The use of thin-tube open-ear fitting has become increasingly popular in the last few years. This type of fitting offers improvements in the wearers' perception of their own voice, especially for people with a mild hearing loss or a high-frequency hearing loss. It also provides an instant fit, is easy to demonstrate, and improves the appearance of behind-the-ear (BTE) hearing aid to wearers. Both the acceptance rate and the return rate of these models are more favorable than the average for hearing aids overall. Although the openness of the fitting may reduce the potential improvement in signal-to-noise ratio (SNR) offered by a directional microphone, Kuk et al. have shown that an open-fit directional microphone provides an SNR improvement of 2–3 dB over the same hearing aid in its omnidirectional microphone mode.1
On the other hand, thin-tube open-ear fittings may compromise the audibility of the high frequencies for some wearers.2 In this paper, we will review some of these potential compromises and discuss why frequency transposition may be a solution for the limitations. We will also show that frequency transposition improves the perception of everyday sounds and may enhance consonant recognition for some people fitted with thin-tube open-ear devices.
ISSUES WITH THIN-TUBE, OPEN-EAR FITTINGS
The majority of today's open-ear fittings use a miniature BTE hearing aid that is coupled to a thin tube with an inner diameter 1 mm or less. At the end of the thin tube is a small eartip (of various sizes to fit the diameter of the ear canal) that typically leaves the ear canal open. Acoustically, this results in a real-ear occluded response (REOR) that remains the same as the real-ear unaided response (REUR). This is the type of fitting addressed in this paper.
Thin-tube, open-ear fitting is distinguished from “open-fitting,” which may include the use of a size #13 tubing (inner diameter of 1.96 mm). It is also distinguished from “thin-tube fitting,” which does not specify the amount of occlusion (or openness) of the ear canal. Increasingly there are more thin-tube fittings that leave the ear canal only partially open by means of a vented occluding earmold. It should also be noted that thin-tube, open-ear fitting is different from “thin-wire” fitting, which places the hearing aid receiver inside the wearer's ear canal. This type of fitting has been termed commercially as “receiver-in-the-ear” (RITE) technology.
Output and bandwidth limitations
Thin-tube, open-ear fittings result in a limited output and a narrower bandwidth than regular (#13)-tube, closed-ear fittings because of the openness of the fitting and the use of a narrower tubing diameter.
(1) Open versus closed fitting: Kuk and Keenan provided a detailed discussion on the effects of opening up the vent diameter on hearing aid performance.2 They include:
* Loss of low-frequency output (especially below 1000 Hz) of the hearing aid.
* Reduction of maximum available gain before feedback in the mid- and high frequencies. The maximum gain at 3000 Hz decreased from 50 dB to about 25 dB as the vent diameter increased from 0 mm (total occlusion) to an open fitting (#13 tubing used). At this time, even when the most advanced feedback-cancellation algorithm is used, the situation may be improved by only about 15 dB to a maximum gain of 40 dB.3
* Increased contribution from direct sounds interacting with the amplified sounds to result in phase cancellation.1
* Potential decrease in speech recognition of soft sounds in quiet from the loss in low-frequency output and the limitation in high-frequency gain. In a systematic study on the effect of vent diameter on speech recognition in quiet at a 50-dB-SPL presentation level, Kuk and Keenan showed an average decrease of 12%-15% in speech-recognition score between a closed mold and an open-ear fitting.2
(2) #13 tube versus thin tube: The output of a BTE hearing aid is specified using a #13 tubing and a standard earhook. The acoustic effect of replacing the #13 tubing/earhook with a thin tube commonly used in open-ear fittings is seen in Figure 1. The output from the hearing aid is altered in two ways. First, the frequency of the tubing resonance is moved downward, from the typical 1000-Hz region to a lower frequency around 800 Hz. Secondly, the high-frequency output is reduced significantly. As much as 15 dB of output reduction is seen at 1500 Hz and a reduction of 5–10 dB is experienced above 2000 Hz.
Can RITE or other technology overcome the problems?
The previous discussion suggests that unless wearers of a thin-tube, open-ear device have only a mild degree of hearing loss, they will probably not receive adequate gain compensation in the high frequencies.
To ensure adequate gain, an open-ear hearing aid must use a highly effective, active feedback-cancellation algorithm. The issue of limited bandwidth from the use of a thin tube may be partially overcome by using the RITE coupling (assuming that the proper receiver is used). The latter possibility has not been documented convincingly and is likely dependent on the type of receivers used.
Cochlear “dead” regions
Another related issue is the candidacy for this type of fitting. As indicated earlier, most candidates for this type of fitting have a high-frequency hearing loss. Indeed, many have a severe-to-profound degree of hearing loss, which may be considered “dead.” These are cases in which the hearing loss results from a complete loss of inner hair cells such that the wearers cannot use the amplified sounds for speech understanding.4
In such cases, acoustic stimulation not only will fail to improve speech understanding, but it may actually distort sounds and lead to poorer speech understanding. Cochlear implants, especially the experimental hybrid ones, may offer the possibility of directly stimulating these high-frequency fibers. However, the invasive nature of the intervention (i.e., surgery is required) may deter the widespread acceptance of this solution.
FREQUENCY TRANSPOSITION: A LIKELY ALTERNATIVE?
Frequency transposition moves the information carried by the high frequencies to a lower frequency region for decoding.5 Doing this makes audible the high-frequency sounds that are otherwise unaidable (as in a dead region) or unreachable (as in a thin-tube, open-ear fit). Because of its non-invasive nature, frequency transposition may be a more desirable solution than a hybrid cochlear implant.
Inteo Audibility Extender
The Widex Inteo hearing aid has an optional linear frequency-transposition algorithm called the Audibility Extender (AE). The mechanism behind this algorithm, and the details of its fitting were described in previous papers.6,7 Briefly, based on the in situ hearing loss (or sensogram) of the patient, this algorithm identifies a start frequency and transposes sounds above the start frequency by one octave to be mixed with the original sounds below the start frequency.
There are several reasons why this algorithm may result in a more positive outcome than those reported in the past (see Braida et al. for a review5). First, sounds are transposed down by one octave (for example, from 4000 to 2000 Hz or from 3000 to 1500 Hz). This means that in a high-frequency hearing loss the transposed sounds will fall on the slope of the audiogram where the survival of the hair cells and neurons may be more ensured. Furthermore, recent studies showed that frequency discrimination around the slope of the audiogram is more sensitive than at other regions in people with a steeply sloping hearing loss.8,9 It is logical to speculate that the transposed sounds will have the highest likelihood of being utilized.
Secondly, because transposition does not alter the temporal and spectral structures below the start frequency, the speech cues within the original sounds would be better preserved. This minimizes the extent of distortion and the potential degradation of sound quality and speech intelligibility.
Lastly, because people fitted with a thin-tube, open-ear device in general have aidable residual hearing up to 2000–3000 Hz, the start frequency for transposition could be high (4000 Hz). This would leave much of the original sounds unaltered and minimize the amount of overlap of the transposed sounds on the original sounds. This may better preserve the sound quality and lead to a higher acceptance for the transposed sounds.
To determine the efficacy of the AE program, we conducted a preliminary study of 13 hearing-impaired wearers with primarily high-frequency hearing losses. Their averaged individual hearing loss configurations (left and right ears combined), as well as the mean audiogram of all subjects, are shown in Figure 2.
All subjects were fitted bilaterally with Inteo open-ear Elan hearing aids (thin-tube, open-ear) for which a start frequency was measured individually.7 The start frequency for all but one subject was 4000 Hz (the exception had a start frequency of 3200 Hz). A listening check showed that this start frequency maintained an acceptable sound quality for almost all the subjects at the time of fitting.
After the initial fitting, we measured baseline performance on the Nonsense Syllable Test (NST) at input levels of 30 dB HL and 50 dB HL. Afterwards, subjects were assigned both the AE Off and AE On programs in a double-blind manner (neither the study audiologist nor the subjects knew the identity of the programs). They were asked to wear the hearing aids in their daily environments for 2 weeks. Subjects were instruct-ed on the operation of the hearing aids and the rationale and mechanisms of frequency transposition were explained.
Subjects were provided with a checklist of sounds to direct their attention in order to familiarize them with the perception of the processing from the two programs. Subjects were instructed to try both programs in all environments. However, they were allowed to use the program they preferred in their daily environments after they had tried both programs. The Sound Diary, a data logging feature on the Inteo, was activated so that it logged the percentage of use of each program. Subjects returned in 2 weeks for a follow-up evaluation.
During the second visit, eight subjects were evaluated on the NST in quiet at 30 dB HL and five subjects at 50 dB HL. This was done to limit the duration of each session to less than 2 hours. In addition, their preference when listening to 10 species of birds, 10 musical passages (some instrumental and some with lyrics), and 10 female discourse passages at a conversational level of 68 dB SPL was also measured. In all testing, subjects were tested with the master program (or AE Off) and the transposition program (AE On) in a blinded, counterbalance order.
Seven of the 13 subjects used the AE On program over 80% of the time and five used the AE Off program over 70% of the time. Only one subject used both programs roughly half the time. It is difficult to judge from the “time of use” data because subjects were instructed to try both programs and not use only their preferred program. However, 11 subjects indicated that the AE On program was as acceptable or more so than the AE Off program during their 2-week experience.
Figure 3 shows the averaged consonant (initial and medial) scores on the Nonsense Syllable Test (NST) measured at 30 dB HL and 50 dB HL at the initial and second visits. At a 30-dB-HL presentation level, the averaged consonant score was 47% AE Off vs. 53% AE On during the initial visit and 50% AE Off vs. 62% AE On during the follow-up visit. At the 50-dB-HL presentation level, the consonant scores were 58% AE Off vs. 62% AE On during the initial visit and 61% AE Off vs. 64% AE On during the follow-up visit.
Because of the limited number of subject data during the second visit, statistical analysis on the NST performance was completed on the initial visit only. The noted differences were statistically significant at the p <0.05 level at both presentation levels. These data showed that: (1) Transposition resulted in a higher consonant score; (2) improvement was seen even without experience (i.e., a difference was seen even at the first session); (3) additional improvement was seen during follow-up visits in both programs; and (4) more improvements were seen at the lower input level.
Figure 4 shows the averaged vowel scores on the NST at both presentation levels during the initial and follow-up visits. At the 30-dB-HL presentation level, the vowel score improved by 3% with the AE On over the AE Off during the initial visit and remained the same during the follow-up visit. At the 50-dB-HL level, a similar difference was seen between the AE On and AE Off programs during both visits. The difference noted during the first visit was statistically significant (p <0.05). These results showed that the use of transposition for this group of subjects did not negatively affect vowel perception; rather it may improve vowel identification.
Figure 5 summarizes the averaged preference for the AE program for the birds, music, and discourse stimuli. The height of each bar represents the percentage of time the AE On program was rated the same as (shown by patterned bar) or higher than (shown by solid, filled bar) the AE Off program.
For the “bird” stimuli, more than 70% of the preference was directed toward the AE On. When the criterion of similar preference was included, the percentage increased to 88%. This shows that almost 9 out of 10 subjects would prefer (or at least not be bothered by) transposition when the spectrum of the stimulus was relatively simple. The preference decreased to about 65% when music stimuli were used and 30% when female discourse passages were used as the stimuli. This pattern is similar to the results reported in a different group of subjects,6 and is in line with the expectations for frequency transposition for people with a high-frequency hearing loss.
The subjective data using female discourse passages contrast with those measured with the objective NST. Whereas the average subject improved by 5% to 12% in consonant identification with the AE On over the AE Off, the same subjects rated the speech quality with the discourse to be poorer with the AE On than the AE Off. This subjective-objective dichotomy is not new in the hearing aid literature and could result from many factors. Nonetheless, the implication of this observation in counseling the wearer, especially in the initial acceptance of this type of processing, is critical. Additional studies are being conducted to examine such possibilities.
The results of this study indicate that the use of linear frequency transposition can further improve the performance of and user satisfaction with a thin-tube, open-ear hearing aid. Such improvement was seen in improved perception of everyday sounds (seen with the birds, music stimuli, etc.) and improved identification of consonant sounds. This type of processing algorithm may be offered as a non-invasive alternative to people with a high-frequency hearing loss to enhance the audibility of high-frequency sounds beyond that provided by conventional amplification.