A Comparison of Intracochlear Pressures During Ipsilateral and Contralateral Stimulation with a Bone Conduction Implant

Abstract

Objectives:

To compare contralateral to ipsilateral stimulation with percutaneous and transcutaneous bone conduction implants.

Background:

Bone conduction implants (BCIs) effectively treat conductive and mixed hearing losses. In some cases, such as in single-sided deafness (SSD), the BCI is implanted contralateral to the remaining healthy ear in an attempt to restore some of the benefits provided by binaural hearing. While the benefit of contralateral stimulation have been shown in at least some patients, it is not clear what cues or mechanisms contribute to this function. Previous studies have investigated the motion of the ossicular chain, skull, and round window in response to bone vibration. Here, we extend those reports by reporting simultaneous measurements of cochlear promontory velocity and intracochlear pressures during bone conduction stimulation with two common BCI attachments, and directly compare ipsilateral to contralateral stimulation.

Methods:

Fresh-frozen whole human heads were prepared bilaterally with mastoidectomies. Intracochlear pressure (P_IC) in the scala vestibuli (P_SV) and tympani (P_ST) was measured with fiber optic pressure probes concurrently with cochlear promontory velocity (V_Prom) via laser Doppler vibrometry during stimulation provided with a closed-field loudspeaker or a bone conduction implant (BCI). Stimuli were pure tones between 120 – 10240 Hz, and response magnitudes and phases for P_IC and V_Prom were measured for air and bone conducted sound presentation.

Results:

Contralateral stimulation produced lower response magnitudes and longer delays than ipsilateral in all measures, particularly for high frequency stimulation. Contralateral response magnitudes were lower than ipsilateral response magnitudes by up to 10–15 dB above ~2 kHz for a skin-penetrating abutment, which increased to 25–30 dB and extended to lower frequencies when applied with a transcutaneous (skin drive) attachment.

Conclusion:

Transcranial attenuation and delay suggest that ipsilateral stimulation will be dominant for frequencies over ~1 kHz, and that complex phase interactions will occur during bilateral or bimodal stimulation. These effects indicate a mechanism by which bilateral users could gain some bilateral advantage.

1. Introduction

Bone-conduction implants (BCIs) have been utilized for many years to treat patients with unilateral and bilateral conductive and mixed hearing losses (Bento et al. 2012; Flynn et al. 2009), as well as single-sided deafness (SSD) (Kim et al. 2017). Increasingly, bone conduction (BC) stimulation is used for contralateral routing of acoustic signals in cases of SSD, and for bilateral stimulation in cases of bilateral conductive hearing loss. In both, the BCI(s) are used to restore some of the advantages of binaural listening, which represents a more complicated stimulation condition (Kim et al. 2017; Priwin et al. 2004).

Binaural hearing utilizes sound location dependent differences in input between the two ears (Middlebrooks et al. 1991). These cues underlie the ability to localize (and discriminate) sounds in space, and provide substantial benefits when listening in noisy environments (e.g. the cocktail party effect) (Blauert 1997; Hawley et al. 2004). Although binaural BC stimulation and contralateral BC stimulation in SSD restores some binaural benefit (Zeitooni et al. 2016), these represent substantially more complicated listening condition than air conduction (AC). Critically, both cochlea are stimulated by both BCIs, thus interaction will occur during bilateral stimulation (Farrell et al. 2017; Stenfelt 2005; Stenfelt et al. 2013; Zeitooni et al. 2016). Similarly, contralateral BC stimulation in SSD likely produces a complex cochlear stimulation pattern resulting from the bimodal (i.e. ipsilateral AC and contralateral BC) stimulation presented to the hearing ear. These interactions may provide cues that co-vary with sound source location (Stenfelt 2012; Tringali et al. 2015).

Recently, transcutaneous BCIs (TCBCIs), which are held in place across intact skin with magnets, have gained popularity relative to percutaneous, direct-connection BCIs (DCBCIs) (Dimitriadis et al. 2016). DCBCIs provide efficient auditory coupling of the sound processor to bone via an osseointegrated titanium abutment, but have cosmetic and medical disadvantages relating to the percutaneous abutment (House et al. 2007). TCBCIs address some of these drawbacks (skin infections/reactions, implant loss), but at the cost of auditory signal attenuation due the intervening soft tissue (Hakansson et al. 2008; Iseri et al. 2015). We have previously characterized the changes in signal transmission to the cochlea (i.e. the intracochlear sound pressure levels generated) with ipsilateral TCBCI stimulation (Mattingly et al. 2015), it remains unclear the extent to which contralateral stimulation will impact TCBCI signal transmission to the inner ear.

Previous reports have investigated the effects of BC stimulation lateralization on subjective reports or skull vibrations (Eeg-Olofsson et al. 2011; Eeg-Olofsson et al. 2013; Nolan et al. 1981; Stenfelt 2012; Stenfelt et al. 2005). To more directly characterize the differences in sound transmitted to each cochlea for ipsilateral and contralateral BC stimulation, and describe potential binaural cues to sound source location, here we combine laser Doppler vibrometry (LDV) and intracochlear sound pressure measurements during BC stimulation on each mastoid to measure the stimulation from BC stimulation to the cochlea directly. We go on to compare DCBCI and TCBCI stimulation to quantify the decrease in transmitted signals with contralateral TCBCI stimulation. Intracochlear pressure measurements are well suited to the study of BC hearing as they enable quantification of the total input signal to the cochlea (Nakajima et al. 2009; Olson 1998, 1999), including non-ossicular pathways (Banakis Hartl et al. 2016; Greene et al. 2018; Greene et al. 2016; Greene et al. 2015; Mattingly et al. 2015).

2. Materials and Methods

2.1. Specimen Preparation

The use of cadaveric human tissue complied with the University of Colorado Institutional Biosafety Committee, and was conducted with approval of the Colorado Multiple Institutional Review Board (COMIRB Exempt #14–1464). Four fresh-frozen whole heads with intact external and middle ears, and no history of middle ear disease were obtained and evaluated (Lone Tree Medical, Littleton, CO, USA). Responses of these ears to AC stimulation, and ipsilateral BC stimulation, have been reported previously (Greene et al. 2015; Mattingly et al. 2015).

The surgical preparation, experimental apparatus, stimuli, and data analysis was similar to methods described previously by our laboratory and others (Banakis Hartl et al. 2016; Farrell et al. 2017; Greene et al. 2015; Jones et al. 2017; Nakajima et al. 2009). Results have been described for a simultaneous set of measurements in a previous report (Mattingly et al. 2015), except for the contralateral stimulation location.

The general experimental setup is illustrated in Figure 1. The specimens were thawed in water, and inspected for damage and anatomical irregularities. A canal-wall-up mastoidectomy and extended facial recess approach was performed to visualize the incus, stapes, and round window (Nakajima et al. 2009). The cochlear promontory near the oval and round windows was thinned with a small diamond burr in preparation of pressure sensor insertion into the scala vestibuli (SV) and scala tympani (ST). A BI300 4 mm titanium implant fixture (Cochlear Americas, Centennial, CO) was placed contralateral to the test ear on the temporal line approximately 55 mm from the external auditory canal (EAC). The full cephalic specimens were fastened to a Mayfield head clamp (Integra Lifesciences Corp., Plainsboro, NJ) attached to a stainless-steel baseplate. Cochleostomies into the ST and SV were created ~1 mm anterior and superior to the oval and round windows (illustrated in Figure 1A) under a droplet of water (large enough to cover the cochlear promontory) using a fine pick, and enlarged until the tip of the fiber optic pressure probe just fit (<350 μm diameter). Pressure probe positions were similar to prior reports from our laboratory (e.g. Alhussaini et al. 2018; Banakis Hartl et al. 2016; Greene et al. 2016; Greene et al. 2015). Pressure sensors (FOP-M260-ENCAP, FISO Inc., Quebec, QC, Canada), were inserted into the SV and ST using micromanipulators (David Kopf Instruments, Trujunga, CA) mounted on the Mayfield Clamp, and sealed to the cochlea with alginate dental impression material (Jeltrate; Dentsply International Inc., York, PA). Sensors were flexibly secured to the micromanipulators via short sections of fine, bent copper wire, and were held in place in the cochleostomies with the tight fit of the pressure probe and the cochleostomy wall, and the alginate. The flexible attachment to the micromanipulators thus allowed the sensor tip to move with the bone, rather than with the micromanipulator; however, it is possible that the sensor tip could move somewhat within the cochleostomy due to the compliance of the alginate seal, which we discuss further in the “Limitations of this study” section in the discussion.

Figure 1.

Illustration of the experimental setup. A) A canal-wall-up mastoidectomy with facial recess was performed to expose the ossicular chain and cochlear promontory. Pressure probes were inserted into the scala vestibuli (P_SV) and scala tympani (P_ST) through small cochleostomies near the stapes footplate and round window. The LDV laser was focused, through the mastoidectomy onto glass beads placed on the stapes capitulum (V_Stap), round window membrane (V_RW), and cochlear promontory (V_Prom) between the stapes, round window, and P_SV cochleostomy. AC stimuli were presented via a closed-field acoustic system, and the sound pressure level in the external auditory canal (P_EAC) near the tympanic membrane assessed with a probe tube microphone. B) Acoustic stimuli were presented ipsilateral, and BC stimuli with a BCI transducer mounted either ipsilateral (BC_Ipsi) or contralateral (BC_Contra) to the measured ear.

Velocities of middle ear structures were measured, in line with the laser beam, with a single-axis LDV (OFV-534 & OFV-5000; Polytec Inc., Irvine, CA) mounted to a dissecting microscope (Carl Zeiss AG, Oberkochen, Germany). The LDV beam was directed through the mastoidectomy, and reflected off of microscopic retro-reflective glass beads (Polytec Inc., Irvine, CA) placed on the stapes capitulum (Stap), round window membrane (RW), and cochlear promontory between the oval and round windows (Prom) to ensure a strong LDV reflection. Stapes velocity (V_Stap) was assessed to AC stimuli before and after making the cochleostomies to verify no change in cochlear input impedance was induced by probe insertion. Similarly round window velocity (V_RW) was compared to V_Stap after pressure probe insertion to verify a 180° phase difference at low frequencies (~400 Hz). Baseline AC and BC response magnitudes, normalized to the input stimulus (i.e. transfer functions), for these specimens have been reported previously (Mattingly et al. 2015). In all LDV measurements, the position and orientation of the laser was held constant between experimental conditions, though slight shifts were unavoidable when moving the implant’s position.

2.2. Stimuli Presentation and Data Acquisition

All experiments were performed in a double-walled sound-attenuating chamber (IAC Inc., Bronx, NY). Stimuli were generated digitally, and presented to the specimen via a bare (i.e. no sound processing) BC transducer (BAHA 4, Cochlear Bone Anchored Solutions AB, Mölnlycke, Sweden), or a closed-field magnetic speaker (MF1; Tucker-Davis Technologies Inc., Alachua, FL), powered by one channel of a stereo amplifier (SA1), and driven by an external sound card (Hammerfall Multiface II, RME, Haimhausen, Germany) modified to eliminate high-pass filtering on the analog output. Stimuli were generated and responses recorded at 44100 Hz, and controlled by a custom-built program in MATLAB (MathWorks Inc., Natick, MA). During baseline AC stimulation, sounds were delivered to the ear canal through a custom-made foam and rigid rubber insert earplug inserted into a speculum, secured in the ear canal with cyanoacrylate adhesive, and sealed with dental impression material. The sound pressure level in the ear canal (P_EAC) was measured with a Brüel & Kjær probe-tube microphone (type 4182; Brüel & Kjær, Nærum, Denmark) and signal conditioner (type 2690). The microphone probe tube was inserted through a small hole in the rubber tubing, and placement near the tympanic membrane was verified by visual inspection through the tubing prior to earplug insertion. Stimuli were twenty short tone pips (twenty cycles at each frequency) presented two frequencies per octave between 120 and 10240 Hz. Input from the microphone, LDV, and pressure sensors were simultaneously captured via the sound card analog inputs.

2.3. Data Analysis

Responses measured were chosen in order to assess features of the transmission pathway to the inner ear. Particular attention is focused on V_Stap during AC, V_Prom during BC, and P_Diff (defined as the complex difference: P_Diff = P_SV – P_ST; Nakajima et al. 2009, Mattingly et al. 2015) during both AC and BC. P_Diff has been shown to directly represent the drive to the cochlea during AC stimulation in small animals (Dancer et al. 1980). P_Diff may likewise directly correlate with perception during BC stimulation, since the mechanism of traveling wave generation on the basilar membrane appears consistent across AC and BC (Dauman 2013; Stenfelt et al. 2003), however this relationship has not been directly assessed. Responses are shown as transfer functions (H), i.e. velocities (V_Prom) and pressures (P_SV, P_ST, & P_Diff) are presented normalized to SPL in the EAC (P_EAC) for AC stimuli (e.g. $H_{SV}^{AC}$ = P_SV/P_EAC), and to the voltage input to the BC transducer (E_IN) for BC stimuli (e.g. $H_{SV}^{BC}$ = P_SV/E_IN), consistent with ASTM F2504 (Rosowski et al. 2007). Resulting AC and BC transfer functions (H^AC & H^BC) were computed from the responses of these measures to pure tone stimuli. The LDV angle was held fixed as much as possible across stimulation conditions, thus minimizing potential differences due to LDV measurement angle. All acquired signals were band-pass filtered between 15 and 15000 Hz with a second order Butterworth filter for data analysis. Velocities are shown as the average of at least three repetitions; pressure was recorded continuously with all velocity measurements, thus pressures shown are the average of at least six repetitions. Responses (of each specimen) are only included in analyses for frequencies with a signal-to-noise ratio greater than 3 dB (and usually greater than 10 dB; noise estimated from the segment of the recording immediately preceding each stimulus presentation; (Mattingly et al. 2015)).

Responses to BC stimuli were compared across experimental conditions using a method inspired by Rosowski et al. and the ASTM Standards for Middle Ear Implants F2504–05 (Rosowski et al. 2007). Responses were compared between AC and BC stimulation to derive the equivalent SPL in the EAC (L_Eq) necessary to elicit the observed response magnitude (Rosowski et al. 2007). The equivalent P_EAC (L_Eq) is calculated by first dividing each electro-vibrational transfer function (H^BC) by the acoustical transfer function (H^AC) of each measure (X; e.g. SV), to derive the equivalent ear canal sound pressure transfer function (H^ET):

$$H_X^{ET}=\frac{H_X^{BC}}{H_X^{AC}}=\frac{(P_X^{BC}or{V}_X^{BC})}{(P_X^{AC}or{V}_X^{AC})}\frac{P_{EAC}}{E_{IN}}$$ (1)

The equivalent ear canal sound pressure level (P_Eq) is then computed from the product of H^ET and the electrical output to the transducer (E_IN):

$$P_{Eq}=H_X^{ET}{E}_{IN}=\frac{(P_X^{BC}or{V}_X^{BC})}{(P_X^{AC}or{V}_X^{AC})}\frac{P_{EAC}}{E_{IN}}{E}_{IN}=\frac{(P_X^{BC}or{V}_X^{BC})}{(P_X^{AC}or{V}_X^{AC})}{P}_{EAC}$$ (2)

Finally, the equivalent sound pressure level (in dB SPL) is computed by:

$$L_{Eq}=20{log}_{10}(H^{ET}\frac{E_{IN}}{20\mathrm{\mu }Pa})=20{log}_{10}(\frac{P_{Eq}}{20\mathrm{\mu }Pa})$$ (3)

P_EAC is thus effectively scaled by the ratio of intracochlear pressures measured during BC and AC stimulation, to derive an equivalent sound pressure (in Pa; P_Eq), and the equivalent sound pressure level (in dB SPL; L_Eq) is computed in the normal way. This method allows for intuitive comparisons across experimental conditions as all measures are represented as an estimate of the perceived sound level produced by the transducer, which may then be directly compared to AC stimulation.

2.4. Experimental conditions tested

Experiments were designed to assess the DCBCI, as well as the TCBCI both within and beyond the range of attachment conditions recommended by the manufacturer, to test the role of soft tissue thickness on stimulation effectiveness. Responses are described in the following paper with the following naming scheme: superscripts identify the stimulation method (e.g. AC: air conducted; DBC: direct BC; 0–9mmBC: BC with 0–9 mm thick skin flaps) and laterality (ipsi[lateral] and contra[lateral] re: the measured ear), while subscripts identify the measurement location (e.g. SV: scala vestibuli pressure). Experiments began by assessing each measure in response to AC sound presentation (AC baseline recordings; H^AC). Five BC conditions were subsequently tested contralaterally. First, the BC transducer was attached directly to a titanium abutment (BC baseline recordings; DCBCI; H^DBC) via a standard snap coupling (Cochlear Baha Connect; Cochlear Americas, Centennial, CO). Second, the BC transducer was attached to the external magnet of the TCBCI (Cochlear Baha Attract, Centennial, CO), which was placed in direct contact (except for a foam pad) with the internal magnet (H^0mmBC). Finally, 3 mm (H^3mmBC), 6 mm (H^6mmBC), and 9 mm (H^9mmBC) thick soft tissue flaps were placed between the magnets of the TCBCI. Cadaveric temporoparietal skin with subcutaneous tissue used for this purpose was harvested from the neck of the specimen, and thinned to the target thickness at the beginning of each recording session. A #5 external BAHA Attract magnet with a BAHA Softwear Pad attached was used in all TCBCI conditions. Note, stimulation was provided by the DCBCI unless explicitly stated otherwise.

3. Results

3.1. Baseline AC responses

Closed-field AC responses normalized to the sound pressure level in the ear canal (i.e. transfer functions; not shown) were computed in order to assess the condition of each ear. Six of seven ears (from 4 heads) tested met inclusion criteria as their responses were consistent with the 95% confidence interval (CI) band for stapes acoustic transfer functions in normal, healthy specimens reported previously (Rosowski et al. 2007), and thus are included in further analysis. The seventh ear was excluded from the study due to the stapes acoustic transfer function consistently lying outside of the 95% CI band. The eighth ear was not tested due to pre-existing damage noted during visual inspection. Ipsilateral response magnitudes measured during DCBCI and TCBCI stimulation were computed from the remaining specimens and are available in a previous report (Mattingly et al. 2015). In the following sections, the responses of V_Prom, P_SV, P_ST, and P_Diff, normalized to the drive voltage to the BC transducer (E_IN; i.e. BC transfer functions), to contralateral stimulation will be compared to this previously published data in these same specimens. In general, the responses of these four measures will be discussed as a group except where individual measures show divergent responses. Responses of V_Stap and V_RW are excluded from comparisons below as they may provide a poor representation of the stimulation to the cochlea during BC (Dobrev et al. 2018), but are provided as supplemental information for an interested reader.

3.2. BC Responses

Figure 2 shows response magnitudes to ipsilateral and contralateral BC stimuli via the DCBCI (V_Prom, P_SV, P_ST, and P_Diff normalized to E_IN). Responses of each specimen are shown as thin lines with filled (ipsilateral) or open (contralateral) markers. Superimposed on each plot are the mean contralateral response magnitudes (± SEM; heavy light gray line ± light gray area), as well as the mean ipsilateral response magnitudes (± SEM; heavy dark gray line ± dark gray area), across specimens. In all measures, responses to contralateral stimulation show equivalent or lower (particularly for high frequencies) response magnitudes than for ipsilateral stimulation. Note, response magnitude drops below the 3 dB signal-to-noise cutoff at some frequencies, thus are excluded from display and further analysis.

Figure 2.

Normalized response magnitudes, as a function of stimulation frequency, in response to BC stimuli via the DCBCI for V_Prom and P_SV/ST/Diff (magnitude of each response normalized to the drive voltage to the BC transducer, E_IN). Thin lines with markers indicate responses of individual specimens, whereas heavy lines and shaded areas represent the mean and SEM. Shaded markers indicate responses to ipsilateral, open marker to contralateral stimulation.

Figure 3 similarly compares response phases, (V_Prom, P_SV, P_ST, and P_Diff relative to E_IN) for ipsilateral and contralateral stimulation with the DCBCI. Once again, responses of individual specimens are shown with thin lines with filled (ipsilateral) or open (contralateral) markers, on which the mean ipsilateral and contralateral phase (± SEM; dark and light gray bands, respectively) of each set of responses across specimens are superimposed. In most measures (particularly velocities), ipsilateral stimulation produced a phase of ~ +90° whereas contralateral stimulation was offset by ~ −90° (re: E_IN) for low frequency stimulation (< 1 kHz), thus contralateral stimulation is offset from ipsilateral stimulation by ~−180°. For higher frequency stimulation, phase decreases rapidly with increasing frequency dependent upon the group delay of the stimulation, with contralateral stimulation decreasing more rapidly than ipsilateral. A notable exception to both of these generalizations is P_Diff (defined as the complex difference, P_Diff= P_SV-P_ST), which shows very similar response patterns to stimulation on both sides: a phase of 0–90° for low frequencies, and a similar rate of decrease with increasing frequency.

Figure 3.

Response phases, as a function of stimulation frequency, in response to BC stimuli via the DCBCI for V_Prom and P_SV/ST/Diff (phase of each response relative to the drive voltage to the BC transducer, E_IN). Thin lines with markers indicate responses of individual specimens, and heavy lines and shaded areas represent the mean and SEM. Shaded markers indicate responses to ipsilateral, open marker to contralateral stimulation.

Group delay is estimated as the slope of the phase as a function of frequency. If we assume that the phase response of the skull is linear (i.e. that the group delay is constant with frequency), we can estimate the group delay by fitting a straight line to the response phase (on a linear scale) relative to the input (Mattingly et al. 2015; Nakajima et al. 2009). The slopes of each set of measurements for 960 Hz and above were reasonably linear, thus were fit (least squares) to a first order polynomial using the Matlab function polyfit for each specimen (mean R² values were 0.85±0.04, except for P_SV which were ~0.65±0.1). The mean (across specimens) slopes are shown for each measure in Table 1 (in μs). In general, delays for unilateral BC stimulation (either ipsilateral or contralateral) were longer than the middle ear group delay reported for AC stimulation in humans (~84 μs; Nakajima et al. 2009; Puria 2003), which may be related to the calculation of phase relative to the transducer drive voltage (including transducer delay), rather than SPL at the tympanic membrane. Importantly, contralateral stimulation was delayed longer than ipsilateral in all four measures, showing delays > 100 μs in P_SV and P_ST, but < 50 μs in V_Prom and P_Diff. The mean intracranial delay was 107 ± 71 μs, which is comparable to a prior estimate of 0–200 μs for skull vibration near the cochlea (Stenfelt and Goode 2005).

TABLE 1.:

Transcranial delay estimated as the difference between ipsilateral and contralateral group delays.

	Group Delay (μs; mean ± std. dev.)
	VStap	VRW	VProm	PSV	PST	PDiff	Mean
Ipsilateral:	−155±6	−159±6	−213±29	−138±24	−137±6	−292±25	−182±61
Contralateral:	−385±23	−264±40	−244±40	−255±41	−248±5	−339±7	−289±58
Difference:	230±25	105±37	31±36	117±65	112±5	47±30	107±71

3.3. Equivalent EAC sound pressure levels

In order to compare effectiveness of each stimulus condition, in Figure 4 we show the equivalent sound pressure level in the EAC (L_Eq) required to elicit each response as described previously (Rosowski et al. 2007), and in equation 1. In other words, L_Eq represents, for the given voltage presented to the BC transducer (i.e. 10 V amplitude), the level of an AC sound that would generate the same magnitude of each response. Responses are shown as the mean (± SEM) L_Eq for contralateral (heavy light gray line, light gray area), and ipsilateral (heavy dark gray line, dark gray area) stimulation. Note, here contralateral and ipsilateral refer to the side of the BC stimulation; both are compared to ipsilateral AC stimulation during the L_Eq calculation. Full scale (10 V), stimulation produced L_Eq of ~100–120 dB SPL in most measures, and somewhat larger estimates of ~160 dB SPL in V_Prom. This higher response in V_Prom results from the low skull vibration (i.e. V_Prom magnitude) elicited by AC stimulation relative to BC stimulation, since L_Eq is dependent upon the ratio of response magnitudes to BC/AC stimulation (see equation 1). Contralateral stimulation was generally comparable to ipsilateral, except at high frequencies (> 1 kHz), which showed a somewhat lower (up to ~20 dB) equivalent sound pressure level in all four measures.

Figure 4.

Equivalent sound pressure levels in the ear canal (L_Eq) required to elicit response magnitudes in each of the four measures, calculated for responses to BC stimuli via the DCBCI for V_Prom and P_SV/ST/Diff. Thin lines with markers indicate responses of individual specimens, whereas heavy lines and shaded areas represent the mean and SEM.

3.4. Transcutaneous stimulation

3.4.1. Gain of contralateral re: ipsilateral stimulation

We recently reported comparisons between stimulation via DCBCI and TCBCI (with varying skin flap thicknesses) for ipsilateral stimulation, revealing skin flap thickness is inversely correlated with stimulation magnitude (Mattingly et al. 2015). Here, we report on the responses to contralateral stimulation for comparison to those ipsilateral data. Figure 5 shows mean (±SEM) transcranial gain (contralateral re: ipsilateral stimulation in dB, calculated for each specimen individually) across specimens, as a function of stimulation frequency. Responses are shown for each contralateral condition normalized to the corresponding ipsilateral measurement. Response gain was largely unchanged by contralateral attachment for low frequencies, and decreased for higher frequencies (> ~1 kHz). High frequency gain was generally consistent across stimulation condition, showing a 10–15 dB reduction for contralateral stimulation re: ipsilateral in all four measures. Contralateral stimulation, therefore, provides a consistent attenuation across BCI attachment conditions for high frequency stimulation (above ~1–2 kHz).

Figure 5.

Response magnitudes for contralateral relative to ipsilateral stimulation (ipsilateral values reported previously (Mattingly et al. 2015)) for each attachment condition. The mean (± SEM) change in responses to contralateral stimulation relative to ipsilateral stimulation across specimens is represented as gain (in dB) relative to ipsilateral stimulation. Colored lines and shaded areas represent the DCBCI and TCBCI with various skin flap thicknesses. Colored markers indicate differences that are significantly different than zero in a two-tailed Student’s t-test (p < 0.05).

3.4.2. Gain of contralateral re: ipsilateral abutment stimulation

The interaction between skin flap thickness and contralateral stimulation site was assessed by calculating the magnitude of each contralateral response with respect to ipsilateral stimulation with the abutment. Figure 6 shows these values (shown as mean (± SEM) contralateral gain re: ipsilateral abutment stimulation in dB across specimen) for all contralateral stimulation conditions. Low frequency stimulation produced a contralateral gain near zero (i.e. no attenuation) for all stimulation conditions. High frequency stimulation produced reductions in response magnitudes varying from 10–15 dB for stimulation via the abutment (same as Figure 6), up to 25–30 dB via the magnet with the thickest skin flap (9 mm). This trend is consistent with a summation of magnitude decreases observed with contralateral stimulation (Figure 6) and increasing skin flap thickness (Figure 6 in (Mattingly et al. 2015)).

Figure 6.

The interaction between skin flap thickness and contralateral relative to ipsilateral stimulation via the DCBCI (ipsilateral values reported by (Mattingly et al. 2015)). The mean (± SEM) change in responses to contralateral stimulation relative to ipsilateral DCBCI stimulation is represented as gain (in dB) relative to ipsilateral abutment stimulation. Colored lines and shaded areas represent the DCBCI and TCBCI with various skin flap thicknesses. Colored markers indicate differences that are significantly different than zero in a two-tailed Student’s t-test (p < 0.05).

3.4.3. Statistics

Statistical comparisons were performed on all measured responses with a 3-way ANOVA, with response magnitude (i.e. Figure 2) as the dependent, and BCI attachment (i.e. abutment or magnet with various skin flaps; d.f. = 4), stimulus frequency (d.f. = 10), and stimulation side (i.e. ipsilateral or contralateral; d.f. = 1) as independent variables. For brevity, we summarize, but do not detail all results here. Main effects of all three independent variables were observed in all three primary measures (H_Prom/SV/ST; F > 5, p < 0.0013; except: H_SV vs attachment, F =3.59, p = 0.0068). Likewise, significant interactions were observed between stimulation side and frequency (H_Prom/SV/ST; F > 2.84, p < 0.0001), stimulation side and attachment (H_Prom; F = 2.75, p = 0.027), and attachment type and stimulation frequency was observed in H_Prom/ST (F > 1.78, p < 0.0003). P_Diff responses were less reliable, thus the ANOVA table was incomplete, but excluding BCI attachment resulting in significant main effects of frequency (F = 4.24, p ~= 0), but not attachment side (F = 2.04, p = 0.1542), and a significant interaction term (F = 2.01, p = 0.009).

Post-hoc pairwise comparisons were computed using the Tukey honest significant difference (HSD; p < 0.05) test. Comparisons across BCI attachment conditions revealed that stimulation with the abutment, 0mm, and 3mm skin flaps were generally not significantly different from one another, whereas 6 mm and 9 mm flap thicknesses generally showed significantly lower response magnitudes. Contralateral stimulation was significantly lower than ipsilateral in all measures. Finally, comparisons across frequencies generally revealed a band of frequencies between 640 Hz and 1920 Hz that showed significantly higher response magnitudes than lower (120–480 Hz) or higher (2560–3840 Hz), which is likely an effect of the transducer output (which has a peak near 1 kHz), though specific frequency comparisons varied somewhat across response measures.

Further post-hoc analysis was conducted on the mean difference in stimulation magnitude of contralateral relative to ipsilateral (i.e. Figure 5), and relative to the ipsilateral abutment (i.e. Figure 6). 2-tailed Student’s t-tests were conducted on each the magnitude of each comparison (in dB), and conditions showing mean magnitudes significantly (p < 0.05) different than zero (no difference) are indicated with filled markers on the appropriate figures. Variability was substantial (particularly in intracochlear pressures), thus many of these conditions are not significant following Bonferonni correction (not shown); the mean differences remain significantly different than zero (particularly for V_Prom) for high frequency stimuli though.

4. Discussion

Contralateral BCI implantation is routinely used in two applications in the United States, SSD and bilateral conductive loss; however, the implications of contralateral BC stimulation remain poorly understood. The goal of such placement in SSD is to provide acoustic information from the side of the head contralateral to the hearing ear, thereby limiting the head-shadow effect and/or restoring some of the lost binaural advantage compared to single-sided listening alone. Likewise, the goal of bilateral stimulation is to provide acoustic information from both sides of the head to improve outcomes. The current results provide insight into the factors affecting BC hearing by providing estimates of transcranial attenuation and delay, which are generally consistent with prior reports (e.g., Stenfelt and Zeitooni 2013). Additionally, these results expand upon results in our previous report (Mattingly et al. 2015) by suggesting that stimulation with a TCBCI results in decreased high frequency transmission to the cochlea for contralateral in addition to ipsilateral stimulation (Mattingly et al. 2015).

4.1. Effects of Contralateral Stimulation on Signal Intensity

Our results show that contralateral response magnitudes were lower than ipsilateral BCI stimulation by 10–15 dB for frequencies above ~1 kHz. This attenuation increased to 25–30 dB and extended to lower frequencies when stimulating with the TCBCI (relative to an ipsilateral DCBCI). These results are consistent with other reports demonstrating transcranial attenuation up to 15 dB (Nolan and Lyon 1981; Stenfelt 2012), and also demonstrate that the increased attenuation seen with a TCBCI (Mattingly et al. 2015) can be approximated reasonably well by summing the transcranial and transcutaneous attenuations.

Recent reports have shown transcranial attenuation at both the audiometric and BCI positions to be variable between patients and within individuals at adjacent frequencies (Nolan and Lyon 1981; Stenfelt 2012). For those patients with higher baseline transcranial attenuation, the additional signal loss associated with the TCBCI may result in significantly poorer audibility than expected. This may be compensated for by increasing signal processor output, but power limitations due to feedback, inadequate high frequency gain, and decreased battery life could limit performance, particularly above 1,000 Hz (Kurz et al. 2014; Mattingly et al. 2015). Nevertheless, the clinical considerations, such as reduced skin reactions, may outweigh the shortcomings of TCBCIs, especially in sensitive populations (e.g. children).

Overall, our results support the use of a DCBCI in SSD patients due to the decreased attenuation and improved access to high frequency sound when compared to stimulation with a TCBCI. That is, SSD individuals may experience poorer audibility than otherwise expected when implanted with a TCBCI, similar to our results with ipsilateral stimulation (Mattingly et al. 2015).

4.2. Effects on Binaural Sound Localization Cues

Contralateral stimulation revealed significant attenuation for high, but not low frequency stimulation, suggesting that bilateral BC stimulation will generate a substantial interaural level difference for these high frequencies. In addition, phases reveal an additional delay relative to ipsilateral stimulation, suggesting that bilateral stimulation may also provide interaural time difference cues.

Binaural cues are important for sound source localization and discrimination, and arise due to sound location dependent differences in input to the two ears (e.g., Greene et al. 2014; Middlebrooks and Green 1991; Tollin et al. 2009). Benefits provided by these cues include sound source localization and discrimination, binaural summation, and elimination or reduction of the head shadow and squelch effects. These benefits, and the mechanisms giving rise to these cues, have mainly been described for AC sound (Blauert 1997; Sargent et al. 2001), and have not been thoroughly described for BC. Binaural cues with BC hearing are more complicated, as unilateral BC sound simulates both cochleae, and this bilateral stimulation results in complex interference patterns in cochlear stimulation (Farrell et al. 2017; Tringali et al. 2015). In particular, prior reports from intracochlear pressure measurements in human cadavers and human psychophysical studies suggest that transcranial delays during bilateral BC stimulation produce interference within the cochlea, resulting in a frequency-dependent conversion of ITDs to ILDs for stationary signals (Farrell et al. 2017; Rowan et al. 2008). The current results cannot provide direct evidence of such an occurrence, but do provide evidence for, and estimates of transcranial attenuation and delay that could lead to such interactions.

Human psychoacoustical studies have shown bilateral BC implantation improves sound localization performance and auditory awareness (Agterberg et al. 2011; Bosman et al. 2003; Grantham et al. 2012; Nolan and Lyon 1981; Stenfelt 2012; Wazen et al. 2003; Zeitooni et al. 2016), and provides some binaural advantage (Priwin et al. 2004; Stenfelt and Zeitooni 2013; Zeitooni et al. 2016) in some individuals. Support for this effect is provided by prior reports demonstrating differences in the timing and spectral content of vibrations recorded in the bone near the two cochleae, suggesting interaural cues are present (Stenfelt 2005, 2012). In bilateral stimulation, both ears have some useable sound sensitivity, thus bilateral stimulation provides access to interaural differences that the auditory system can use to improve spatial hearing abilities.

4.3. Implications for SSD

Conventional understanding of the benefit of BCI use in the SSD population is a decrease in the head-shadow effect, thus increasing the access to high frequency sound content presented to the deaf side (Wazen et al. 2003). Contralateral stimulation will result in complex interactions in the hearing ear thus additional benefit could be provided by the interaction. However, there remains debate regarding the extent and mechanisms of this contralateral stimulation: while no binaural cues can be generated in a single ear (by definition), there may nevertheless be monaural cues generated that co-vary with sound source location (such as timbre changes from the combined AC & BC stimulation).

Stimulation in an SSD patient will result in simultaneous stimulation of the hearing ear via ipsilateral AC (normal transmission through the ear canal and middle ear) and contralateral BC (on the contralateral mastoid) that will result in a complicated cochlear stimulation (interference) pattern (Tringali et al. 2015). Since the BC and AC stimulation mechanisms fundamentally differ, it is possible that unilateral cues may be generated that can restore some spatial hearing abilities. Supporting this, recent studies in human and chinchilla suggests that AC and BC signals interact, resulting in interference patterns in the resulting stimulation that produce a frequency dependent comb-filtering effect on cochlear stimulation that may allow some spatial awareness (Stenfelt and Zeitooni 2013; Tringali et al. 2015). Additionally, recent reports in human subjects conclude that SSD individuals may obtain some spatial benefit from additional BC stimulation (Kim et al. 2017; Stenfelt 2005).

4.3. Limitations of this study

Studies of BC sound transmission are difficult and complicated by limitations in available data collection system. Here, we assessed stimulation to the inner ear with two measurement systems. The first was a single-axis, single-point LDV. The velocity measured by a 1-D LDV includes only the component of the vibration in-line with the laser beam, excluding any orthogonal velocity components, thus provides an incomplete measure of complex vibrational modes. Furthermore, a single-point LDV only measures velocity at a single location, thus may not accurately represent the overall motion of a flexible (e.g. RW) or mobile (e.g. stapes footplate) structure (Sim et al. 2010; Stenfelt et al. 2004). These deficits limit our interpretation of stapes and RW velocity measurements for both AC and BC stimulation, thus we have chosen to withhold stapes and RW velocity measurements from the main document, including them in supplemental figures. BC stimulation, however, results in complex three-dimensional motion of the skull that is dependent upon the stimulation frequency and location, as well as skull geometry and material properties, thus a single-axis measurement is insufficient to fully describe skull vibration patterns at the cochlear promontory (Dobrev and Sim 2018). We attempted to minimize these limitations by holding the LDV orientation fixed across stimulation conditions, thus providing a consistent basis upon which to make comparisons; however, it is possible that substantial components of these responses were overlooked in our LDV measurements. Overall, the velocity responses presented here, and in our preceding report (Mattingly et al. 2015) should be assessed with these limitations in mind.

The second measurement technique employed in this study was the measurement of intracochlear pressures with miniature, fiber optic pressure probes. These sensors are inserted into the fluid of the inner ear and measure intracochlear pressures directly, thus will not be susceptible to artifacts from sensor position in the same way as the LDV measurements. The sensors used here are commercially available, robust to mechanical manipulation, and have a flat frequency response across a high bandwidth; however, these sensors show relatively low sensitivity, thus relatively high stimulation levels are required to elicit a suitable signal to noise ratio. Responses in this report were to moderate level stimuli (~100–120 dB SPL equivalent), which are below levels at which the human middle ear motion saturates (Greene et al. 2017), but may nevertheless result in different responses than at lower stimulation levels. Furthermore, insertion of the pressure probes requires creating fenestrations into the cochlea, and introducing the possibility for air to infiltrate the cochlea. We attempt to minimize this possibility by making the openings as small as possible (while still allowing the probe to fully enter the cochlear duct), opening the cochlea under water, and sealing the opening with alginate. Furthermore, we compare stapes and RW responses before and after probe insertion, and compare stapes, RW, SV & ST to responses reported in prior literature, to ensure responses are consistent with normal function.

An additional potential confound is motion of the fiber optic pressure probe relative to the bony wall of the otic capsule into which cochleostomies are created. Prior reports have suggested that, since the alginate dental impression material used to seal the probes in the cochleostomies cures to a relatively soft rubbery consistency, relative motion of the probe and the bone can result during the vigorous BC stimulation (Stieger et al. 2018). We cannot rule out the possibility that such motion has occurred; however, we believe three factors minimize the effect of such an effect on our measurements. First, in independent, follow-up measurements to those conducted in this study, no substantive differences were noted in velocities measured from the sensor tip and the adjacent cochlear promontory bone when sealed with either alginate alone or in conjunction with polymethyl methacrylate dental cement. Nevertheless, this analysis was not conducted as thoroughly as in the report by Stieger et al. (2018), nor was it conducted on the specimens contained within this report, thus the possibility of this artifact biasing our results remains. Second, differences in experimental setup and methodology may reduce the prevalence of movement artifacts in our results. Here, we conduct measurements in whole cadaveric heads, rather than in isolated temporal bones. The mass and composition of these specimens more closely simulates the conditions experienced by living individuals, and provides mass and moisture that we believe improves stability of the specimen’s condition over the experimental period. Similarly, we believe the use of a flexible probe attachment to the supporting micromanipulators helps decouple the fiber from the support structure, thereby improving coupling with the bone itself. Finally, the principle results of this report relate to changes in stimulation effectiveness across varying BC conditions. Relative motion of the probes relative to the bone would likely be relatively consistent from one condition to another (at each stimulus frequency), thus the effect of this artifact will be minimized or eliminated by such direct comparisons.

A final limitation that must be considered, while we thoroughly assessed responses in each measure reported here, we did not directly characterize the output of the BC driver, although that information is readily available from the manufacturer. Comparing intracochlear pressures to skull vibration (via V_Prom) reveals a constant relationship across frequency for all three pressure measures, across both stimulation sides, suggesting that V_Prom is a good measure of cochlear stimulation (not shown).

5. Conclusion

We have demonstrated attenuation of contralateral BC stimulation relative to ipsilateral stimulation, which is exaggerated by use of a TCBCI. These results suggest that careful audiometric assessment should be performed to determine candidacy for contralateral TCBCI placement, and that patients with poor high frequency sensitivity would likely achieve improved outcomes with ipsilateral and/or DCBCI stimulation.

In addition to the transcranial attenuation, we demonstrate a consistent delay with both DCBCIs and TCBCIs. These findings are in agreement with prior studies, and suggest that bilateral implantation, as well as contralateral implantation in SSD, may generate cues that are useful for binaural/spatial hearing.