Computed-tomography estimates of interaural mismatch in insertion depth and scalar location in bilateral cochlear-implant users

Abstract

Hypothesis:

Bilateral cochlear-implant (BI-CI) users will have a range of interaural insertion-depth mismatch because of different array placement or characteristics. Mismatch will be larger for electrodes located near the apex or outside scala tympani, or for arrays that are a mix of pre-curved and straight types.

Background:

Brainstem superior olivary-complex neurons are exquisitely sensitive to interaural-difference cues for sound localization. Because these neurons rely on interaurally place-of-stimulation-matched inputs, interaural insertion-depth or scalar-location differences for BI-CI users could cause interaural place-of-stimulation mismatch that impairs binaural abilities.

Methods:

Insertion depths and scalar locations were calculated from temporal-bone computed-tomography scans for 107 BI-CI users (27 Advanced Bionics, 62 Cochlear, 18 MED-EL).

Results:

Median interaural insertion-depth mismatch was 23.4° or 1.3 mm. Mismatch in the estimated clinically relevant range expected to impair binaural processing (>75° or 3 mm) occurred for 13–19% of electrode pairs overall, and for at least three electrode pairs for 23–37% of subjects. There was a significant three-way interaction between insertion depth, scalar location, and array type. Interaural insertion-depth mismatch was largest for apical electrodes, for electrode pairs in two different scala, and for arrays that were both pre-curved.

Conclusion:

Average BI-CI interaural insertion-depth mismatch was small; however, large interaural insertion-depth mismatch—with the potential to degrade spatial hearing—occurred frequently enough to warrant attention. For new BI-CI users, improved surgical techniques to avoid interaural insertion-depth and scalar mismatch are recommended. For existing BI-CI users with interaural insertion-depth mismatch, interaural alignment of clinical frequency tables might reduce negative spatial-hearing consequences.

I. INTRODUCTION

The purpose of a single (unilateral) cochlear implant (CI) is to partially restore sound and speech perception to people with poor hearing, typically those with bilateral moderate to profound hearing loss. Bilateral CIs (BI-CIs) are becoming increasingly prevalent^1,2. The primary purpose of BI-CIs is to convey cues for auditory spatial perception. Although BI-CI users have better sound localization and speech recognition in the presence of competing sounds compared to unilateral CI users^3–6, the spatial-hearing benefits that they receive from two ears are far less than those experienced by normal-hearing (NH) listeners^7–10, and they continue to struggle to communicate in noisy environments^11–13.

In typical auditory systems, binaural sensitivity is computed in brainstem neurons that receive tonotopically symmetric input from the auditory periphery^14–16. For BI-CIs to convey maximally useful binaural cues, it is imperative to minimize interaural place-of-stimulation mismatch^14,17–21. This can be partially achieved by surgically matching the insertion depths (i.e., minimizing interaural insertion-depth mismatch) for the two CIs in a BI-CI user. In reality, perfectly matching insertion depths is difficult to achieve in complex CI surgeries due to a high prevalence of relatively shallow insertion depths^22,23 and scalar translocations²⁴. If information about the interaural insertion-depth mismatch were available, the audiologist could potentially realign the ears using asymmetric frequency mappings in the two CIs. In addition, interaural place-of-stimulation mismatch may be caused by mismatch in the scalar location of electrodes (i.e., interaural scalar mismatch). The electrode array is commonly inserted in Scala Tympani (ST) through the round window and is intended to remain entirely in ST²⁵, without damaging the basilar membrane or Reissner’s membrane. The apical end of the electrode array can sometimes unintentionally puncture one or both membranes, producing electrodes at least partially located in Scala Media (SM) or Scala Vestibuli (SV)^25–28. Finally, it is sometimes not possible to insert the array completely within the ST cavity, in which case the array can occupy the SM or SV cavities immediately upon initial insertion. While interaural insertion-depth mismatch for BI-CI users has most commonly been examined in small-N studies using psychophysical^{17–21,29,30} and electrophysiological techniques¹⁹, our understanding of the prevalence and characteristics of this mismatch is limited at the population level³¹, and interaural scalar mismatch has not been examined at all.

This study used computed-tomography (CT) scans and computer-model rendering^32–34 to assess the extent to which interaural insertion-depth mismatch exists in a population of 107 BI-CI users. By characterizing the prevalence of interaural insertion-depth and scalar mismatch, and their dependence on array type and tonotopic location, the intent was to provide guidance to surgeons and audiologists in choosing array types and programming strategies to optimize BI-CI hearing outcomes. First, we hypothesized that BI-CI users would have a range of interaural insertion-depth mismatch because of different physical placements of the arrays or different array types in the two ears. Based on previous reports of tuning for binaural sensitivity^17,30, we defined the estimated clinically relevant range for which binaural sensitivity is expected to be negatively affected to include any interaural insertion-depth mismatch >3 mm along an average basilar membrane (BM), which is equivalent to an angular difference of approximately 75°. The population standard deviation in insertion angle for a unilateral CI is (coincidentally) about 75°, based on the summary of the literature provided in Table I of Landsberger et al.²². If an individual BI-CI user’s two insertions were independently selected at random from this assumed normal distribution, the mismatch between the two arrays would have mean zero and be normally distributed with a standard deviation a factor of √2 larger, or 106°, and thus 48% of cases would have clinically relevant absolute mismatch (>75°). However, the cochleae for a given individual are interaurally symmetric, and thus the two insertion depths are not independent. We therefore hypothesized that the prevalence of clinically relevant interaural insertion-depth mismatch would be smaller than 48%.

Second, we hypothesized that scalar translocation might be prevalent in BI-CI users, particularly at the array’s apical end^27,35. This should increase the prevalence of interaural scalar mismatch, predominantly in cases when the arrays follow different intended trajectories through the cochlea (i.e., they are a mix of pre-curved and straight types), which should in turn increase the amount of interaural insertion-depth mismatch. In other words, when considering the shape of the cochlea, the different array types, and different potential insertion trajectories, the geometry of the situation leads one to consider greater potential for interaural insertion-depth mismatch near the apex.

II. METHODS

Post-operative CT scan and image analysis^{32–34,36–38} was performed on 107 BI-CI users from Vanderbilt University and University of Maryland-College Park. The study population included 27 Advanced Bionics users (13 both pre-curved, 10 both straight, 4 mixed), 62 Cochlear Ltd. users (49 both pre-curved, 3 both straight, 10 mixed), and 18 MED-EL users (all both straight). Behavioral data for 19 of these BI-CI users (all Cochlear Ltd.) are reported in Bernstein et al.³⁹

To determine the intra-cochlear locations of the electrodes, a patient-specific cochlea shapes were derived from the patient CT scans using a model that includes segmentation of the intra-cochlear anatomy^{32–34,36–38}. This analysis quantifies the estimated electrode position in three dimensions. For the present analysis, we were interested in (1) the insertion angle based on a coordinate system defined by a line drawn between the round window and the modiolus and (2) the scalar location of the electrode (ST, SM, or SV)⁴⁰.

Figure 1 shows CT scans from two example subjects, one with a small amount of interaural insertion-depth mismatch and no interaural scalar mismatch (top row) and one with a large amount of interaural insertion-depth mismatch and an interaural scalar mismatch (bottom row). Electrode scalar location was categorized as ST, SM, or SV based on the anatomical landmarks in the cochlea; ST and SV are shown as different colors in the leftmost and rightmost panels in Fig. 1.

Figure 1:

CT-scan model analyses for two example BI-CI users, with the left two columns showing the left ear and the right two columns showing the right ear. Insertion depth can be seen in the center panels. Scala location can be seen in the leftmost and rightmost panels; ST is shown in red and SV is shown in blue. The example BI-CI user in the top row (A) has a relatively small amount of interaural insertion-depth mismatch (center panels). The electrode arrays are both located in ST (red regions, leftmost and rightmost panels), and thus there is no interaural scalar mismatch. The example BI-CI user in the bottom row (B) has a relatively large amount of insertion-depth mismatch (the right-ear array is much deeper than the left-ear array) and there are numerous electrodes located in the blue regions outside of ST (i.e., some electrode pairs have an interaural scalar mismatch).

Interaural insertion-depth mismatch for number-matched electrode pairs was characterized in three ways. First, insertion depth was estimated in angular degrees around the cochlear spiral, with the modiolus at the center of the coordinate system and the round window defined as zero degrees. The statistical analyses of the dataset were based on this standard approach from the CI CT literature. Second, for comparison with previous estimates of the dependence of binaural sensitivity on mismatch^{17–21,29,30,41–44}, the insertion depth was translated into BM distances (in millimeters) based on the spiral-ganglion correction⁴⁵ to the Greenwood⁴⁶ frequency-to-place map, assuming a 35-mm cochlear length. For both angular and BM distance, interaural insertion-depth mismatch was calculated as the absolute value of the difference for a given number-matched electrode pair. Third, to provide an acoustic-equivalent measure, interaural mismatch was calculated in octaves based on standard center-frequency (CF) allocation tables for each manufacturer. This interaural mismatch was calculated as the ratio of the CF for a given electrode to the effective CF for an electrode at the same insertion depth in the other ear, obtained by linearly interpolating the electrode insertion angles and allocation tables (log-frequency).

The dataset was statistically analyzed with a linear mixed-effects model (R version 4.0.3) using the buildmer (v1.7.1)⁴⁷ and lme4 (v1.1–26)⁴⁸ packages. The best-fitting model included interaural insertion-depth mismatch as the dependent variable and three fixed effects: insertion depth (in degrees, continuous variable), scalar location (categorical variable: ST-ST, SV-SV, and ST-SV), and array type (categorical variable: both pre-curved, both straight, and mixed). Electrodes located in SM were omitted because there were relatively few (7.3%) and because there are several different types of trauma that could be occurring for electrodes near SM, but without direct visualization of scalar structures the specific trauma is unclear. Electrodes located outside of the cochlea were also omitted. Random slopes and intercepts were included for scalar location by subject in the model of best fit chosen by the buildmer function. For the categorical variables, the model was referenced to the ST-ST and both pre-curved categories.

III. RESULTS

Figure 2 shows probability distribution functions (pdf; bars) and cumulative distribution functions (cdf; dotted curves) for interaural insertion-depth mismatch pooled across all electrodes and subjects. The median interaural insertion-depth mismatch was 23.4° (Fig. 2A), 1.3 mm (Fig. 2B), or 0.27 octaves (Fig. 2C). Figure 2A includes a vertical line at 75°, which denotes the lower boundary of the estimated clinically relevant range. Figure 2B includes a similar vertical line at 3 mm of interaural insertion-depth mismatch^{17–21,29,30,41–44}. Interaural insertion-depth mismatch >75° occurred for 13.0% of electrode pairs and >3 mm occurred for 19.0% of electrode pairs. Mismatch >75° was observed for at least one electrode pair for 32.7% of the subjects, and for at least three electrode pairs for 23.4% of the subjects. Mismatch >3 mm was observed for at least one electrode pair for 43.0%, and for at least three electrode pairs for 37.4% of the subjects.

Figure 2:

Distributions of interaural insertion-depth mismatch evaluated with respect to angle (degrees; panel A), distance (millimeters; panel B), and center frequency (octaves; panel C). Bars represent the probability distribution function (pdf) and the dotted curve shows the cumulative distribution function (cdf; i.e., the cumulative summation of the pdf). The vertical dashed lines depict an interaural insertion-depth mismatch of 75° (panel A) and 3 mm (panel B), respectively. These represent the lower boundary of the clinically relevant range where interaural insertion-depth mismatch is expected to degrade binaural perception.

Figure 3 shows the percentage of individual electrodes located in ST or SV as a function of the insertion depth. The percentage of scalar translocation increased with insertion depth. For depths <180°, fewer than 20% of electrodes were in SV. This increased to 25–35% for depths between 240°–420° and about 50% for depths >420°.

Figure 3:

Histogram of individual electrodes located in Scala Tympani (ST) and Scala Vestibuli (SV) as a function of insertion depth in degrees, illustrated in terms of the percentage of electrodes in each scala for a given insertion depth range. The numbers (N) at the top of the plot show the number of electrodes at each depth, and the numbers inside the legend indicate the total number of electrodes in each category. Error bars show +1 standard deviation.

Figure 4 shows the absolute interaural insertion-depth mismatch as a function of the average insertion depth for each electrode pair, plotted separately for each combination of scalar location (rows) and array type (columns). Data are included for the 1715 pairs from 105 subjects. Figures 4E and 4F had insufficient counts to assess the relationship between interaural insertion-depth mismatch and insertion depth, with electrodes from only 1 or 2 subjects in each case. For the other categories, the fitted regression lines show a tendency for more interaural insertion-depth mismatch with greater insertion depth (i.e., positive slopes) when at least one of the arrays was pre-curved (left and right columns).

Figure 4:

Interaural insertion-depth mismatch as a function of the average insertion depth for a given electrode pair. Each panel represents a different combination of scalar location (rows) and array type (columns). Solid lines show linear fits to the data from the linear mixed effects model. Dashed lines show the 95% confidence interval of the fits. The number of electrode pairs (the data points) and the number of subjects contributing to each condition are reported in each panel. Arrows indicate cases with >240° interaural insertion-depth mismatch that fell outside the plot range to better show the fits.

The results of the linear-mixed effects analysis are reported in Table I. There were significant main effects of insertion depth (more angular mismatch for more apically located electrodes) and scalar location (ST-SV pairs showed more angular mismatch than ST-ST pairs) on interaural insertion-depth mismatch. There were, however, many significant two- and three-way interactions, suggesting that the relationship between interaural insertion-depth mismatch and insertion depth varied across array and scalar categories.

Table I:

Results of the linear mixed-effects model.

Fixed Effects	Estimate	Std. Err.	df	t	p
Intercept	28.74	4.14	111.40	6.94	<0.001
Insertion Depth	0.04	0.01	1593.00	5.96	<0.001
Scala (SV-SV)	−4.18	8.58	29.79	−0.49	0.629
Scala (ST-SV)	−27.08	4.75	74.39	−5.71	<0.001
Array (Both STR)	0.32	7.33	112.10	0.04	0.965
Array (Mixed)	−11.70	9.74	114.50	−1.20	0.232
Insertion Depth × Scala (SV-SV)	0.06	0.02	1560.00	3.55	<0.001
Insertion Depth × Scala (ST-SV)	0.11	0.01	1541.00	9.96	<0.001
Insertion Depth × Both STR	17.39	56.67	441.80	0.31	0.759
Insertion Depth × Mixed	52.13	12.55	103.90	4.15	<0.001
Scala (SV-SV) × Both STR	−6.97	42.03	118.60	−0.17	0.869
Scala (ST-SV) × Both STR	48.77	11.83	79.57	4.12	<0.001
Scala (SV-SV) × Mixed	−0.01	0.01	1585.00	−1.02	0.309
Scala (ST-SV) × Mixed	0.02	0.02	1590.00	1.54	0.125
Insertion Depth × Scala (SV-SV) × Both STR	−0.04	0.11	1543.00	−0.34	0.733
Insertion Depth × Scala (ST-SV) × Both STR	−0.17	0.03	1137.00	−5.36	<0.001
Insertion Depth × Scala (SV-SV) × Mixed	0.16	0.14	1548.00	1.13	0.260
Insertion Depth × Scala (ST-SV) × Mixed	−0.14	0.03	1596.00	−4.54	<0.001
Random Effects	Variane
Subject	942.6
Scala (SV-SV)	957.4
Scala (ST-SV)	546.4
Residual	229.0

Significant effects and interactions are highlighted in bold. ST=Scala Tympani; STR=straight; SV=Scala Vestibuli.

To examine the interactions, seven post-hoc models were run (the previous analysis and six additional models with different reference levels), one for each combination of scalar location and array type, excluding the two SV-SV combinations with insufficient counts/subjects. Bonferroni corrections were applied for seven analyses (criterion: p<0.0071=0.05/7). Table II reports the slopes for the different scalar-location and array combinations. It also reports p-values for pairwise tests of the difference in the slopes of the fits, which was obtained by comparing the interactions between insertion depth and different categorical conditions (bottom four lines of Table I) for each of the re-leveled models.

Table II:

Slopes for the different scalar and array categories (left columns) and a comparison of slopes across categories (right columns).

						Slope Comparison
Slopes						ST-ST	ST-ST	ST-ST	SV-SV	SV-SV	SV-SV	ST-SV	ST-SV	ST-SV
Scala	Array	Intercept	Slope	p		Both PC	Both STR	Mixed	Both PC	Both STR	Mixed	Both PC	Both STR	Mixed
ST-ST	Both PC	28.740	0.036	<0.0001		--	0.309	0.125	0.0004	--	--	<0.0001	<0.0001	<0.0001
ST-ST	Both STR	29.050	0.025	0.005			--	0.038	0.733	--	--	<0.0001	0.040	0.374
ST-ST	Mixed	17.032	0.059	<0.0001				--	0.260	--	--	<0.0001	0.374	0.356
SV-SV	Both PC	24.554	0.098	<0.0001					--	--	--	0.009	0.205	0.038
SV-SV	Both STR	--	--	--						--	--	--	--	--
SV-SV	Mixed	--	--	--							--	--	--	--
ST-SV	Both PC	1.653	0.147	<0.0001								--	<0.0001	<0.0001
ST-SV	Both STR	54.105	−0.038	0.194									--	0.061
ST-SV	Mixed	38.721	0.033	0.173										--

Bolded p values are significant at the p<0.0071 level. PC=pre-curved; ST=Scala Tympani; STR=straight; SV=Scala Vestibuli.

Figure 4 and Table II show that interaural insertion-depth mismatch increased with increasing insertion depth for four combinations of scalar location and array type: all three scalar-location cases with both-pre-curved arrays (Fig. 4A, D, G) and the ST-ST mixed-array case (Fig. 4C). There was no significant effect of insertion depth in the ST-ST both-straight case (Fig. 4B). Here, the slope of the regression line was shallow (0.025), meaning that the interaural insertion-depth mismatch increased by only 9° for a full 360° turn around the cochlea towards the apex. There was also no effect of insertion depth in the ST-SV both-straight (Fig. 4H) and ST-SV mixed-array cases (Fig. 4I), although the confidence intervals for these cases were large because of relatively few data points.

The right half of Table II reports the pairwise comparisons of the slopes in Fig. 4. There was no significant effect of array type on the slope for the ST-ST conditions (top row of Fig. 4). There was a significant effect of array type on the slope for the ST-SV conditions (bottom row of Fig. 4), where there was a significantly greater slope for the both-pre-curved (Fig. 4G) condition compared to the both-straight (Fig. 4H) and mixed (Fig. 4I) conditions. There was a significant effect of scala location on the slope for the both pre-curved conditions (left column of Fig. 4); the slopes were significantly greater for the SV-SV (Fig. 4D) and ST-SV (Fig. 4G) conditions compared to the ST-ST condition (Fig. 4A). There were no significant changes in slope as a function of scalar type for both-straight (middle column of Fig. 4) and mixed array (right column of Fig. 4) conditions, although this might be partially due to the relatively few counts for the both-straight ST-SV (Fig. 4H) and mixed ST-SV (Fig. 4I) cases. Finally, there was a significantly greater slope for the both-pre-curved ST-SV condition (Fig. 4G) than all the other conditions, except for the both-pre-curved SV-SV condition (Fig. 4D).

To summarize, the significant three-way interaction reflected different dependencies of interaural insertion-depth mismatch on insertion depth. When both electrodes were in ST, there was a slight dependence of mismatch on insertion depth for all array-type categories (Fig. 4, top row). However, when both arrays were pre-curved, there was a greater dependence of mismatch on insertion depth when at least one of the electrodes was outside of ST (Fig. 4, left column). The largest dependence on insertion depth occurred for the cases with two pre-curved arrays and an interaural scalar mismatch (ST-SV scalar locations; Fig. 4G). For these cases, the slope was 0.147, which means that the interaural insertion-depth mismatch increased by 52.9° for a low-frequency electrode pair located one full turn toward the apex relative to a high-frequency pair.

IV. GENERAL DISCUSSION

A. Summary of Results and Relationship to Previous Studies

Many factors contribute to relatively small binaural benefits in BI-CI users, including device-related and surgical factors^49–52. Interaural place-of-stimulation mismatch can reduce binaural functioning in BI-CI users because the binaural differences are computed at the level of the brainstem using frequency-matched inputs¹⁴. The prevalence of interaural place-of-stimulation mismatch systematically analyzed in a large number of patients is important to report; most psychophysical or electrophysiological studies investigating interaural place-of-stimulation mismatch have involved relatively small sample sizes (N≈10)^{18–21,29,30,50,53–61}. Using a larger database of CT scans from N=107 BI-CI users, physical interaural place-of-stimulation mismatch was estimated based on interaural insertion-depth mismatch (insertion depth differences of individual electrode pairs, measured in degrees).

Our main finding was that interaural insertion-depth mismatch was common (Fig. 2), with a median interaural insertion-depth mismatch of 23.4° or 1.3 mm. This median value is slightly smaller than a recently reported median interaural insertion-depth mismatch of 39° in 50 children with simultaneously implanted BI-CIs, as estimated from x-rays³¹. More important than the magnitude of physical mismatch is the effect this would be expected to have on binaural perception. Previous reports suggest that the estimated clinically relevant range—where significant decrements in perceptual binaural sensitivity and fusion occur—includes any mismatch larger than about 3 mm^{17–21,29,30,41–44}, which is equivalent to about 75°. In the current study, 19% of electrode pairs had >3 mm and 13% of electrode pairs had >75° of interaural insertion-depth mismatch, consistent with our hypothesis that fewer than 48% of cases would show clinically relevant mismatch.

While perceptual data are not available for the full set of subjects analyzed here, Bernstein et al.³⁹ compared CT-scan and ITD-tuning estimates of interaural insertion-depth mismatch for up to five reference electrodes for a subset of N=19 of the BI-CI subjects from the current study. Only one subject (subject BI9) had interaural mismatch >75° based on CT-scan estimates of electrode insertion angle. For this subject, ITD sensitivity improved from chance performance (ITD discrimination threshold >5000 μs) for number-matched electrodes to thresholds of 500–1000 μs for a place-matched interaural pair (see their Fig. 2A, middle column). However, the incidence of mismatch in the estimated clinically relevant range was larger for the full set of N=107 subjects (13–19%) in the current report than for the smaller N=19 sample (5%) in Bernstein et al.³⁹ Given the lack of perceptual data available for the relatively larger proportion of subjects who had large mismatch in the current study, the expected effect of mismatch on spatial-hearing outcomes can instead be deduced from published estimates of the shape of the binaural-sensitivity tuning curve. ITD discrimination thresholds are expected to increase by approximately a factor of two to four for a 3-mm mismatch^17,30 or a factor of six for a 6-mm mismatch¹⁷.

It is important to note that the estimated impact of mismatch on binaural perception is derived from studies involving single-electrode stimulation in each ear. However, it has been suggested that BI-CI users may be less tolerant to interaural insertion-depth mismatch using multi-electrode stimulation⁶¹. There are also likely neural contributions to interaural mismatch resulting from interaurally asymmetric electrode-to-neural interfaces^62,63. Finally, few data exist to relate mismatch to more ecologically relevant perceptual outcomes such as binaural speech recognition and sound localization, apart from studies that have used vocoder simulations to examine these relationships^42,64–67. Given the incidence of mismatch in the BI-CI population, this is an important future direction.

Our second main finding was a significant three-way interaction between insertion depth, scala location, and array type. There was a larger interaural insertion-depth mismatch for increasing insertion depths when the electrodes had an interaural scalar mismatch and when there was at least one pre-curved array (i.e., both pre-curved or mixed arrays), but particularly when there were two pre-curved arrays (Fig. 4, Tables I and II). While this broadly confirms the second hypothesis of the study, our initial hypothesis did not include the both-pre-curved condition. Consistent with previous reports^25,27,68, the probability of non-ST electrode location for a single array increased with the insertion depth in the current data set (Fig. 3). Pre-curved arrays also tended to have more mismatch (Fig. 4, Tables I and II). The instance that caused the most interaural insertion-depth mismatch was when there were two pre-curved arrays and translocation occurred, typically at the apex (Fig. 4G). Previous reports suggest that pre-curved arrays tend to undergo an initial translocation at a more basal location than straight arrays²⁷. An early translocation likely causes a deviation in the path of the array that amplifies insertion-depth mismatch throughout the translocated portion, leading to large interaural insertion-depth mismatch with pre-curved arrays and more apical locations.

These results might account for some of the dependence of binaural sensitivity on tonotopic place that is sometimes reported in the literature⁶, in particular the tendency for the poorest ITD discrimination performance to occur near the apical end of the array^50,51,69,70. Sensitivity might be degraded by translocations⁷¹ and interaural insertion-depth or scalar mismatch (Figs. 3 and 4), all of which become more prevalent toward the apex. These occurrences might be inconsistent across different studies with small study populations, explaining the mixed conclusions.

B. Limitations and Future Directions

While we have shown that interaural insertion-depth mismatch occurs frequently in BI-CI users, this analysis reflects only the anatomical location and does not account for other factors that might affect binaural sensitivity. Peripheral biological factors such as neural degeneration, dead regions, and tissue scarring may alter the relative populations of peripheral neurons being stimulated^62,63,72. Furthermore, it is possible that binaural circuits could adapt to address interaural insertion-depth mismatch^73–75, which could affect binaural functioning and/or binaural fusion^44,76; however, little physiological evidence has supported such an idea to date^19,39.

To assess the relationship between physical and functional mismatch more directly, future work should consider relating binaural sensitivity to CT measurements of intracochlear electrode location. Our recent measurements in a smaller group of 19 BI-CI users (a subset of the 107 BI-CI users examined here) and 23 SSD-CI listeners suggest that CT and ITD-based estimates of interaural insertion-depth mismatch are closely aligned³⁹. In contrast, perceptual^19,39,76 or electrophysiological binaural sensitivity show incongruence with pitch¹⁹. Another limitation of the current study was that even though it included >100 BI-CI users, a larger sample would have provided more statistical power to examine some of the existing subgroups (e.g., use of two straight arrays, or a mix of pre-curved and straight arrays, Fig. 4) or additional subgroups (e.g., separating out different generations of pre-curved arrays within a manufacturer), as well as accounting for the different lengths of the arrays.

C. Clinical Implications

Interaural place-of-stimulation matching has the potential to improve binaural performance in BI-CI users. During implantation, improved surgical techniques to avoid interaural insertion-depth and scalar mismatch are recommended, for example, by using CT-based surgical guidance⁷⁷.

Post-implantation, knowledge of the interaural cochlear locations could guide the audiological frequency-mapping process to optimize spatial hearing. While this might be facilitated by psychophysical and electrophysiological measurements of binaural performance^{17–19,29,30,44}, these measurements require ITD sensitivity, which is not possible for some CI users^78–80, and can be laboriously time consuming³⁰. CT scans, on the other hand, are rapid and precise, albeit with the downside of any risks associated with increased radiation exposure, especially for children⁸¹. This approach acknowledges that the binaural system requires coordination across the ears, rather than treating the two devices as separate systems.

Pre-implantation, this study also guides decision-making with respect to array selection. There are competing issues involved in the selection of an array type, including the length of the array and the intended intra-scalar location. First, one should consider that longer arrays (typically straight arrays) that might have deeper insertions would be more inclined to have interaural insertion-depth and scalar mismatch toward the apex. Second, some studies suggest that a pre-curved array might provide an advantage for monaural speech-understanding outcomes and provide better channel independence^82,83, although translocations with such arrays tend to occur more basally than straight arrays²⁷ and appear to be a potential problem linked with decreased performance⁷¹. Our results showed that pre-curved arrays were more likely than straight arrays to result in more interaural insertion-depth and scalar mismatch toward the apex. The potential consequence of interaural mismatch is reduced binaural sensitivity (although frequency-mapping programming techniques might address this issue). It is important to qualify this conclusion because the observed significant differences between array types were small and it is unclear if these differences would substantially affect perception.