone-anchored device studies demonstrate that a posterior auricular placement of the sound processor increases sensitivity to sound to the rear of a listener and may degrade performance in noise relative to the more anterior placement of a BTE hearing device. Introducing improved directionality with dual microphones has successfully compensated for reduced speech understanding in noise for bone-anchored device users. Additional signal processing capabilities may also provide a range of outcomes at this posterior placement.

Cochlear implants

The first cochlear implant (CI) sound processors available to profoundly deaf recipients in the late 1970s were bulky body-worn devices. Advancements in design enabled rapid progression to smaller, more portable devices which could more readily fit in a pocket, with a wire-connected microphone headset worn behind the ear. The introduction of smaller and more powerful microprocessors enabled further miniaturisation of the sound processor resulting in the commercial release of a BTE sound processor in late 1990s (Helms et al, 2001; Totten et al, 2000). The BTE processor included the microphone and signal processing electronics connected via a short cable to the transmitting coil worn over the implant site (Kompis et al, 2002). This processor configuration is still the most widely available one today. Dual-microphone directional processing was introduced into sound processors in 2005 with the release of the Nucleus® Freedom® BTE device which provided significant improvement in speech understanding in spatially separated noise (Spriet et al, 2007).

Recently, an off-the-ear (OTE) sound processor for CIs was introduced which integrates the coil, processor and microphones into a single unit worn over the implant site (Mertens et al, 2015; Wimmer et al, 2015). This sound processor placement is very similar to the posterior auricular placement of bone conduction implants. For this reason, the OTE placement would be expected to have similar positive aesthetics advantages, but also limitations from head shadow effects due to the microphone position, particularly for single-microphone processors.

In a study with ten single-sided deaf recipients, Mertens et al (2015) compared performance of this OTE processor (Rondo®) against a BTE device (Opus 2). No difference in outcomes between the OTE and BTE processors was found for free-field threshold, speech understanding and localisation assessments. This is probably because the subjects were tested with the non-test ear un-occluded, meaning advantage provided by their normal-hearing ear would easily obscure any subtle differences that may have existed between the processors.

Wimmer et al (2015) tested unilateral hearing performance with these same processors, but in subjects with profound bilateral hearing loss. In this study they found significantly poorer hearing performance with the OTE compared to the BTE sound processor when tested using a speech signal from the front and competing noise from the rear. The mean SRT score was 4.4 dB worse with the OTE processor, which is not surprising since the microphone location of the OTE is closer to the noise source relative to the BTE, and both processors used a single microphone with an omni-directional response. Physical and numerical modelling by the authors confirmed that the degree of observed differences in SNR between the processors was consistent with their expectations based on the differences in calculated head shadow and pinna effects resulting from the different microphone locations. The authors suggested that future studies should investigate the use of directional microphone technologies in OTE devices (Wimmer et al, 2015).

Kanso sound processor

The Kanso™ (model CP950) sound processor is the first OTE sound processor released by Cochlear Limited (Figure 1). The sound processor houses dual microphones, user controls and two Zinc-Air batteries together with the magnet and transmitting coil in a single unit. It is designed to be worn on the recipient’s head directly over the site of the internal CI. Kanso offers the same SmartSound iQ signal processing options as the Nucleus® 6 BTE sound processor including automatic sensitivity control (ASC), adaptive dynamic range optimisation (ADRO®), wind noise reduction (WNR) and background noise reduction (SNR-NR) (Mauger et al, 2012). Dual omni-directional microphones allow the full range of microphone directionality patterns as in Nucleus 6. These are a moderately directional microphone pattern called “Standard”, a fixed highly directional microphone called “zoom”, and an adaptive directional microphone called “Beam®” (Hersbach et al, 2012). The automatic scene classifier (SCAN) automatically detects a user’s listening environment, and selects appropriate technologies in each listening environment, including the selection of directional processing options (Mauger et al, 2014). It also supports wireless connectivity to a range of assistive listening accessories (Wolfe et al, 2016).

Figure 1. Photograph of Kanso (CP950, left) and Nucleus 6 (CP920, right) sound processors showing form differences and relative size.

This study

This study compared the preference and performance outcomes with a new dual-microphone OTE sound processor (Kanso) against the latest commercially available BTE Nucleus 6 (CP900 series) sound processor. Speech understanding was assessed in quiet as well as in co-located and spatially separated speech and noise. A self-report survey was used to assess user preference and performance ratings across a range of real-world listening situations.

Specific research questions for the study include:

1. Is there a performance difference in quiet between the Kanso and Nucleus 6 sound processor?

2. Is there a performance difference in co-located speech and noise (S0N0) between the Kanso and Nucleus 6 sound processor?

3. Does directional processing in the Kanso processor provide performance benefits over Standard directionality in spatially-separated speech and noise?

4. Does dual-microphone directional processing provide benefit over a single microphone in spatially separated speech and noise with the Kanso sound processor?

5. Are qualitative preference or perception ratings significantly different between Kanso and users’ own BTE sound processors?

Materials and methods

Research participants

Twenty-three research participants implanted with a Nucleus CI system were recruited in the study at a single site. The study was approved by the Royal Prince Alfred Hospital (Sydney, Australia) Human Research Ethics Committee (X14-0375). All participants in the study gave written informed consent. The inclusion criteria for all research participants were; at least 18 years old, implanted with either the Nucleus® 24, Nucleus® Freedom® or Nucleus® CI500 series implant, current users of the a Nucleus® 5 or Nucleus 6 BTE sound processor, at least six months CI experience after activation (with both implants for bilateral participants), native Australian English speaker, willing to participate and to comply with all requirements of the protocol, and able to score 30% or more at +15 dB signal-to-noise ratio (SNR) in an open-set speech recognition test with their CI alone, and no additional needs that would prevent participation in the evaluations. No in-kind or monetary incentive was offered or given for participating in this study. Biographical details for each participant are shown in Table 1. Four participants were bilaterally deaf, and were tested with their unilateral CI. Seven participants were tested as bilateral CI users. In these cases both ears were fitted with the Kanso and Nucleus 6 sound processors and both processors were used in testing. Nine subjects had some usable hearing in their contralateral ear being either single-sided deaf (SSD), or aided with a hearing aid (Bimodal users). In these cases, the contralateral device was removed and the contralateral ear was occluded as needed during speech recognition testing.

Table 1. Biographical data of study participants.

Participant	Gender	AGE (years)	Hearing mode	Own processor	Angle (Degrees)	Distance (cm)	Magnet strength
1	M	44	Bimodal	CP900	45	8	2
2	M	65	Unilateral	CP810	45	7	5
3	M	32	Unilateral	CP900	30	7	3
4	M	85	Bilateral	CP900	60	7	2, 2
5	F	78	Bimodal	CP900	45	7	1
6	M	30	SSD	CP900	45	8	6
7	M	53	SSD	CP900	60	6	6
8	F	49	Bilateral	CP810	52.5	8	2, 2
9	F	62	Unilateral	CP810	60	7	3
10	F	65	Bilateral	CP810	45	7.5	5, 4
11	F	57	Unilateral	CP810	60	9	2
12	F	57	Bilateral	CP810	37.5	6.5	1, 1
13	F	60	Bilateral	CP810	30	8	4, 1
14	F	56	Bilateral	CP810	52.5	8	4, 2
15	M	74	Bimodal	CP810	30	7	3
16	F	36	SSD	CP900	30	7	2
17	M	38	Bilateral	CP900	52.5	7.5	3, 3
18	M	76	Bimodal	CP900	60	8	4
19	M	73	Bimodal	CP900	45	8	2
20	F	35	Bimodal	CP900	60	6	3

The location of the Kanso sound processor is described as the elevation from horizontal as well as the distance relative to the ear canal. Participant BTE magnet strengths for unilateral and bilateral (Left, Right) participants.

Study design

The study used repeated measures, single-subject design in which subjects served as their own control. Before session one, three participants withdrew from the study. Twenty research participants were tested in sessions one and two and three.

CI recipients’ performance was compared between the new Kanso (CP950) sound processor and the Nucleus 6 sound processor. A range of SmartSound iQ programmes were tested in quiet and in noise. In all cases, ASC, and ADRO were enabled. Programmes listed in this paper are named after the directional microphone technology selected (Standard, zoom or Beam), or SCAN for the programme with automation. Kanso was additionally tested using a research-only programme that used the front microphone only, providing an omni-directional response (Single microphone), to demonstrate the performance of a single-microphone sound processor.

In a first session, participants using their BTE sound processor (either the Nucleus 5 or Nucleus 6 processor) were fitted with Kanso (Table 1). Participants were fitted with the Kanso default programme and one or two custom programmes depending on their own BTE programme usage. For users of the Nucleus 5 sound processor, upgrading to Kanso enabled access to wireless accessories for the first time. Participants wore the Kanso sound processor for an acclimatisation period of approximately three weeks. Three test sessions were conducted, spaced approximately three weeks apart. During speech perception testing with BTE programmes, participants were all fitted with a Nucleus 6 sound processor. At the end of the final session, recipients received a survey to rate the performance of the Kanso processor compared to their own BTE processor.

Test procedures in quiet and noise

In session one, both words in quiet and sentences in noise were used to compare outcomes between Kanso and Nucleus 6. Both sound processors used the automatic SCAN programme. Two lists of Australian CNC words (50 words per list) were presented at 50 dB SPL in quiet from in front of the listener. Two lists of open-set BKB-like sentences (20 sentences per list) spoken by an Australian female were presented at 65 dB SPL in four-talker babble noise (Dawson et al, 2011). Both the speech and noise were presented from a loudspeaker in front of the listener (S0N0; co-located noise). Depending on the listener’s response, the noise level was adjusted up or down by 4 dB for the first four sentences, and by 2 dB for the remaining sixteen sentences. Background noise was presented for 12 seconds before the first sentence. Following each noise level adjustment was a three second period of noise at the new level and a beep cue before the next sentence was presented. The adaptive test determines each participant’s speech reception threshold (SRT), defined as the SNR for 50% sentence understanding. Each SRT was calculated as the mean SNR for sentences five to twenty and also the SNR at which sentence 21 would have been presented based on the subject’s response to sentence 20. All speech and noise configurations had loudspeakers 1 m from the listener.

In session two, sentences in spatially-separated noise was used to compare between Kanso and Nucleus 6 with clinically available directional microphone programmes. In session three, sentences in spatially-separated noise was used to compare an extended range of directional microphone programmes with the Kanso sound processor. In both sessions speech was presented from in front of the listener, and individual babble noise maskers were dynamically and randomly presented through the seven loudspeakers located 30 degrees apart in the rear hemi-field (spatially-separated noise) (Hersbach et al, 2012). This spatially-separated noise condition was chosen to provide a more realistic simulation of a noisy listening situation such as listening to someone speaking in a café with other competing conversations happening to the sides and rear of the listener. The same adaptive protocol as in session one was used. The sound field produced by each loudspeaker was calibrated at the location of the listener, using 1/3 octave narrowband noise centred at 1 kHz presented at 65 dB SPL. The levels of the individual babble noise maskers were equally reduced so the total noise level was 65 dB SPL calibrated at the listening position without the listener present.

User rating survey

At the end of the study, a survey was provided for participants to rate the Kanso sound processor relative to users’ own BTE processor. Thirteen questions were posed using the format “Compared to your own sound processor, how do you rate the CP950 performance (including accessories)?”. Participants responded on a 5-point Likert scale with the ratings of “much better”, “somewhat better”, “no difference”, “somewhat worse”, and “much worse”.

Data analysis

Individual percent correct (words in quiet) or SRT scores (sentences in noise) with the Kanso and Nucleus 6 sound processors were obtained by averaging the performance of the two lists in a single test session for each programme. In all statistical tests, an alpha of 0.05 was used.

For test session one, a two-tailed repeated measures t-test was conducted for percent correct scores in quiet and SRT sentences scores in noise.

For test session two, a two-way repeated measures analysis of variance (ANOVA) was conducted with the main factors, “programme type”, and “processor type”. There were four levels within the “programme type” factor comparing Standard, zoom, Beam and SCAN, and two levels within the “processor type” factor comparing Kanso and Nucleus 6. Post-hoc Newman-Keuls comparisons were used for comparisons within the interaction factor, “programme type × processor type”, to evaluate the difference between Nucleus 6 and Kanso within each of the different programme types, as well as the benefits of directional microphone settings.

To determine if processor location affected performance outcomes, the distance between the ear canal and the implant magnet, and the angular elevation from the horizontal, using the ear canal as the vertex, were measured (see Table 1). A performance difference was calculated as the SRT of Kanso minus the SRT of Nucleus 6 for each programme. Correlation analysis consisted of a multiple linear regression analysis using the improvement as the dependant variable, and the distance and angle as two independent variables. For bilateral users, the left and right distance and angle measurements were averaged.

For test session three, a one-way repeated measures ANOVA on ranks was used to determine the effect of the main condition of “programme type” comparing Single microphone, Standard, zoom, Beam and SCAN. Post-hoc Newman-Keuls comparisons were used to determine differences between programme types.

Sentence scores were collected over a wide range of SNR levels in each session. These levels spanned the range of individual performance-intensity functions. It was therefore possible to calculate performance-intensity functions for each sentence list, and to convert from dB SRT scores to per cent correct scores (Dawson et al, 2013). The average between the maximum and minimum SRT scores of sentence lists was found for individuals in both sessions two and three. This score was then used as the conversion SNR for the individual’s test session. Percent correct scores were then calculated through the performance-intensity curve for each sentence list at the conversion SNR. Two per cent correct scores were averaged for each programme. This score represents individual per cent correct scores for each programme when tested at an SNR in the middle of their performance range.

Survey results were analysed using a one-sampled signed rank test, for a test mean comparison of “no difference”, to determine if there was a difference between Nucleus 6 and Kanso ratings.

Results

Speech testing in quiet and co-located noise

Session one results were analysed for words in quiet and sentences in noise. Group mean per cent correct scores in quiet were 52.4% and 54.8% for Kanso and Nucleus 6, respectively. The paired t-test did not show a significant difference between the programmes (p = 0.143) for words in quiet (Figure 3(A)). Group mean SRT scores in co-located four-talker babble noise were 4.0 dB and 3.7 dB for Kanso and Nucleus 6, respectively. The paired t-test did not show a significant difference between the programmes (p = 0.210) for sentences in noise (Figure 3(B)).

Processor testing

Session two results were analysed to assess processor performance, as well as programme performance. Group results for sentences in spatially-separated noise comparing Kanso and Nucleus 6 are shown in Figure 4. The two-way repeated measures ANOVA showed a significant main effect of “processor type” averaged across the four directional microphone types (F[1,19] = 14.4, p = 0.001). Post hoc Newman–Keuls comparison within processor showed significantly better SRTs for the Nucleus 6 processor compared to the Kanso processor (p < 0.001) with a mean difference of 1.1 dB. There was a significant main effect of “programme type” averaged across the two processors (F[3,57] = 88.7, p < 0.001). Post-hoc comparisons within programme showed significantly better SRTs for zoom, Beam and SCAN compared to Standard (p < 0.001). SCAN was significantly better than zoom (p < 0.05). No differences were found between zoom and Beam, or Beam and SCAN. The interaction effect (“processor type” ד programme type”) was significant (F[3,57] = 11.9, p < 0.01), indicating that the “processor type” effect was different across the various programmes and similarly that the “programme type” was different across the two processors.

Assessing the processor type within the Standard microphone comparison, post-hoc analysis showed that there was no significant difference in mean sentence perception between Kanso and Nucleus 6 (p = 0.242). Assessing the processor type within zoom, Beam and SCAN comparisons, post-hoc analysis showed that mean sentence perception was significantly better for Nucleus 6 compared to Kanso for zoom (p < 0.001), Beam (p < 0.001) and SCAN (p < 0.001) with a mean difference in SRT of 1.4 dB (19 percentage points), 2.0 dB (23 percentage points), and 1.4 dB (16 percentage points) respectively.

Location correlation with performance

Correlations between both distance and elevation angle of the Kanso processor and the difference scores between Kanso and Nucleus 6 were investigated. No significant correlations were found for any of the four programmes; Standard (r²=0.027, p = 0.790), zoom (r²=0.162, p = 0.222), Beam (r²=0.059, p = 0.595), and SCAN (r²=0.154, p = 0.242).

Programme testing

Session three results were analysed to assess programme performance with the Kanso processor. Group results for spatially-separated noise comparing Kanso programmes are shown in Figure 5. Group mean SRT scores of 2.3, -2.3, -3.9, -4.6 and -4.6 dB were found for Single microphone, Standard, zoom, Beam, and SCAN respectively (Figure 5). These scores corresponded to per cent correct scores of 12%, 59%, 75%, 79%, and 80% for fixed level testing. The one-way repeated measures ANOVA on ranks showed a significant main effect of programme (χ²[4,19] = 48.2, p < 0.001). Post-hoc pair-wise comparisons showed the Single microphone to be significantly poorer than all other programmes (p < 0.001). Post-hoc pair-wise comparisons found Standard to be significantly poorer than zoom, Beam and SCAN (p < 0.001). There was no significant difference found between zoom, Beam or SCAN.

Survey results

Survey results comparing user responses for Kanso against their own BTE processor are shown in Figure 6. Comparisons of “Overall Hearing Performance” (p < 0.05), “Listening to Music” (p < 0.05), “Comfort” (p < 0.001), “Look and Feel” (p < 0.001), “Learn to Use” (p < 0.05), and “Ease of Use” (p < 0.01) were all found to be significantly more favourable for Kanso compared to users’ own BTE processor. There was no significant difference found for any other survey questions.

Discussion

This study assessed the performance of the new Kanso OTE sound processor with respect to the Nucleus 6 BTE sound processor. Both sound processor models have the same microprocessor and SmartSound® iQ technologies, but differ in their form factor and wearing configuration.

Speech understanding discussion

Previous studies have not presented speech understanding results in quiet for OTE processors. However, Wimmer et al (2015) reported on a physical and numerical model which estimated that in the range of likely OTE positions on the head, a 10 degree increase in azimuth would result in approximately 1 dB attenuation for signals presented from the front. Given a previously reported bone anchored processor location of 50 mm BTE canal, there is an approximate 22 degree increase in azimuth for an OTE processor (Flynn, 2011; Stenfelt, 2005), producing an estimated 2.2 dB attenuation for this OTE location. The OTE placement therefore poses a challenge regarding both signal sensitivity, and directionality in spatially separated noise environments.

The Kanso processor was optimised to provide equal frontal sensitivity to a BTE sound processor. Additionally, it aimed to achieve a similar polar characteristic to the Standard directional processing of the BTE processor. When measured on KEMAR, the polar characteristics of the Kanso sound processor are shown to be very similar to the BTE polar characteristics (Figure 2(A)). This equivalent sensitivity is supported by the present study which showed no significant difference in scores between Kanso and the Nucleus 6 BTE in quiet or in co-located speech and noise (Figure 3), and no significant difference in spatially separated noise (Figure 4) with the Standard programme. These results confirm that dual-microphone processing works effectively in both sound processors.

Figure 2. Polar patterns for Kanso and Nucleus 6 sound processors while worn by a head-and-torso simulator (KEMAR). Zero degrees (upward) are taken as the direction the listener is facing. Processors were placed on the right ear. Top: For the dual-microphone Standard directional microphone programme. Bottom: For the dual-microphone zoom highly directional microphone programme.

Figure 3. (A) Group mean per cent word understanding in quiet. (B) Group mean speech understanding for speech in co-located four-talker babble. Error bars show the standard error of the mean.

Figure 4. Group mean speech understanding of the Kanso and Nucleus 6 sound processors for speech in spatially separated four-talker babble noise. Both processors were tested in four SmartSound iQ programmes. Error bars show the standard error of the mean.

Figure 5. Group mean speech understanding for the Kanso sound processor for speech in four-talker babble in spatially separated noise. Four SmartSound iQ programmes were used, as well as one research programme which used a single microphone only. Error bars show the standard error of the mean. Asterisks show significance with *** = p < 0.001.

Figure 6. Survey results showing user ratings of Kanso relative to the users’ own sound processor. The column on the left (> =) shows the proportion of responses rating Kanso equal or, better than the users’ own sound processor, as well as the total number of responses for each question. Asterisks show significance compared to “No Difference”; * = p < 0.05, ** = p < 0.01, *** = p < 0.001. Box plots show 25th, 50th and 75th percentiles. Error bars show the 5th and 95th percentiles.

The directional processing algorithm for Kanso included an important change from the BTE processor in order to compensate for its location on the head. Directional processing for zoom in Kanso was implemented with the null at 150 degrees instead of a null at 120 degrees as used in the Nucleus 6 BTE (Figure 2(B)). Mean results with the fixed highly directional (zoom), adaptive directional (Beam), and automatic scene classification (SCAN) programmes on Kanso, showed significant improvements in speech understanding compared with Standard directionality ranging from 2.3 dB (27 percentage points) to 3.1 dB (35 percentage points) (Figure 4). These improvements over Standard directionality are consistent with previous studies evaluating spatially separated noise performance (Hersbach et al, 2012). This result confirms that both OTE and BTE processors provide directional benefit in difficult spatially separated noise environments when dual microphones are activated. Group mean performance using zoom, Beam and SCAN programmes with Nucleus 6 did, however, show additional benefit (of between 1.4 dB to 2.0 dB) over that measured with Kanso in spatially separated noise. This is likely a result of the limited ability of directional microphones to fully compensate for different microphone locations when using fixed highly directional or adaptive directional processing. SCAN is an automatic programme, which detects the listening environment and automatically selects appropriate sound processing technologies. The clinical performance of SCAN (which predominantly classifies this noise test as “speech in noise”) was not found to be different to the performance of the Beam programme. This result is similar to previous Nucleus 6 testing since SCAN automatically enables Beam in the “speech in noise” scene (Mauger et al, 2014).

Previous CI OTE publications were limited to processors utilising only a single microphone providing an omni-directional response pattern. To enable comparisons with published outcomes from those studies, a single-microphone programme was implemented on the Kanso processor via research programming software. This programme used the front microphone only providing an omni-directional microphone condition for speech testing. The early bone-anchored device experience with a single microphone showed degraded performance in noise due to the more posterior location of the processor. With the use of dual-microphone directional processing, a mean listening improvement of 4.2 dB in noise was provided (Oeding et al, 2010). Directional microphone polar characteristics in this study were found to significantly improve listening from in front with Standard and zoom directionality (Figure 2). Clinical results with dual microphones using Standard directionality showed a mean 4.5 dB (47 percentage points) improvement over a single-microphone programme. A large mean 6.9 dB (68% points) improvement was found for dual microphones using SCAN over the single-microphone programme. This result is higher than reported by Hersbach et al (2015) with mean improvements of 5.6 dB and 4.4 dB for zoom and Beam programmes over an omni-directional microphone programme on a CI BTE sound processor. This is likely due to the more difficult noise environment in Hersbach et al (2015), which included noise locations both behind and in front of the listener, compared to only in the rear hemi-field in the current study. The current study results reinforce the large and clinically important improvements that directional microphone processing technologies provide for listening in noise, and the superiority of dual microphones over single-microphone programmes.

Discussion of user ratings

After transitioning from their own BTE processor to Kanso, and following a 2 week period of take home experience, study participants rated the relative performance of Kanso. It should be noted that it was not possible to blind the users to the processor being rated, given that Kanso was easily identified as the new device due to the different form factor and wearer placement. Users rated their Overall Hearing Performance as significantly better with Kanso. Given that ratings for listening in noise were not significantly different between the processors, we suspect additional conveniences with Kanso not available in some participants BTE processors (such as access to wireless accessories or SCAN) may have influenced the overall hearing performance ratings in favour of Kanso. It is reassuring that although testing in a sound booth revealed some differences between processors in spatially separated noise, participants did not report any functional differences in their real-world speech in noise experience as evidenced by the survey responses. Ratings for listening to music, however, were significantly more favourable for Kanso than for BTE processors. A possible explanation for this is that some participants used older generation BTE processors which, unlike Kanso and the Nucleus 6 BTE processors, did not include a SCAN programme to automatically identify music and select appropriate technologies for this scene.

Several expected benefits of switching from a BTE to an OTE processor are related to humanistic factors like comfort, usability and aesthetics. The survey revealed a significantly higher mean rating of comfort for Kanso compared to participant’s own BTE processor. Although the survey did not expand upon this to examine which specific factors were attributed to the increased comfort ratings, other OTE studies suggest it could include general improved wearing comfort, reduced skin pressure, and better comfort for those who wear glasses (Mertens et al, 2015; Távora-vieira & Miller, 2015). While the appearance of the Kanso processor is quite distinct and the form factor different to that of a BTE processor, survey ratings for the Look and Feel of Kanso rated higher than for the BTE processor.

Retention on the head is a consideration of an OTE sound processor given the lack of surrounding structures to which the processor may be easily anchored. Previous studies have reported results for OTE retention rates. Mertens et al (2015) reported on retention for a group of nine subjects, of which 44% reported the OTE processor unexpectedly falling off their head up to several times a day. In this study, somewhat surprisingly, no significant difference was found in retention ratings for Kanso compared to their own BTE sound processor despite the additional security of its position over the pinna. This may in part be due to the variability in use of retention accessories for Kanso. In the first Kanso fitting session of this study, almost half of the subjects (8/20) elected to use a retention accessory, however, this number dropped to 25% (5/20) by the end of the first 2 weeks of Kanso use as retention fears decreased and device confidence increased. Some subjects did comment that OTE retention was better than the BTE in some specific situations like lying down or bending over where the BTE would typically fall off much more easily. It is difficult to control for the variations in activity levels and other real-world lifestyle factors that impact directly on device retention. Nevertheless, there is likely to be specific situations in which retention with an OTE processor may be more successful than with a BTE processor and vice versa.

Sound processor comfort and retention are directly impacted by the magnet “Strength” (sometimes referred to as magnet size or magnet number) selected for the user. A range of magnet Strengths are offered with BTE sound processors to accommodate a range of user skin flap thicknesses, as well as user specific retention requirements. For clinical consistency, Kanso magnet Strengths have been designed to provide similar retention and comfort characteristics as BTE magnet Strengths, given they are not directly interchangeable across processor types. This harmonisation between OTE and BTE magnet Strengths accounts for differences in retention forces required due to the different physical attributes of the sound processors. Therefore in most cases, the magnet Strength used with a BTE and a Kanso sound processor are expected to be the same. To further improve wearing comfort, Kanso can be worn with the addition of a SoftWear™ pad. This thin foam pad is attached to the underside of the Kanso processor to more evenly distribute the pressure of the processor on the head. It also increases the distance between the coil and the implant. Similar accessories have been used by magnet retention bone vibratory devices such as the Baha® Attract System (Clamp & Briggs, 2014). In this current Kanso study, participants were provided with the same magnet Strength for their Kanso sound processor as they used with their BTE sound processor. It was explained that if there was a retention issues experienced that they should contact the study coordinator, and if there was a comfort issue that participants should add the SoftWear pad provided to the Kanso sound processor, and if the comfort issues persisted they should contact the study coordinator. Of the 20 participants in this study, 19 maintained the same magnet Strength as their BTE sound processor and one required a reduction from Strength 3 to Strength 2. At the end of the study, five participants had elected to use the SoftWear pad. These results suggest that initially fitting the same magnet Strengths as for their BTE sound processor is a suitable starting point for recipients upgrading, and that some users may find improved comfort by use of the Kanso SoftWear pad. For newly implanted users being fitted for the first time, these results indicate that the same process in selecting initial magnet Strength as for BTE fittings can be used, due to the harmonisation of magnet Strength between devices.

An OTE may be appealing to some CI candidates because relative to a BTE processor, the all-in-one design (without cables and with a single button) appears simpler to operate and manage on an ongoing basis. The results from the user survey showed that users rated the ease of use and learning about how to use the Kanso processor higher compared to the BTE processor. This is a positive finding suggesting potential benefits for patient counselling and device management for OTE recipients. Overall, the user survey results suggest some important considerations that may influence an individual’s preference towards choosing an OTE rather than a BTE solution include comfort, cosmetics, ease of use and lifestyle factors pertaining to retention. Therefore, counselling regarding sound processor choices should include consideration of these factors in addition to audiological and medical criteria.

Surgical considerations for OTE devices

Consideration regarding placement of the internal implant during CI surgery is important for users of OTE and BTE processors, but especially so for bilateral recipients where alignment of the internal devices is important to maintain symmetry of the external processors for maximum aesthetics. If a CI candidate has elected to use an OTE processor, then additional care should be given to implant positioning to ensure the placement of the external processor over the implant does not interfere with performance of the microphones. With another implant system, Kuzovkov et al (2016) recommended the use of a flatter implant bed and use of wells for fixation pins to increase stability of the implant and minimise the implant angle in relation to the skull.

There were no newly implanted recipients recruited in this study. Due to differences in surgical technique and anatomy, there was a range of implant locations within this patient group, resulting in variations in OTE processor placement on the head. While a previous study reported a significant correlation between implant/processor location and speech perception in noise for unilaterally OTE implanted users (Wimmer et al, 2015), this study failed to show a correlation. Future studies should further investigate location effects on speech perception as processor location has the potential to influence performance. Particularly, asymmetric bilateral users should be investigated as they have the potential to be affected by both overall sensitivity, and relative sensitivity between devices. While some sensitivity differences may have occurred as a result of surgical location of the implant, they were either too small to affect speech perception, or were perhaps neutralised by minor, unintended adjustments in head position during speech testing in this study. Therefore, implant recipients considering an upgrade to the Kanso processor, but who have a less common implant location or head shape should not be discounted as potential upgrade candidates.

Conclusions

Kanso is the first dual-microphone sound processor for cochlear implant recipients that is worn entirely off the ear (OTE). As with previous studies using bone conduction devices, this study has also demonstrated that the difference in processor location for OTE and behind-the-ear (BTE) devices can be largely compensated for by utilising dual-microphone technology. The study results showed no difference in speech performance (in quiet and co-located speech in noise) with Kanso compared to the Nucleus 6 BTE sound processor. In spatially separated noise performance, decrement was found for highly directional and adaptive programmes, but not in the moderately directional programme. With the default SCAN programme, Kanso demonstrated significant improvements in co-located and spatially separated noise compared to Standard directionality and single-microphone programmes. Despite previous experience with BTE sound processors, subjects rated Kanso as significantly better than their own BTE processor on measures of comfort, look and feel, ease of use, music and overall hearing performance. Results suggest that for CI candidates, the Kanso OTE processor may offer an attractive alternative to conventional BTE sound processors, especially for those who highly value discretion and simplicity.

Declaration of interest: The authors are employees of Cochlear Limited, the manufacturer of the technology described in the article.

Acknowledgements

The authors would like to thank Pam W Dawson for assistance with statistical analysis and Phyu Khing for technical support. Thanks also to Anne Beiter, Ryan Carpenter, Aaron Parkinson, Xueling Zhu, and Ingrid Mauger for reviewing the manuscript.