Sound Source Localization Testing in Single-sided Deafness Following Bone Conduction Intervention

Summary

We present a protocol to evaluate the impact of bone conduction intervention on sound localization ability in patients with single-sided deafness (SSD). This protocol can be applied to assess the efficacy of bone conduction devices in restoring sound localization abilities and improving the overall quality of life for individuals with SSD.

Abstract

Single-sided deafness (SSD), where there is severe to profound hearing loss in one ear and normal hearing in the other, is a prevalent auditory condition that significantly impacts the quality of life for those affected. The ability to accurately localize sound sources is crucial for various everyday activities, including speech communication and environmental awareness. In recent years, bone conduction intervention has emerged as a promising solution for patients with SSD, offering a non-invasive alternative to traditional air conduction hearing aids. However, the effectiveness of bone conduction devices (BCDs), especially in terms of improving sound localization abilities, remains a topic of considerable interest.

Here, we present a protocol to assess the impact of bone conduction intervention on sound localization ability in patients with SSD. The protocol includes the experimental setup (a sound-treated room and a semicircular array of loudspeakers), stimuli, and data analysis methods. Participants indicate the perceived direction of noise bursts, and their responses are analyzed using root mean square error (RMSE) and bias. The results of sound localization testing before and after bone conduction intervention are reported and compared. Despite no significant differences, most patients (71%) had a localization bias clearly toward the intervention side after bone conduction intervention. The study concludes that bone conduction intervention can promptly enhance certain sound localization skills in patients with SSD, offering evidence to support the efficacy of BCDs as a treatment for SSD.

Introduction

Sound localization, the capacity to pinpoint the precise origin of auditory stimuli, is a critical auditory skill that underpins a host of essential functions in daily life, including effective communication, safe navigation through environments, and the ability to orient oneself in space. When an individual experiences Single-sided deafness (SSD), the auditory system's ability to localize sounds is severely compromised. This is because our brains typically rely on the comparison of sound information received by both ears to calculate the location of sound sources accurately.

The human auditory system employs sophisticated signal processing techniques to localize sound sources, relying on interaural time differences (ITDs) and interaural level differences (ILDs) as primary cues. ITDs refer to the slight time delay between the arrival of sound at each ear, which provides information about the sound source's azimuth. ILDs, on the other hand, represent the difference in sound levels between the two ears. The auditory system integrates these cues with other factors, such as spectral cues and head movements, to form a precise spatial representation of the auditory environment^1,2. These binaural cues are processed and integrated to allow us to determine the direction from which a sound is coming. However, when hearing in one ear is impaired, this bilateral processing is disrupted, leading to difficulties in localizing sounds.

Bone conduction devices (BCDs) offer a promising solution for individuals with SSD^3,4. These devices work by transmitting sound vibrations directly to the cochlea through the bones of the skull, thereby circumventing the damaged outer and middle ear. BCDs are particularly useful for those with conductive or mixed hearing loss, as well as for individuals with SSD. The benefits of bone conduction technology for SSD patients have been documented in previous research. For instance, a study by Chandrasekar et al. demonstrated that bone conduction devices significantly improved speech recognition in noise for individuals with SSD³. Similarly, a meta-analysis review by Huang et al. highlighted the positive effects of BCDs on speech perception and quality of life for these patients⁴.

Despite this evidence, the specific impact of bone conduction intervention on sound localization abilities in SSD patients is not as well understood. For example, Agterberg et al. reported that the sound-localization performance of patients with single-sided deafness is not improved when listening with a bone-conduction device⁵. Some systematic reviews, such as the one by Kim et al., have reported that six previous studies with 139 cases with Bone-Anchored Hearing Aids (BAHA) have shown the percentage of correct sound localization identification to be between 13% and 65.8% before BAHA implantation and between 15% and 68.5% after the implantation but without statistical significance⁶. Because these studies used the percentage of sound source localization accuracy where scoring required accurately identifying the emitting speaker out of multiple speakers, we believe the difficulty level is relatively high. In contrast, our assessment method evaluates the angular error of sound source localization and uses the root mean square for scoring. Therefore, we consider our method to be more suitable for the demands of acute testing.

To address this gap in the literature, the current study aims to evaluate the effectiveness of BCD in restoring sound localization abilities in patients with SSD. We are using the speaker configuration that is described by van de Heyning et al.⁷.We have developed a protocol for testing sound localization that involves pre and post intervention assessments. Participants will be tested in both aided (using the BCD) and unaided conditions to compare their localization performance. By examining the changes in sound localization abilities before and after the implementation of bone conduction intervention, this study will provide valuable insights into the potential benefits of BCDs for SSD patients. The findings could contribute to a better understanding of how these devices can be optimized to improve spatial awareness and auditory function more broadly, thereby enhancing the overall quality of life for individuals with SSD.

Protocol

In this study, the participants were 14 children with congenital SSD, equipped with bone conduction hearing aids. The inclusion criteria for the participants were a confirmed diagnosis of SSD. The participants were recruited from a specialized audiology clinic and were informed about the study's purpose, procedures, and potential risks and benefits. Informed consent was obtained from the parents or legal guardians of the participants before their enrollment in the study.

1. Setup

NOTE: This section describes the procedure for conducting a sound localization experiment using the referenced software tool. The experiment is designed to assess the ability of participants to localize a sound source within a free-field setup. Localization testing was conducted in a sound-treated room with seven loudspeakers (see Fig. 2 in Van de Heyning et al.)⁷ equally distributed along a semicircle between -90° (left) and 90° (right) azimuth. Speaker configuration is chosen because of practical considerations. The materials needed for this experiment are included in the Table of Materials.

1. Ensure that a Windows PC with a compatible audio driver and multi-channel soundcard is available.
2. Connect actively powered speakers to the soundcard using balanced cables.
3. Configure the audio hardware according to the manufacturer's instructions, ensuring glitch-free playback and sufficient channel separation.
4. Position the speakers in a circular setup following the guidelines⁶. Place the subject in the center of the semicircle, facing the frontal loudspeaker. Use the software to configure the loudspeakers in the desired semicircle arrangement with a 30° angle between each adjacent speaker (see Figure 1). Ensure the center of the sound-emitting part of the loudspeakers is at the level of a hypothetical plane going through the subject's ear canals by adjusting the height of the chair according to the height and size of the subject.

2. Calibration

1. Choose the appropriate audio driver in the software.
2. Select the ASIO-compatible soundcard from the list of available devices.
3. Review and configure the necessary parameters in the setup menu, including:
   1. ShowResults: Choose when to display the results during the experiment (live, final, silent, or closed).
   2. DummyLSwarning: Enable or disable the warning message indicating the presence of dummy speakers.
   3. trainingMode: Enable or disable the training mode, where the target speaker is highlighted until a response is given.
   4. includeTrainingModeResults: Choose whether to include training mode results in the summary tables and figures.
   5. includeDemoModeResults: Choose whether to include demo mode results in the summary tables and figures.
   6. quickMode: Enable or disable quick mode, which reduces presentation levels and the number of presentations per speaker.
   7. colormap: Select the color maps for the data set and confusion matrix plots.
   8. nLS: Specify the total number of clickable speakers (real and dummy).
      NOTE: Dummy speakers mean that no sound will be emitted from the speaker during the whole process of sound source localization. Real speakers have sounds.
   9. nRep: Specify the number of repetitions per speaker.
   10. LSCircleStart / End: Specify the angle span of the circular setup.
   11. colormapDataSet: Select the color map for the data set plot.
   12. colormapConfusion: Select the color map for the confusion matrix plot.
4. Refer to the calibration instructions provided in the software to calibrate the system using a CCITT noise signal and an SPL meter with A-weighting setting.
   1. Review the driver settings of the sound device.
   2. Start the calibration procedure by clicking Extras | Calibrate.
   3. Verify the loudspeaker to sound cards channel output mapping. Assign response-only dummy speakers to channel 0.
   4. Click on a speaker button to play the calibration noise of 10 se on that loudspeaker.
   5. Measure the sound pressure level with the tip of the SPL meter at the virtual head position of the test subject pointing towards the active speaker. Consult the manual of the SPL meter about the correct measurement position. Set the sound level meter to measure the A-weighted equivalent sound level LAeq (Slow integration time).
   6. Adjust the loudspeaker/system gain(s) to achieve a noise level of approximately 70 dBA (LAeq 67-75) dB allowed). Enter the actually measured LAeq noise level in the respective calibration field.
   7. Repeat steps 2.4.3-2.4.6 for each of the remaining loudspeakers.
   8. Complete the calibration by clicking on Done.
5. Click the calibration verification button to validate the setup. With this step, the loudspeakers will represent the stimulus of signals (1,2) for operator to evaluate SPL compared to the results of calibration.

3. Experiment

1. Specify Meta Data: Enter the participant's information, including Subject ID, type of hearing aid, and optional comments.
2. Configure any response-only dummy speakers by assigning them to channel 0 during calibration. A yellow box at the top will indicate the presence of dummy speakers. If the setup parameter DummyLSwarning is true, then a text inside the box will show the number of dummy speakers.
3. Choose the study folder where the results will be saved.
4. Click the Start button to begin the experiment.
   1. The participant will be presented with auditory stimuli and prompted to respond by selecting the perceived sound source location. Have children who are able to identify numbers verbally report the corresponding loudspeaker number, and ask those lacking this ability to point directly to the loudspeaker they believe is producing the sound.
   2. Let the software randomly present two spectrally shaped noise stimuli with a duration of 1 s, including 20 ms rise and fall times. The stimuli will be consecutively presented at one of three randomly selected levels: 60 dB HL, 65 dB HL, and 70 dB. HL. The number of presentations is six per speaker (two stimuli at three levels).
      NOTE: The use of two types of noise aims to confound monaural spectral cues and prevent overestimation of localization performance.
5. View the results in real time (live mode) or after the experiment is completed (final mode). The results include the confusion matrix, root mean square (RMS), BIAS, and standard deviation (STD) of the angular error.
   NOTE: Positive values indicate a rightward bias, while negative values indicate a leftward bias. The further the value deviates from 0, the more pronounced the lateral bias is, indicating poorer localization ability.

4. Data analysis

1. Load and analyze previously saved results using the Load & Analyze function. Select from Menu | File | Load & analyze to load the MAT file of an old measurement. The result figure will be shown with the confusion matrix in a separate figure.
2. Generate summary tables and figures for all individual results in the study folder by clicking File | Create Summary.
   NOTE: The function scans for all valid Excel and MATLAB files matching the pattern LOC*.xlsx and LOC*.mat in the study folder and all subfolders.
3. Visualize the data set by plotting the number of measurements for each participant over the clinical visit tag and clinical visit number.
4. Export the summarized data as spreadsheets, including raw data and calculated statistics. The output spreadsheets are named Summary_of_all_LOC_measurements.xlsx and Summary_of_all_LOC_measurements_RAW.xlsx
5. Export scatterplots and boxplots for RMS, BIAS, and STD of angular errors, grouped by clinical visit tag and clinical visit number. Scatterplots show all the RMS, BIAS, and STD of angular errors over time. Subject IDs are color-coded, and the clinical visit tags are coded by marker symbols as shown in the legend.
6. Perform batch analysis and export confusion matrices as PNG images for all MAT files in the study folder.

5. Factory reset

1. Use the Factory Reset function to reset the software to its default settings.

Results

In this study, the participants were 14 children with SSD, equipped with bone conduction hearing aids. The age range of the participants (9 boys, 5 girls) was from 5 to 12 years old, with a median of 7.78 years (see Table 1). Without bone conduction device on the right side in Figure 2, the result of this child with left-sided deafness showed a clear rightward bias (BIAS = 53.6°) and RMS = 95.5°). With bone conduction device on the right side in Figure 3, the overall sound localization ability of this child with left-sided deafness improved to some extent, with the rightward bias reduced to BIAS = 10.7°, and the localization accuracy improved (RMS = 82.9°).

Among the 14 children tested, there were no significant differences in RMSE (P = 0.63) and standard deviation of angular error (STDE) (P = 0.15) before and after the intervention (nonparametric paired tests, and a P-value < 0.05 is statistically significant). However, 71% (10 of 14) of the children showed a shift in bias toward the side of the intervention, indicating a shift in the direction of sound source localization towards the side of the intervention. However, due to large individual differences and a relatively small sample size, there was no significant difference in the overall BIAS (P = 0.32) before and after the intervention. In summary, the utilization of BCDs can promptly enhance the sound localization capabilities among individuals to some extent. However, the degree of improvement may vary among individuals.

Figure 1: Example for setup and a single stimulus and matching response record. Localization testing was conducted in a sound-treated room with seven loudspeakers equally distributed along a semicircle between -90° (left) and 90° (right) azimuth. The subject was placed in the center of the semicircle, facing the frontal loudspeaker. The angle α is the angular error for the presentation. Please click here to view a larger version of this figure.

Figure 2: Example for result without BCD intervention. Without bone conduction device on the right side, the result of this child with left-sided deafness showed RMS = 95.5°, BIAS = 53.6°, STD = 80.6°. Please click here to view a larger version of this figure.

Figure 3: Example for result with BCD intervention. With bone conduction device on the right side, the result of this child with left-sided deafness showed RMS = 82.9°, BIAS = 10.7°, STD = 83.8°. Compared to Figure 2, the RMS and STD values are close, but the BIAS has decreased to 10.7°, where 0° represents no lateral bias. Please click here to view a larger version of this figure.

Subject No	Sex	Side	Age (years)
1	Male	L	6
2	Male	L	12
3	Male	L	7
4	Male	L	11
5	Male	L	8
6	Male	L	6
7	Female	L	5
8	Female	L	5
9	Male	L	9
10	Female	R	10
11	Female	R	9
12	Male	R	7
13	Male	R	7
14	Female	R	7

Table 1: Demographic characteristics of participants.

Discussion

Children aged 5 and older with hearing loss are able to successfully undertake this test. For those with SSD, the acute application of bone conduction hearing aids during sound source localization testing demonstrated a level of improvement in bias, although this enhancement did not achieve statistical significance in terms of RMSE STDE reduction. The improvement can also be a learning effect.

The potential for more significant improvements with prolonged use of the device, driven by neural and central processing, suggests an area that deserves further investigation. Long et al. make the point that sound source localization ability may improve with age⁸. The primary rehabilitation need of the vast majority of parents of children with SSD is to improve their ability to localize sound sources, followed by their ability to recognize speech in the presence of noise, so it is necessary to explain sound source localization and the rationale of the intervention program in detail to patients with SSD prior to the intervention and help them establish a reasonable expectation of the effect.

The impact of cochlear implant (CI) intervention on sound source localization ability in single-sided deafness (SSD) has been well established, with significant improvements demonstrated in previous studies⁹. Sound localization abilities in SSD patients are not fully understood. However, for various reasons such as fear of surgery, auditory nerve developmental anomalies, or economic constraints, many SSD patients opt for bone conduction devices (BCDs) as a form of intervention. The question of whether BCD intervention can enhance sound source localization ability in SSD patients requires careful evaluation.

Previous studies have predominantly used a scoring system that requires participants to make completely accurate judgments of the source of sound, leading to conclusions of no significant differences in localization ability before and after bone conduction device intervention^{10,11,12,13,14,15}. However, Monini et al. reported an improvement in sound localization when listening with a BCD¹⁶. Our study sought to explore whether the method adopted in our research could yield different or more nuanced results. This protocol has some critical points to be kept in mind. The maximum sound pressure level must not exceed 75 dBA and the external amplification should not change after calibration. Any other applications that may access the sound device during the experiment must be disabled. This examination method can effectively evaluate the sound source localization ability of children with SSD, providing reference for selecting treatment plans. A limitation of this test is that in children, it requires prolonged attention and obedience to the operator's commands, especially in younger children.

In conclusion, while BCDs have already demonstrated their value in improving communication for those with SSD, this study will shed light on their ability to address one of the most challenging aspects of hearing loss-sound localization. We hope that future research with larger sample sizes will provide further insights that can inform clinical practice and offer more detailed advice to SSD patients seeking guidance.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
2015a x32 or MATLAB R2018a runtime environment			1
Audio driver			1
Focusrite Scarlett 18i20 3rd Gen or other ASIO compatible multi-channel soundcard			1
Height ajustable Chair			1
LOC software tool for sound localization with a license			1
M-Audio BX5 D3 Loudspeaker			7
Microsoft EXCEL			1
Millenium BS-500 Monitor Stand			7
Pro snake 17620/10 Audio Cable 10m(Balanced TRS audiocable)			7
SPL meter			1
Tape			1
Windows PC			1