Using a cochlear implant processor as contralateral routing of signals device in unilateral cochlear implant recipients

As CROS processor, we used a CP910 Cochlear Nucleus 6 System (Cochlear Ltd. Sydney, Australia). The processor was worn behind the ear in the usual manner, but without an antenna. The CI and CROS processors were connected by custom-made bilateral connection cables that were plugged into the accessory sockets. They were symmetrically designed and could be used in either direction. All cables were tested before use. The specially designed firmware allowed the CROS processor to continuously transmit the audio signal to the CI side. Signals from the CI and CROS sides were mixed at a ratio of 1:1. The audio processing strategies of the CROS processor were identical to those of the CI processor and were set to "SCAN" (n = 6) or "Standard" (n = 4) to match the setting of the CI processor for each patient. “SCAN” is the name of the scene classifier that activates microphone directionality algorithms when appropriate [11]. “Standard” is a fixed microphone directionality.

First, we measured the signal delay caused by the cable connection of the two processors. Second, to test the output of three different setups (CI alone, CROS processor alone, CI-CROS processor), we presented a narrow-band noise centered at 1 kHz calibration signal at 65 dB SPL coming from the front. The combination of the signals was recorded via a custom cable connected to the processor’s built-in acoustic output socket on the CI side. The acoustic output was programmed to deliver the input signal without amplification and with a flat response on the whole frequency spectrum. The output was recorded with an Audio Analyzer (Rhode & Schwarz, Gemany). Third, we evaluated the directionality effect of the CI and CROS processor in an acoustic chamber (6.0 × 2.2 × 4.1 m³, frequency-independent reverberation time of 0.14 s) with a head and torso simulator (Brüel & Kjaer, Germany).

Ten German-speaking unilateral CI users using the Cochlear Nucleus System (Cochlear Ltd. Sydney, Australia) were included (5 females, 5 males, aged 48–79 years, mean age 64 years). All participants were experienced CI recipients implanted at least two years before the start of the study. All of them had good speech understanding with the CI (≥ 75% understanding of monosyllabic words at 65 dB SPL).

Study participants had two appointments. On the first visit, the CI-CROS system was adjusted and participants were instructed in the handling of the device. We advised to wear the CROS system during the entire day for at least two weeks. At the control visit, all participants stated that they had followed these instructions. During the trial period at home, participants filled out the client oriented scale of improvement (COSI) [11]. The questionnaire includes 11 hearing situations (e.g., “following conversations with 1 or 2 in quiet” and “familiar speaker on the phone”) and 4 descriptions about the emotional state (e.g., “feeling left out” and “feeling upset or angry”). Listening situations and emotional descriptions must be ranked on a scale of −5 (much better with the CI only) to + 5 (much better with the CI-CROS system). At the beginning of the trial period, participants were given a booklet with the COSI hearing situations and emotional descriptions. All questions were studied before starting the experiment. We instructed participants to answer at least one question per day. As listening situations may vary depending on the day, study participants were able to choose the question to be answered. In addition, study participants were able to add notes under each question. At the follow-up visit, the questionnaire was re-examined with the investigator and any questions noted were clarified.

As half of the study participants used a conventional hearing aid in the non-CI ear outside the study (which was not used during the trial period), for the analysis of subjective ratings, two subgroups were formed accordingly: Individuals without a hearing aid on the non-CI ear (group non-HA, n = 5; range of pure tone average (PTA): 84 to 120 dB HL) and individuals who normally wore hearing aids (group HA; n = 5; range of PTA: 49 and 116 dB HL).

On the second visit, speech understanding was evaluated in an acoustic booth. For speech perception in quiet, we employed the German Freiburg monosyllabic word test at 50 and 65 dB SPL [12]. Speech perception in noise was assessed using an adaptive German matrix test (Oldenburger Satztest—OLSA) [13]. During the OLSA test, a quasi-diffuse noise (N_D) at 65 dB SPL was presented by 4 loudspeakers (JBL, USA) arranged in 90-degree intervals with 1 m distance to the listener [3]. Sentence tests were presented from 3 different spatial configurations: speech from the front (S₀N_D), speech from the CI side (S_CIN_D), and speech from the CROS side (S_CROSN_D). For data visualization and analysis, we used GraphPad Prism (GraphPad Software, version 8.4.3 Inc., USA).

Using the cable connection, the bilateral link delay was 3 ms. The frequency spectrum in response to the calibration signal of the CI and CROS processor was very similar (Fig. 1). When the signals of the CI and CROS processor were mixed, there was a 3 dB signal increase for signals from the front. Using the Kemar simulator, the compensation of the head shadow by the additional CROS processor was observable.

With the activated CROS device, speech perception thresholds in quiet slightly increased at 50 dB SPL (Fig. 2; mean 43% without additional CROS device, 50% with the CROS device). With a higher presentation level (65 dB SPL), this effect leveled out (mean 75% without additional CROS device, 78% with the CROS device).

For speech perception in noise, with the additional CROS processor, the greatest benefit was obtained when the signal was presented to the non-CI side (Fig. 3; S_CROSN_D; mean improvement 8.4 dB). However, when the signal was applied at the CI side, on average, speech perception decreased by 4.8 dB SNR. For speech perception from the front, patients showed an almost equal performance with or without the activated CROS processor (mean difference – 0.3 dB).

Figure 4 displays the subjective ratings grouped in participants who usually do not wear hearing amplification on the non-CI side (group non-HA) and participants who do wear a hearing amplification on the non-CI side (group HA). Overall, most subjects in group non-HA were satisfied with the additional CROS device. The biggest improvements in sound quality were measured when watching television (median = 2) and listening to music (median = 1). In contrast, car journeys resulted in a deterioration in overall assessment (median =  – 1) and spatial hearing (median =  – 0.5). Contrary to our expectations, a deterioration was also observed when the CROS device was directed towards the passenger. This could be explained by a reverberation effect, which is created in a small driver's cab and, thus, leads to an amplified background noise.

In group HA, the satisfaction with the CROS device was considerably lower. Although some subjects (2 out of 5 participants) felt a benefit through the additional CROS device, others felt a deterioration in many hearing situations. Especially, a deterioration in ratings occurred during conversations in noise (median =  – 1.5), using the telephone (median =  – 1.5), and in confined spaces (e.g., in a car).

Regarding emotional states and social aspects, the idea of an additional CROS device was well tolerated by all the subjects. Overall, the patients who were already accustomed to a hearing aid on the non-CI side were more familiar with the CROS device than the non-hearing aid group.

The connection of two devices without time delay remains a challenge. In our test setup, there was a delay of 3 ms. This short delay time could only be achieved using a cable connection. Recent publications have mostly employed a wireless connection of the CROS device [9, 10] resulting in significantly longer delay times. With the 3 ms delay, our figures still showed a comb filter effect for signals from the front. In real-life situations, we anticipate this effect to be much smaller as (i) the reverberation paths of natural sounds vary according to their environment. [14] and (ii) the comb filtering quickly disappears at different stimulation angles due to the head shadow effect. However, the technical advantage of a cable connection may not outweigh the practicality and aesthetics aspects of a wireless connection. Furthermore, the problem of longer delay times could be solved by signal processing strategies. The CI and CROS signal could be delayed in such way that the devices are synchronous.

In our figures, the measured output of the CI and CROS processor had the same spectral characteristics as the CI-only signal. When both, CI and CROS processor signals were mixed, there was a 3 dB signal increase. This is expected due to signal summation and could provide a performance benefit for signals from the front. Lastly, with the switch-on of the CROS processor, the polar plot showed a more constant sensitivity across all azimuths compared to the CI-only situation; the amplification on the CI side was shifted slightly to the opposite side with suppression of the head shadow effect.

At 50 dB SPL, speech perception in quiet was slightly better when the additional CROS processor was activated. At louder stimulation levels (65 dB SPL), this effect leveled out. This finding could be due to the above-mentioned summation effect which is more pronounced at lower presentation levels [15, 16].

In noisy situations and consistent with existing literature, the CROS processor amplified speech signals when they were applied to the non-CI side (median improvement 8.4 dB SNR) [9, 17]. It is interesting to note, that a gradual increase of the signal to noise ratio seems possible over time due to a learning effect [9]. However, our experiment was too short to allow such learning curve. In our figures, in the unfavorable situation S_CIN_D, we observed that the CROS processor amplified the noise. This stands in contrast to earlier findings with the same study setup where the quasi-diffuse noise was not amplified by the additional CROS device [3]. Lastly and as expected, when speech signals came from the front, there was only a small difference whether the additional CROS processor was added or not.

Study findings from different research groups are summarized in Table 1. It must be taken into account that study protocols varied and, thus, the results cannot be compared directly. In our current figures and in comparison to our earlier findings [3], there was a significant improvement in the S_CROSN_D situation and a concurrent deterioration in the S_CIN_D situation. This can be explained by the fact that the directionality settings of the currently used system amplify lateral signals. This leads to an improved speech understanding in the situation S_CROSN_D, but also to an increased deterioration in situation S_CIN_D. With a fixed directional microphone setting to the front, we would expect less pronounced differences.

We observed a difference between the non-HA and HA group. Patients who only had a CI without an additional hearing aid were generally satisfied with the idea of a CROS device. This was reflected in the subcategory’s overall satisfaction, sound quality and following conversations. Our study results stand in contrast to findings of [9] where two different questionnaires (i.e., the Abbreviated Profile of Hearing Aid Benefit [18] and the abbreviated Speech Spatial Qualities Questionnaire [19]), did not show a difference when using the CROS device. The explanation for these controversial findings could be diverse. Most apparent, we used a questionnaire that focuses on specific everyday listening situations. Participants had to fill in at least one hearing situation per day but could decide for themselves which one it was (depending on acoustic conditions during a day). Regarding the technical configurations, our system was technically evaluated before use. A certain time delay between the two devices was present but was shorter compared to wireless systems. Lastly, the sound processing strategy of the CI and CROS processor was identical, which significantly improved the interaction between the CROS device and the CI processor.

In our experiment, in noisy environments and in rooms with a lot of reverberation, there was an adverse perception with the additional CROS device in group non-HA and HA. These situations are most challenging for a CROS device [4, 8]. The problem could be solved by a so-called mute button, which is already successfully used in certain devices [9]. This feature seems to work well in noisy situations and can significantly improve the performance of CROS devices.

In the group HA, individuals did not experience satisfaction with the CROS processor in many listening situations. This was expected, since the combination of electric and residual acoustic hearing has proven to be beneficial in many studies [20‐22]. In the future, in the subgroup of patients with residual hearing in the non-CI ear, an electro-acoustic CROS fitting could be technically feasible; the non-CI ear could be supplied with acoustic signals while at the same time the signal could be transmitted contra-laterally to the CI.