1.1.

Cochlear Implants and Their Challenges

Cochlear implants (CIs) are considered one of the most successful neural prostheses. Today about 350,000 individuals with severe-to-profound hearing loss have received a CI to restore some of their hearing. However, the performance of individual users varies largely. While some patients are able to communicate over the phone in different languages, others receive little benefit from CIs. For all CI users, noisy listening environments and music perception constitute a challenge.^1–3 It has been argued that performance could be improved by reducing the interaction between neighboring CI electrode contacts, subsequently creating more independent channels for stimulation. More spatially selective stimulation with electric current can be achieved through multipolar stimulation, where multiple electrode contacts are used to narrow the current field.^4–7 Another approach to increasing the number of different pitch percepts is called current steering.^8–10 In this approach, neighboring electrodes are used simultaneously to “steer” the current to selected neuron populations between the two contacts. However, this technique does not introduce more independent channels for parallel stimulation.^10–14

1.2.

Optical Stimulation

More recently, the use of photons has been suggested as an approach to evoke responses from small populations of neurons,^15–18 because optical radiation can be delivered spatially selectively.^19,20 It is anticipated that optical stimulation decreases interactions between neighboring channels, allowing for the development of neural prostheses with enhanced neural fidelity. Two methods for direct neural stimulation with light are currently considered: optogenetics and infrared neural stimulation (INS).^15,16,18,21 Optogenetics requires the delivery of a viral vector to express photosensitive ion channels in the membrane of the spiral ganglion neurons (SGNs).^16,22 INS does not require such treatment because during INS, the fluid in the target tissue absorbs the photons and the energy is converted into heat.^23–27 Spatially and temporally confined heating evokes action potentials in the SGNs (see below for mechanism). Although both methods appear promising, they also have challenges. For optogenetics, the neurons must be manipulated genetically. This requires targeting of a selected population of neurons with a viral vector to induce stable expression of light-sensitive ion channels. The rate by which the ion channel or optogenetic tool is expressed is crucial since low expression of the optogenetic tool will require larger photon flux rates and high expression of the ion channel may damage the cell. Moreover, tissue in the beam path largely scatters the incident photons resulting in broad response profiles and a significant reduction of the transmitted radiant energy. Published results have shown that the energy required to evoke an action potential on the murine auditory nerve is about 10-times larger than for electrical stimulation.¹⁶ The challenge for INS is the delivery of heat to the target structure(s), which needs to be removed or dissipated to prevent thermal damage during stimulation. Tissue heating limits the rate of stimulation to about 250 pulses per second (pps) at a maximal radiant energy of $25\mu \mathrm{J}/\text{pulse}$.^28–31 Optical sources, small enough to be inserted into the cochlea, have a low-wall-plug efficiency (ratio in converting electrical power to optical radiation). The energy required to stimulate with INS is about 100 times larger when compared to electrical stimulation.^18,32 Optoacoustic events resulting from INS must be considered for patients with residual hearing.

When comparing optogenetics and INS, the wavelengths of the radiation used for each of the two methods should also be considered. The incident radiant energy is reduced by the tissue in the beam path through scattering and absorption of the photons. At wavelengths used for optogenetics, $\lambda <1064\mathrm{nm}$, the extinction coefficient for the radiation is governed by the scattering of the photons, whereas above $\lambda =1064\mathrm{nm}$, the absorption becomes the dominant factor in tissue. Hence, for optogenetics, tissue reduces the transmitted radiant energy and broadens the beam by scattering. For INS, the energy is mostly reduced through absorption of the photons and less through scattering.^33,34

1.3.

INS-Neurons are Activated by Temporally and Spatially Confined Heating

One of the first reports on laser irradiation as a method to stimulate neurons came from Fork’s study on Aplysia Californica.³⁵ Irradiation of the tissue with blue ($\lambda =488\mathrm{nm}$) light ($\text{spot size}=10\mu \mathrm{m}$) evoked action potentials at stimulus levels above 12.5 mW radiation power.³⁵ Wells and coworkers studied light tissue interactions using the free-electron laser in great detail. They determined radiation wavelengths that could be used for safe neural stimulation, which are in the near-infrared and infrared.³⁶ One of the wavelength ranges for which optical sources exist for stimulation is between 1840 and 2100 nm. Upon the absorption of the photons by the water, their energy is converted into heat,²³ which then evokes an action potential. Temporally and spatially confined heating depolarizes the cell by changing the membrane capacitance^37–40 resulting in a depolarizing inward current. The change in capacitance might result from changes in membrane thickness⁴¹ or from small-diameter nanopores in the membrane.⁴² Furthermore, it has been demonstrated that transient receptor potential cation channels of the vanilloid group (TRPV) are involved.^26,43–45 They are temperature sensitive and are highly calcium selective.^46–55 Published results demonstrated that intracellular calcium homeostasis changes during INS.^56–61 Spatially and temporally confined heating, which occurs during INS, also results in stress relaxation waves.⁶² Those pressure waves could vibrate the basilar membrane and evoke auditory responses through stimulation of remaining inner hair cells. Results have been presented where cochlear INS did not evoke responses in deaf animals.^63–65 Tan et al. concluded that INS in the cochlea only originates from the generation of a pressure wave. Those findings differ from reports that showed responses in deaf animals missing hair cells^15,66 and in congenitally deaf mice.⁶⁷ One of the deaf mouse models lacks the vesicular glutamate transporter-3 ($\mathrm{VGLUT}{3}^{-/-}$) and does not release glutamate at the inner hair cell afferent synapse.^67–70 This mouse model shows no auditory response to sound stimuli but responds to INS, indicating direct interactions between INS and SGNs. Other gene manipulated mice, which show no auditory brainstem response (ABR) response to acoustical stimuli, but respond to INS,⁶⁷ are the Atoh1-cre; ${\mathrm{Atoh}1}^{\mathrm{f}/\mathrm{kineurog}1}$ mice.^71,72

To use INS in a CI, photons must be delivered to selected sites along the cochlea. This can be achieved by inserting light delivery systems (LDSs) into scala tympani of the basal turn of a cochlea. LDSs can be arrays of optical sources, such as side-emitting laser diodes (SELDs) or vertical cavity surface emitting lasers (VCSELs), bundles of glass fibers, or bundles of polyimide waveguides. Low ${\mathrm{H}}_2\mathrm{O}$ containing glass fibers are not considered as LDS because they are too stiff and break easily if they are larger than $50\text{-}\mu \mathrm{m}$ in diameter. Stiff optical fiber bundles will also damage the cochlear soft tissue structures during insertion into scala tympani and cannot be inserted at sufficient lengths.⁷³ Polyimide waveguides are flexible and biocompatible. However, challenges remain in coupling the light sources and waveguides. Moreover, the size of the waveguides limits the number of individual optical stimulation sites along the spiral ganglion.⁷³

This paper discusses how to build and test optrodes with small optical sources, which are powerful enough for INS.

2.1.

Light Sources

Currently, three different types of light sources have been used to fabricate optrodes, VCSELs, SELDs, and microlight-emitting diodes ($\mu \text{-}\mathrm{LEDs}$). The light sources are available in sizes that can be assembled into arrays suitable for insertion into scala tympani (Fig. 1). Considering the frequency place map in the human cochlea (organ of Corti and spiral ganglion),^74,75 50 light sources could be assembled into a 24-mm long array and provide a frequency resolution of about 1/8 of an octave.^74,75 Our most recent arrays feature 15 light sources on a 24-mm long array. The number of optical sources will be increased in future optrodes by placing the optical dies closer together.

Fig. 1

Different light sources. (a) The top and (b) the reverse side of a VCSEL ($\lambda =680\mathrm{nm}$). The dimension is $250\times 250\times 200\mu {\mathrm{m}}^3$ and its maximum output power is about 4 mW. (c) and (d) The top and bottom of a $5\times 7$ VCSEL array, ($\lambda =1860\mathrm{nm}$). The dimension is $450\times 250\times 200\mu {\mathrm{m}}^3$ and its maximum output power is about 7.5 mW. Each circle in (c) represents one VCSEL. (d) The light emitting window and the cathode, (e) and (f) a SELD ($\lambda =1850\mathrm{nm}$). The dimension is $450\times 350\times 100\mu {\mathrm{m}}^3$ and its maximum output power is about 50 mW. (g) The original appearance of a blue $\mu \mathrm{LED}$ ($\lambda =470\mathrm{nm}$) before being resized. The dimension is $1000\times 600\times 200\mu {\mathrm{m}}^3$ and its maximum output power is about 34 mW. Scale bars are the same for (a)–(f) [shown in (b)] and (g)–(h) [shown in (h)]: $100\mu \mathrm{m}$. The power ratings are given for continuous wavemode operation.

2.2.

Optrodes and Hybrids

The first step in the fabrication of the multichannel optrode was connecting the cathode of the light sources with conductive silver epoxy (EPO-TEK H20E, Epoxy Technology Inc., Billerica, Massachusetts) to a $125\text{-}\mu \mathrm{m}$ diameter silver wire [Figs. 2(a), 2(c), and 2(e)]. The distance between the sources could be adjusted as needed. The silver wire, which could be replaced by strips of platinum or silver foil, also serves as a heat sink. In the second step, a $25\text{-}\mu \mathrm{m}$ diameter Teflon-coated platinum wire was connected to the anode of each source using conductive silver epoxy. Note that wire-bonding has been tested as an alternate contacting method [Fig. 1(a)], however, the connection is fragile and using epoxy resulted in more reliable connections of the light sources. Following the assembly of the optrode, the function of each light source was tested before and after the optrode was embedded into silicone. For the silicone embedding, the optrode was placed in a custom fabricated mold. The mold was filled with Silastic [MDX4-4210, Medical Grade Elastomere, base and curing agent (LOT 0006932899, Dow Corning Corp., USA)] and was allowed to cure overnight in an oven at 60°C. After the silicone was solidified, the electrode was removed from the mold [Figs. 2(b), 2(d), and 2(f)] and the wires of the optrode were extended to about 10 cm. Then the optrode could be inserted into a Tygon^® Micromix flexible microbore plastic tubing with an inner diameter of 1 mm and an outer diameter of 1.8 mm (LOT 507206, Saint-Gobain Performance Plastics, Portage, Wisconsin) and was connected to a transcutaneous connector [Fig. 2(b)]. During implantation, the transcutaneous connector was secured with Ethilon 3.0 (Ethicon, Cincinnati, Ohio) to the skin incision [Fig. 2(b)].

Fig. 2

Optrodes fabricated with small optical sources. (a) A three-channel optrode. The two infrared sources in (a) are VCSELs ($\lambda =1860\mathrm{nm}$, larger dies to the left). A red source ($\lambda =680\mathrm{nm}$, smaller die to the right) serves as a pilot light, which helps to orient the implant. (b) The completed electrode ready for implantation. At the bottom right, the transcutaneous connector is shown. (c) A picture of a two-channel optical array made of SELDs. The size of the given emitters is $450\mu \mathrm{m}\times 300\mu \mathrm{m}\times 100\mu \mathrm{m}$. The anode of each emitter was connected with a thin gold wire, and then to a Teflon-insulated silver wire. The cathodes of both emitters were connected to a single silver wire, which also served as a heat sink for the light sources. The direction of the light emission is indicated by the arrow. (d) A picture of a silicone-embedded single-channel optical array made with a SELD. The anode and cathode connections are the same as shown in panel (c). The array is embedded in silicone. (e) An array of $15\mu \mathrm{LEDs}$ connected with a silver wire to the cathodes and platinum wires to the anodes. (f) An optrode made with 15 light sources. (g) The insertion of this optrode into a human scala tympani model. (g) and (i) The radiation of a single $\mu \mathrm{LED}$ and multiple $\mu \mathrm{LEDs}$. $\text{Scale bars}=900\mu \mathrm{m}$.

A different optrode design used Flexible Printed Circuit Board (FPCB) technology. The single-layer FPCB was designed as the light source carrier, which renders the optrode fabrication process much easier. The substrate for the CI must be soft, flexible, and biocompatible. It has been demonstrated that polyimide polymers are suitable in stiffness (see also discussion), are biocompatible,^76,77–80 and were selected for the support base and the insulation cover layer. Copper was selected as the conductive material. Adhesive films provide the material to bond the copper foil to the base film. The epoxy and copper contacts were further coated with silicone for biocompatibility.

To fabricate the multichannel optrode carrier, a $25\text{-}\mu \mathrm{m}$-thick copper foil was laminated on the upper surface of the polyimide substrate. Unwanted copper was etched from the copper layer, such that the resulting wire width was $80\mu \mathrm{m}$. To isolate each channel, a $25\mu \mathrm{m}$-thick polyimide film was laminated on the surface. Subsequently, the polyimide film applied for insulation was etched away on the top of the light source mounting areas and solder joints. Light source mounting areas and solder joints were further improved by electroplating a $25\text{-}\mu \mathrm{m}$-thick gold layer on the contact areas [Fig. 3(a)]. Dictated by the number of current sources of our portable diode driver system, we only fabricated three channel optrodes. The number of contacts, however, can easily be expanded. Figure 3(b) shows the FPCB carrier and its tip. Figure 3(c) shows the FPCB-based optrode with a connector. Figures 3(d)–3(f) show the FPCB carrier and a three-channel optrode made with VCSELs, $\mu \mathrm{LEDs}$, and SELDs. This carrier can also accommodate other light sources and metal contacts for electrical stimulation.

Fig. 3

The fabrication of the optrode based on FPCB technique. (a) The vertical fabrication structure and materials of the light source carrier and (b) the FPCB carrier and the tip. The light source mounting area on the tip is $100\mu \mathrm{m}\times 100\mu \mathrm{m}$; the distance between two channels is 1 mm; the width of the carrier is 0.75 mm, and the thickness is $100\mu \mathrm{m}$. (c) An FPCB-based optrode with a connector at the end and (d) the tip of an FPCB-based optrode with three VCSELs. (e) The tip of a FPCB-based optrode with three $\mu \mathrm{LEDs}$. (f) The tip of a FPCB-based optrode with three infrared SELDs.

2.3.

Insertion Force Measurements

The insertion force of the electrodes and optrodes was measured in five cadaveric cat and human cochleae, as well as in a matrix printed model of the human scala tympani ($\sim 0.016\text{-}\mathrm{mm}$ resolution, Objet 260vs Dental, Fisher Unitech Corporation, Chicago, Illinois). The human cochleae were obtained through the Anatomy Gifts Registry. Each cochlea was accessed via a retro-facial approach and a cochleostomy was created 0.5 to 1 mm from the round window. This allowed the insertion of five short custom fabricated optrodes (with $<5$ contacts, Table 1). Four long arrays (with $10+$ contacts, Table 1) were tested in the human scala tympani model. For the experiments, the electrodes/optrodes were either mounted on a Narishige electrical step-motor (MM108; Narishige, Japan) or an LTS150 translation stage (Thorlabs, Newton, New Jersey) to advance the optrode with constant speed at 0.553 or 0.25 mm/s, respectively. At the same time, the insertion force was measured with a Mark-10 Digital Force Gauge (Model M5-012, Mark-10 Corporation, Copiague, New York). The MESUR Lite by Mark-10 software was used to acquire the data at a sampling rate of 10 Hz. The electrode advancement was always monitored through a microscope and was stopped immediately when the electrode would not advance.

Table 1

The different arrays used for the insertion force measurements in a cat cochlea, human cochleae, and a human scala tympani model. The number of optical or electrical sources, the width, the height, insertion force, and distance are also listed. The four arrays equal or longer than 15 mm decrease their width and height from base to apex. Changes are given by the ranges.

2.4.

Testing in an Animal Model

All animal procedures were carried out in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Northwestern University.

2.4.2.

Cochlear implantation in cats

In preparation for the implantation surgery, each animal was premedicated with Telazol (2 to 4 mg/kg, intramuscular), butorphanol (0.4 mg/kg, subcutaneously), and atropine (0.04 mg/kg, subcutaneously). Intravenous catheters (22G) were placed in the left and right cephalic veins, and Ringer’s solution containing 2.5% dextrose was given throughout the length of the procedure. Anesthesia was maintained with isoflurane (1% to 3%). Only the left ear was implanted with the optrode. The surgical area was aseptically prepared. A “C” shaped incision was made behind the left pinna, and the bulla was surgically accessed. An opening, $\sim 5\times 5{\mathrm{mm}}^2$, was created in the bulla with a motorized drill (Micro-Torque II, WPI) and a 3-mm cutting drill bit to visualize the basal turn of the cochlea. The cochleostomy was then made with a cutting drill bit (1 mm) attached to the motorized drill. The optrode was inserted about 5 mm through the bulla and the cochleostomy into scala tympani of the cochlea [Fig. 4(a)]. The optrodes were then secured at the bulla with acrylic. The acrylic not only secured the optrode but also sealed the bulla [Figs. 4(b) and 4(c)].

Fig. 4

The implantation of the optrode into a cat cochlea. (a) The optrode was inserted into a cat cochlea through the cochleostomy, (b) the optrode was fixed to bulla with dental acrylic, (c) the second layer of dental acrylic, and (d) the transcutaneous connector was secured onto the lower neck skin.

A small incision, 1-cm long, was made in the skin between the scapulae. The wire bundles to the optrodes were tunneled under the skin from the bulla to the scapular incision, where the electrical connector was sutured to the skin [Fig. 4(d)]. The incisions were closed in several layers with interrupted sutures. Postoperatively, the animal was monitored daily and received buprenex (0.005 to 0.01 mg/kg, subcutaneous, $2\times /\mathrm{day}$ for 2 to 3 days) and meloxicam (0.1 mg/kg, oral, $1\times /\mathrm{day}$ for 3 to 4 days) for pain management. No vestibular deficits were seen in any of the animals.

2.5.

Imaging at the European Synchrotron Radiation Facility

Tomographic data sets were acquired at beamline ID19 of the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. Due to the 150 m-length of the beamline, excellent coherent properties of the x-ray wave fronts at the position of the experiment allow for very sensitive imaging by means of propagation-based phase contrast. In order to reduce dose to the sample tissues while maintaining a high contrast, the beamline’s single-harmonic undulator u13 (26.3 keV) was chosen. In addition, a 1-mm-thick diamond, a 2.8-mm-thick aluminum absorber, and a 0.5-mm-thick polished Beryllium exit window, the beamline was operated optics-free (pink). For detection, an in-house-developed indirect system was used, combining two commercial lenses (Hasselblad) in tandem-design. Using the ratio of the focal distance of the two lenses, here 100 and 200, the magnification was set to $2\times$. The lenses project the luminescence image of a 500-μm-thick LuAG:Ce (Ce-doped Lu3Al5O12) single-crystal scintillator onto the sensor of a commercial camera (pco.edge, type: 5.5, $2520\times 2160\text{pixels}$, each $6.4\text{-}\mu \mathrm{m}$ pixel size). At the conclusion of the experiment, the projections were used to reconstruct the samples. Custom written phase retrieval software was used for the reconstructions.⁹

2.6.

Statistics

Average and standard deviations were calculated. Statistical analysis was completed on the ABR data to determine any significant elevation of threshold following implantation. An ANOVA was used to determine statistical significance, with the null hypothesis indicating no threshold difference between the two conditions. A one-tailed test was used for the postoperative measurement since a threshold decrease following cochlear implantation was highly unlikely.

3.1.

In Vitro Insertion of Test Electrodes

The five panels of Fig. 5 illustrate the insertion process of a sham optrode into a cadaveric cat cochlea. The sham optrode has five optical sources but the sources are only connected to the backbone. In Fig. 5(a), the electrode tip is just inserted through the cochleostomy and in Fig. 5(d), the entire optrode was inserted. After optimizing the optrode’s shape, functional ones were assembled. The longest insertion depth of an optrode in the cat cochlea was 6 mm, which was acceptable for 5 red VCSELs, or 4 infrared VCSELs or SELDs arrays, or 4 blue LEDs.

Fig. 5

The insertion of the sham optrode into a cadaveric cat cochlea. (a) The landmarks of the magnified view of the cochlea, the round window, the cochleostomy, and the optrode. (b)–(d) The progressive insertion of the optrode into the cochlea. In panel (d), the entire optrode is inserted into scala tympani. The length of insertion is about a 6 mm. Considering the spacing of the optical sources and insertion depth, the maximum number of VCSELs that can be inserted into the cat cochlear at this time is five.

3.2.

Insertion Force Measurements

Insertion force was measured with different custom-made optrodes in cat and human cochleae. For the short and thin arrays, the insertion forces were relatively small ($<5\mathrm{mN}$, Table 1). Insertion force was also tested with a plexiglass model of the human scala tympani, which allowed a direct view of the insertion depth (Fig. 6). Four arrays were tested in this model: two custom-made optrodes and two electrical arrays from contemporary CI systems (Table 1, Fig. 6). The insertion force and depth were comparable between the optrode with 15 contacts and the 16-channel electrode. Of note, the insertion depth was shorter than the entire length of the optrodes and electrodes. All optrodes and electrodes could be inserted with full length by hand [Figs. 2(g)–2(i)].

Fig. 6

The changes in force during insertion in a model of the human scala tympani at different depths of insertion for four arrays. The optrode has 10 (blue line) or 15 (turquoise line) $\mu \mathrm{LEDs}$. The electrical alone arrays have 16 or 12 contacts. The placement of each array in the model is shown in the four corresponding inserts.

The placement of the optrodes in the cat cochleae was also examined by x-ray microtomography with synchrotron radiation. Figure 7(a) shows a typical projection obtained during the scans. The reconstruction of the optrode is shown in Figs. 7(b) and 7(d). Corresponding sketches are shown in Figs. 7(c) and 7(e).

Fig. 7

(a) An x-ray projection of an inserted optical array in situ in a cat cochlea. The thick wire is the backbone and acts as a heat sink. The thin wires connect to the anodes of the optical sources. The scale bar represents $500\mu \mathrm{m}$. (b) and (c) The same array after the reconstruction and its sketch. The optical sources irradiate Rosenthal’s canal. A thin layer of tissue can be seen around the electrode, which has been slightly retracted (dash line in the sketch). The organ of Corti (OC) is also marked. The scale bar represents $500\mu \mathrm{m}$ and is used for the following panels. (d) and (e) The cross section of the same array after the reconstruction and its sketch. The optrode (Op), tissue, bone, and OC are marked in the sketch.

3.3.

In Vivo Functional Testing in the Guinea Pig Animal Model

SELDs were assembled into arrays, which could be inserted into the guinea pig cochlea for in vivo functional testing. At an input current of 600 mA, the maximum radiant energy emitted from the SELDs was $15.3\pm 4.9\mu \mathrm{J}$ per pulse ($n=10$), ranging from 8 to $20\mu \mathrm{J}$ per pulse. The energy of each SELD will be screened in future assemblies of the optrodes to eliminate the variation. The energy was measured in air with a Coherent J-50-LP-1A energy sensor. The SELDs arrays were inserted through the cochleostomy to stimulate the base of the cochlea. Optical pulses were $100\mu \mathrm{s}$ in duration and were delivered at 100 Hz. During the experiments, the test current levels were increased in 6 equal steps from 0 to 600 mA [Fig. 8(a)]. The amplitude of the CAPs dropped with decreasing current amplitude and reached a minimum at 300 mA [Fig. 8(a)]. CAPs also disappeared when the SELD arrays were rotated in or partially extracted from the cochleostomy so that the emitting side was no longer facing the SGNs.

Fig. 8

The traces show CAPs evoked with an SELD and an optical fiber in the same animal (guinea pig). (a) CAP responses evoked by a SELD operated at different current levels. The energy output measured prior to the in vivo test was $20\mu \mathrm{J}/\mathrm{pulse}$ with 600-mA current input. The four traces represent the current input of 600, 500, 400, and 300 mA. The traces are the averaged responses to 20 stimulus presentations. (b) CAPs evoked by delivering the radiant energy with an optical fiber at different energy levels. The traces are the averaged responses to 100 stimulus presentations. (c) The CAP amplitudes at different radiant energies for both the SELD and the optical fiber.

CAP amplitudes in response to INS, which were obtained with the SELDs, were compared to CAP amplitudes measured in the same animal by delivering the radiation with an optical fiber [Figs. 8(b) and 8(c)] coupled to a table-top laser (Lockheed Martin Aculight Corp., Bothell, Washington).

3.4.

In Vivo Test in Cats

For the cochlear stimulation devices, it is important that they are both biocompatible and can evoke neural responses. In cat cochleae, the power of the VCSEL array was too low to evoke measurable ABRs. The results of the chronic experiments demonstrate that the optrodes are biocompatible and that chronic implantation does not damage cochlear function over time. Functional tests with more powerful SELDs were done in the guinea pigs as shown previously.

After the in vitro long-term testing, the optrodes were implanted into the left cochleae of a cat. Figures 4 and 5 show the optrode insertion and the transcutaneous connector to the current source. The current sources were small computer controllable laser diode drivers developed by Lockheed Martin Aculight (LMA), which fit into the backpack of a cat. Alternatively, a commercially available high-power precision source, LDX-32400 (ILX Lightware, Bozeman, Montana), was used. Optically and acoustically evoked ABRs were recorded before implantation (baseline) and two weeks after the surgery (Fig. 9). Thereafter, acoustically evoked ABRs were measured every two weeks up to 26 weeks after the surgery. Note that the cat study aimed to determine the biocompatibility of the optrodes and to determine whether implantation and materials will lead to a deterioration of cochlear function over time. Cat number 13IKB3 was implanted with red VCSELs. Cats 13IMR3 and 13CKC6 were implanted with the low-power infrared VCSELs. 13IKB3 was excluded from measuring the optical ABRs (oABRs). No oABRs were evoked by the low-power VCSELs. In pulsed operation mode, the highest output power of low-power VCSELs is about 70 mW ($7\mu \mathrm{J}/\mathrm{pulse}$), which is about the radiant energy required to reach stimulation threshold as determined in previous experiments.^18,32 Sound levels to evoke an ABR with acoustic clicks were elevated immediately after surgery by about 50, 35, and 25 dB in cat 13IKB3, 13IMR3, and 13CKC6, respectively. Thresholds were determined two weeks after the surgery. No further changes in threshold were observed after the placement of the optrode [Fig. 9(a)]. Threshold elevations occurred mostly at frequencies above 22 kHz, where the optrode was located [Fig. 9(b)]. No acoustic responses could be recorded for stimulation frequencies above 22 kHz in all three cats, and sound levels to reach the threshold for an ABR were elevated at frequencies between 8 and 22 kHz by 25 dB in cat 13IMR3.

Fig. 9

The ABR thresholds to acoustic stimuli pre- and postimplantation. (a) Click evoked ABR thresholds at different time points in three cats. Click thresholds were elevated after optrode implantation, but then remained consistent in the months after; (b) pure tone evoked ABR thresholds at different times after implantation for animal 13IMR3. The optrode implantation caused high-frequency hearing loss (at 32 kHz) and elevated thresholds among 8 to 22.6 kHz, but little change was noted below 8 kHz.

4.1.

Requirements for the LDS

4.2.

Enabling Technology