Vision Research 49 (2009) 825–833

Contents lists available at ScienceDirect

Vision Research

journal homepage: www.elsevier .com/locate /v isres

Focal activation of the feline retina via a suprachoroidal electrode array

Y.T. Wong a,d, S.C. Chen a,b, J.M. Seo c, J.W. Morley b, N.H. Lovell a, G.J. Suaning a,*

a Graduate School of Biomedical Engineering, University of New South Wales, Sydney NSW 2052, Australiab School of Medicine, University of Western Sydney, NSW 1797, Australiac Department of Electrical Engineering, Seoul National University, Seoul, Republic of Koread Center for Neural Science, New York University, New York, NY 10003, USA

a r t i c l e i n f o

Article history:Received 22 August 2008Received in revised form 17 February 2009

Keywords:Retinal prosthesisVisual cortexCatElectrical stimulation

0042-6989/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.visres.2009.02.018

* Corresponding author.E-mail address: g.suaning@unsw.edu.au (G.J. Suan

a b s t r a c t

This paper presents the results of the first investigations into the use of bipolar electrical stimulation ofthe retina with a suprachoroidal vision prosthesis, and the effects of different electrode configurations onlocalization of responses on the primary visual cortex. Cats were implanted with electrodes in the supra-choroidal space, and electrically evoked potentials were recorded on the visual cortex. Responses wereelicited to bipolar and monopolar stimuli, with each stimulating electrode coupled with either six-returnelectrodes, two-return electrodes, or a single-return electrode. The average charge threshold to elicit aresponse with bipolar stimulation and six-return electrodes was 76.47 ± 8.76 nC. Bipolar stimulationusing six-return electrodes evoked responses half the magnitude of those elicited with a single or two-return electrodes. Monopolar stimulation evoked a greater magnitude, and area of cortical activation thanbipolar stimulation. This study showed that suprachoroidal, bipolar stimulation can elicit localized activ-ity in the primary visual cortex, with the extent of localization and magnitude of response dependent onthe electrode configuration.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Microelectronic vision prostheses aim to restore visual perceptsthrough electrical stimulation of the surviving visual pathways inthose that have been blinded through photoreceptor loss (for re-view, see Dowling, 2005). Intense research into these prostheseshas been underway for over half a century, however the optimalsite of implantation of stimulating electrodes has been, and re-mains a topic of intense debate. Visual sensations have been elic-ited through stimulation of the lateral geniculate nucleus (Pezaris& Reid, 2007), the superior colliculus (Nashold, 1970), the visualcortex (Brindley & Lewin, 1968; Dobelle & Mladejovsky, 1974;Marg & Dierssen, 1965), the optic nerve (Brelén et al., 2006) andsites on (epiretinal) (Humayun et al., 2003), under (subretinal)(Zrenner et al., 2007), or near the retina (suprachoroidal) (Kandaet al., 2004; Zhou et al., 2008).

The well-defined retinotopic mapping and easier surgical accessof the retina, provide compelling reasons why this should be thesite of prosthetic intervention for diseases such as retinitis pigmen-tosa when the photoreceptors degenerate, but other retinal ele-ments remain intact (see Santos et al., 1997, but also Jones &Marc, 2005). However, the distinctly-layered structure of the eyelends itself to multiple potential sites for electrical stimulation,each with its advantages and disadvantages. Implantation of elec-

ll rights reserved.

ing).

trodes onto the epiretinal surface is complicated by the delicatenature of the retinal tissue; the high shear forces associated withthe mass of an electrode array and the relatively high accelerationsof the eye; complications in electrode fixation; and blockage ofvisualization of the stimulated tissue (Gerding, Benner, & Taneri2007; Guven et al., 2005; Majji et al., 1999). Subretinal implanta-tion is complicated by the need to partially detach the retina fromthe retinal pigment epithelium (at least) at the electrode site; theneed to penetrate the retina from an intra-ocular approach therebyrisking further retinal detachment; or the need to penetrate thechoroid from an extra-ocular approach thereby risking choroidalhemorrhage; and the need to intervene soon after photoreceptordeath, before retinal remodeling reduces the outer nuclear andplexiform layers to a fibrotic glial seal (Jones & Marc, 2005; Sachset al., 2005).

Suprachoroidal electrode placement offers a number of distinctplacement advantages. These include a relatively simplified surgi-cal approach that requires only a small, transverse incision a fewmillimeters from the limbus; easily controlled insertion depth; ri-gid and firm placement and fixation of the electrode with the col-lagenated scleral wall on one side, and choroid on the other;reduced risk of choroidal hemorrhage; and isolation of electrodesfrom direct contact with delicate, retinal tissue. Perhaps, mostimportantly, this approach lends itself to a robust and stable,life-long implantation (Zhou et al., 2008). These advantages how-ever come at the cost of a possible reduction in the maximum spa-tial resolution that a suprachoroidal prosthesis may provide,

826 Y.T. Wong et al. / Vision Research 49 (2009) 825–833

compared to that theoretically achievable with epiretinal and sub-retinal prostheses.

To-date, studies involving suprachoroidal electrode placementhave all utilized monopolar stimulation in rabbits and rats with re-turn electrodes placed in the vitreous, or on the sclera (Kanda et al.,2004; Yamauchi et al., 2005; Zhou et al., 2008). This study is thefirst to investigate the efficacy of suprachoroidal, bipolar stimula-tion. In addition, hundreds (if not thousands) of electrodes arerequired to allow implantees to perform such tasks as readingand facial recognition (Hallum, Chen, Preston, Suaning, & Lovell,2005). With this many electrodes, monopolar stimulation, and con-ventional bipolar stimulation, (where each stimulating electrode ispaired with one-return electrode), will result in complex spatio-temporal interactions between electrodes resulting in an inabilityto elicit multiple discrete phosphenes (Horsager, Weiland,Greenberg, Humayun, & Fine, 2008; Loudin et al., 2007; Rizzo,Wyatt, Loewenstein, Kelly, & Shire, 2003). The authors posit thatsurrounding each stimulating electrode with six-return electrodeswill reduce the electrical cross-talk between stimulation sites,resulting in a more focused response (Lovell, Dokos, Cloherty,Preston, & Suaning, 2005; Wong et al., 2007). The effect on corticalevoked potentials of six-return electrode stimulation compared toone-return and two-return electrode stimulation was investigated.

2. Methods

2.1. Animal preparation

All experiments were conducted with prior approval from theAnimal Care and Ethics Committee at the University of NewSouth Wales, and in accordance with the National Health andMedical Research Council of Australia guidelines for animalexperimentation.

Adult cats (n = 6) were anesthetized through an intra-muscularinjection of ketamine hydrochloride/xylazine hydrochloride(20 g kg�1/1 mg kg�1). The animals were given atropine sulfate(0.2 mg kg�1, sc) to reduce mucosal secretions, and dexamethasone(1.5 mg kg�1, im) to reduce cerebral swelling. The femoral arteryand vein were catheterized to allow blood pressure to be continu-ously monitored and to deliver fluids (compound sodium lactatesolution, and glucose 2.4 ml kg�1 h�1). A tracheotomy was per-formed and animals were ventilated with 70% N2O/30% O2, andanesthesia was maintained with 0.5–1% halothane. Both eyes ofthe animals were implanted with an electrode in the suprachoroi-dal space. In one animal, two electrode arrays were implanted intothe same eye, with the heads of the electrode arrays separated byapproximately 5 mm.

Fig. 1. (a) The 14-channel platinum stimulating electrode array tip. Electrodes are groupelectrode can be used to stimulate in three different electrode configurations, (top) selectrode stimulation. These configurations can be moved so that any electrode can act

The primary visual cortex was exposed through a craniotomybetween the Horsley–Clarke coordinates +9 to �4 (anterior/poster-ior) and +0.5–+6 (medial/lateral), and the dura mater removed. Thepupils of the animals were dilated by topical application of a mix-ture of phenylephrine hydrochloride (10%), atropine sulfate (1%),tropicamide (1%), and cyclopentolate (1%). Recordings from thevisual cortex were made over a period of three to four days, afterwhich the animals were euthanized with a lethal overdose of so-dium pentobarbital.

2.2. Implant technology and stimulation paradigms

Stimulating electrode arrays consisted of 14 planar platinumelectrodes embedded within a silicone rubber carrier (Schuettler,Stiess, King, & Suaning, 2005; Suaning, Schuettler, Ordonez, &Lovell, 2007). The electrodes were manufactured by initially spin-coating a 25 lm layer of silicone rubber onto a microscope slide,over which Pt foil (12.5 lm thick) is placed and patterned with aNd:YAG laser into the desired electrode shape. The excess Pt is re-moved, and another 25 lm layer of silicone is spin-coated over thepatterned Pt. The electrodes are then opened by laser micro-machining, and removed from the microscope slide. The electrodearray had a thickness of approximately 65 lm, and a maximumwidth at the array tip of 3.8 mm. Electrodes were organized intotwo hexagons (Fig. 1a) with the centers separated by 600 lm.The electrode contacts were circular with 230 lm diameter open-ings. The electrode diameter corresponds to a visual angle of 1.15�at the retinal surface of the cat, and an electrode density of oneelectrode per 3�. The distance between the centers of the two hexa-gons of electrodes corresponds to a visual angle of 11.85�. For threeof the experiments the top right electrode of the left hexagon, andtop left electrode of the right hexagon were replaced with triangu-lar electrodes (horizontal width 800 lm, vertical width 460 lm),for increased surface area, and hence the amount of charge theycould deliver.

To provide electrical stimulation, the electrode arrays were con-nected to the authors’ latest generation Application Specific Inte-grated Circuit neurostimulator (Wong et al., 2007). Primarily,bipolar stimulation was used with the stimulating and return elec-trodes both on the stimulating electrode array (Fig. 1a). However,to compare the effects of bipolar and monopolar stimulation, aspherical, 500 lm diameter, platinum electrode was inserted intothe vitreous to act as a monopolar return in three animals.

For bipolar stimulation, the effects of three different electrodeconfigurations were investigated. The first had each stimulatingelectrode surrounded by six-return electrodes, which effectivelyelectrically isolated each stimulation site (Fig. 1b), (i) (Lovellet al., 2005). Alternatively, each stimulating electrode was either

ed into two hexagons with the circular contacts with a diameter of 230 lm. (b) Theix-return electrodes, (middle) a single-return electrode, and (bottom) two-returnas the stimulating or return electrode.

Fig. 2. A 32-channel platinum recording electrode. Electrode contacts are 350 lmin diameter, and electrodes centers are spaced 1.27 mm apart on the short side and950 lm apart on the long side.

Y.T. Wong et al. / Vision Research 49 (2009) 825–833 827

coupled with a single-return electrode (Fig. 1b), (ii) or with twoadjacent return electrodes (Fig. 1b), (iii) Both monophasic andbalanced, cathodic-first, biphasic waveforms were utilized, withinter-stimulus shorting to a common ground (floating relative tothe animal) used to minimize charge imbalance at the electrode-tissue interface (Wong et al., 2007). Unless otherwise stated, all re-sults for biphasic stimulation were elicited using single pulsesdelivered with both the anodic and cathodic pulse widths fixedat 400 ls, and the stimulus amplitude varied from 10 lA to1 mA. Cathodic and anodic pulses were separated by a constant10 ls delay.

2.3. Surgical implantation

To insert the electrode array into the cat’s eye, a scleral inci-sion approximately 4.5 mm wide and half the thickness of thesclera was made 7 mm behind and parallel to the corneal limbususing a 15� Stab blade (Alcon Inc., H}unenberg, Switzerland). Thisincision was extended 1–2 mm perpendicular to the limbus cre-ating a short tunnel through the sclera, made using a shortcutblade (3.75 mm Angled Satin Shortcut Blade, Alcon Inc.). Thistunnel was then extended down to the choroid by cuttingthrough the remaining sclera at a 45� angle using a 15� stabblade.

The electrode array was then placed on a custom-made 95 lmthick transparent polyvinyl acrylate film (Copier TransparencyFilm, Corporate Express Aust. Ltd., Australia) inserter, and held inplace with sodium hyaluronate (10 mg ml�1 ProVisc, Alcon Inc.).The electrode and inserter were fed through the tunnel into thesuprachoroidal space, and then fed 17 mm into the eye perpendic-ular to the limbus, so that the head of the array (Fig. 1a) was posi-tioned at the posterior pole near the fovea. The inserter was thenremoved, leaving the electrode array in contact with the surfaceof the choroid. The electrode array was held against the choroidby the sclera resulting in a stable and consistent distance fromthe retinal neurons. The scleral tunnel was repaired at the sidesof the electrode array. An overhanging suture over the externalsections of the electrode array onto the sclera was made to avoidexcess strain on the electrode array. The inserter was the samewidth as the electrode array and provided the rigidity necessaryto separate the choroid and sclera.

2.4. Recording setup

Electrically evoked potentials (EEPs) were recorded on the pri-mary visual cortex using a 32-channel surface electrode arraymanufactured in the same manner as for the stimulating electrode.Recordings were made differentially to a single indifferent plati-num electrode placed subcutaneously within the muscles in theneck. The recording array was organized in a 4 � 8 square grid withcenter to center electrode spacings of 1.27 mm (vertically), and950 lm (horizontally). Electrodes on this array were circular, with350 lm diameter openings (Fig. 2).

The electrodes were connected to a 32-channel headstage(NN32AC-Z, Z-Series 32-Channel Neuronexus Headstage Adapter,Tucker Davis Technologies (TDT), Alachua, FL, USA), which thenconnected to an amplifier (�10,000) and analog-to-digital conver-tor (PZ2-256, TDT). Signals were sampled at 25 kHz and low-passfiltered at 10 kHz and high-pass filtered at 1 Hz. An ensemble aver-age of 300 runs at a repetition rate of 997 ms was taken to reducethe noise level. The amplitude of the initial peak of the evoked re-sponses was calculated by finding the difference between theamplitude of the first peak, and the amplitude of the pre-stimulussignal. The threshold for an evoked response was taken as the stim-ulus intensity that yielded a first peak twice the amplitude of thebackground noise.

To assess the viability of the recording arrangement, visuallyevoked potentials (VEPs) in response to a stroboscopic flash (BotexSuper Strobe SP-106MR, NCW Enterprises Co. Ltd., Kowloon Bay,HK) were also recorded using the same electrode array and record-ing setup. An ensemble average of 100 runs was taken to increasethe signal-to-noise ratio.

2.5. Statistical analyses

The amplitude of the initial peak of the evoked response foreach of the 32-electrodes on the recording array were plottedfor different return electrode configurations, and for differentcharge injection values. Analysis of Variance (ANOVA) was usedto assess the significance of charge injection and electrode con-figuration on the amplitude of the initial peaks of the evoked re-sponses. ANOVA was performed on the recorded responsesacross the electrode array to 16 different charge injection andelectrode configurations.

The initial peaks of the response for the 32-electrodes for two-and single-return electrode stimulation were then averaged andnormalized to the corresponding response to six-return stimula-tion for the same charge injection. Student’s two-tailed paired t-test was used to test the hypothesis that the average normalizedresponse to two-electrode and single-electrode return stimulationwas greater than one, and hence greater than the average responsefor six-return stimulation. Finally, changes in recorded responsesto different stimulation paradigms were tested for significanceusing a two-sample Kolmogorov–Smirnov (KS) test. Statisticalanalyses were performed through MATLAB (R14, The MathWorks,Inc., MA, USA) with 95% confidence intervals. Results are expressedas the mean ± standard error of the mean (SEM) except where other-wise noted.

3. Results

3.1. Cortical evoked potentials

VEPs and EEPs were elicited from all animals. Overall, the re-sponses showed little variation over 3–4 days. There were no ob-servable differences in responses elicited with the electrodearrays with triangular and circular return electrodes and thosewith just circular electrodes. Fig. 3 shows a typical electricallyevoked response recorded from the 32-channel cortical array.The EEPs recorded on a single electrode from the 32-channel ar-ray to increasing stimulus charge is shown in Fig. 4. All stimuliin Fig. 4 were delivered with each stimulating electrode sur-

Fig. 3. EEPs resulting from a balanced biphasic, bipolar stimuli of 327 lA and apulse width of 400 ls (130.8 nC) with six-return electrodes. The waveforms arepresented in the same spatially arranged mapping as they were recorded (seeFig. 2). The left four waveforms have been removed for clarity due to excess noiseaffecting these recordings.

Fig. 4. EEPs increasing in amplitude in response to biphasic, bipolar stimuli of107 lA (42.8 nC)(subthreshold, gray line), 245 lA (98 nC), 327 lA (130.8 nC), and512 lA (204.8 nC). In each case the pulse width was 400 ls and each stimulatingelectrode was surrounded by six-return electrodes. The stimulus artifact is visibleapproximately 8 ms after the onset of the recording.

Fig. 5. The amplitude of the peaks of the EEPs in response to two differentstimuli with two electrode arrays separated at the retina by approximately5 mm, are plotted as circles with proportional areas. The response to a 213 lA(85.2 nC) stimulation with (a) Electrode 1, and with (b) Electrode 2, and a 314 lA(125.6 nC) stimulation with (c) Electrode 1, and (d) Electrode 2. All stimuli werebiphasic, bipolar, with a pulse width of 400 ls and each stimulating electrodewas surrounded by six-return electrodes.

Fig. 6. Strength–duration curves for three animals are indicated with ‘‘+”, ‘‘s” and‘‘x” markers. The rheobase was found to be 20 lA, with a chronaxie of approxi-mately 1650 ls. A line has been fitted to the data in the form I = Ir /(1� e�ln(2).s/c)where I is the stimulus current, Ir is the rheobase current, s is the stimulus duration,and c is the chronaxie.

828 Y.T. Wong et al. / Vision Research 49 (2009) 825–833

rounded by six-return electrodes, and as biphasic stimulations.As the amount of charge injected per stimulus was increased,the amplitude of the response increased.

In Fig. 5, the amplitudes of the first peak of the EEPs recorded ateach electrode of the array have been plotted with circles in whichthe area is proportional to the amplitude. This is presented tohighlight the extent of the spread of activity across the cortex,and to allow for visual comparisons of the effect of different stim-ulus configurations. Stimulation of electrodes separated by approx-imately 5 mm in the eye, resulted in a shift in the focus ofactivation in the cortex (Fig. 5a and b). As the stimuli increasedin charge delivered, neural activation spread across the cortex(Fig. 5c and d). All stimuli were biphasic, bipolar waveforms withsix-return electrodes.

The average latency to the first peak when near threshold bi-phasic stimuli and six-return electrodes were used for all the ani-mals was 15.6 ± 0.94 ms, and this decreased with increasingstimulus charge (approximately three times threshold) to

10.5 ± 0.62 ms. The average of the electrode-tissue impedancesacross experiments for a single stimulating electrode and six-re-turn electrodes, measured within 24 h of the electrode insertionwas 14.5 ± 1.15 kX at 1 kHz. This was measured using a 40 lA sinewave.

Fig. 7. The responses to biphasic, bipolar stimuli with six-return electrodes of (a)71 lA (28.4 nC), and (b) 90 lA (36 nC), and biphasic, monopolar stimuli of (c) 71 lA(28.4 nC), and (d) 90 lA (36 nC). All stimuli had pulse widths of 400 ls. For themonopolar stimulation, the return was placed in the vitreous.

Y.T. Wong et al. / Vision Research 49 (2009) 825–833 829

3.2. Stimulus threshold

The average charge injection at threshold required to elicit anEEP for biphasic, bipolar stimulation with six-return electrodes

Fig. 8. The EEP responses to stimuli of 54 lA, 400 ls (21.6 nC) utilizing (a) biphasic, bipwith both the rightmost and top right electrodes, as returns (c) biphasic, bipolar stimulathe top right electrode as a return, and (e) monophasic, bipolar stimulation with all six elelectrode arrangement, and the stimulus waveform.

was 76.47 ± 8.76 nC, when pulse widths of 400 ls were used. Thisequates to a charge density of 184.05 lC cm�2 (through the singlestimulating electrode). The average threshold for biphasic, mono-polar stimulation was 26.27 ± 6.33 nC for 400 ls pulse widths,which equates to a charge density of 63.22 lC cm�2. A strength–duration curve for three animals when biphasic stimulation withsix-return electrodes was used is shown in Fig. 6, and the rheobasewas 20 lA and chronaxie approximately 1.65 ms.

3.3. Responses to stimulus paradigms

For identical biphasic charge injections, monopolar stimulationevoked a greater extent and amplitude of neural activity (Fig. 7cand d), than that of bipolar stimulation with six-return electrodes(Fig. 7a and b). This difference in response diminishes with highercharge injections as the evoked cortical potentials reach a maxi-mum. The KS test showed that the response to the 28.4 nC mono-polar stimulation (Fig. 7c) was significantly different to 28.4 nCbipolar stimulation (Fig. 7a) (P < 0.001), however the 36 nC mono-polar stimulation (Fig. 7d) was not significantly different to the36 nC bipolar stimulation (Fig. 7b). When using biphasic bipolarstimulation, six-return electrodes (Fig. 8a) resulted in a more fo-cused cortical response compared to a single-return electrode(Fig. 8c and d), evidenced by the reduced spread and magnitudeof the activity on the cortex. The KS test showed that responseselicited using a single-return electrode (Fig. 8c and d) were signif-icantly different to the response elicited using six-return electrodes(Fig. 8a) (P < 0.05). When two electrodes were used as returns(Fig. 8b), activity was again recorded from a larger area of the cor-tex, and the magnitude of the response was slightly greater thanfor six-returns. Finally, monophasic, bipolar stimulation pulses(Fig. 8e) were shown to elicit a response from a greater area ofthe cortex when compared to biphasic, bipolar pulses. The in-

olar stimulation with all six electrodes as returns, (b) biphasic, bipolar stimulationtion with the right most electrode as a return, (d) biphasic, bipolar stimulation withectrodes as returns. Next to each figure is a schematic of the stimulating and return

Fig. 9. The peak amplitude of the EEPs for the 32-electrodes on the recording arrayfor four different return configurations at different charge injections. All stimuliwere balanced biphasic, bipolar stimulation.

830 Y.T. Wong et al. / Vision Research 49 (2009) 825–833

creased area and magnitude of activity in response to the mono-phasic stimulation and two-return electrode stimulation werenot significantly different to the six-return stimulation.

Fig. 9 shows the peak responses for different charge injectionvalues, for four different stimulation paradigms. The peak responsewas calculated by averaging the amplitude of the first peak acrosseach electrode of the electrode array. Typically, responses were re-corded on a subset of the 32-electrodes, with responses increasingto a maximum on one or two electrodes on the array. The averageresponse across the array was taken to give an indication of the to-tal evoked response.

Fig. 10. Bars show the average of the peak EEP responses to two-return and single-return electrode stimulation normalized to the peak EEP response to six-returnelectrode stimulation at the same charge injection. Responses to charge injectionvalues (varied from 14.4 nC to 161.2 nC), and presented to four animals were usedto generate the averages. All responses were to biphasic, bipolar stimuli. Error barsindicate SEM. Stars over bars indicate p-values when there was a significantdifference between means (two-tailed, paired t-test): *P < 0.0001. There was nosignificant difference between the means of the two-return responses and one-return responses (P = 0.28).

ANOVA found that charge injection was a significant factor tothe variation in response amplitude (P < 0.0001), as was electrodeconfiguration (P < 0.0001). With each electrode configuration, theresponse over the 32-channel array increased with increasingcharge injection. The responses also increased in magnitude whenelectrode stimulation configurations were changed from six-returnstimulation, to two-electrode return stimulation to finally single-return electrode stimulation.

The presented results in this section have to this point been fora subset of the charge injection values, and for individual animals.These results were presented as there was great variation in theamplitude of the responses across animals. This variation wasdue to slight differences in the stimulating and recording electrodelocations. The results do however show the typical changes in re-sponse due to changes in electrode configurations for all animals.To see the effect of different charge injection values over multipleanimals, responses to two and single-return stimulation were nor-malized to the average response to six-return electrodes stimuli ofthe same charge injection. This normalization allows for the quan-titative analysis of the change in evoked response due to changesin electrode configuration alone. Fig. 10 shows the average of re-sponses across animals to two and single-return stimulation, nor-malized to the average response to six-return electrodes withstimuli of the same charge injection that varied from 14.4 nC to161.2 nC.

The response for two-return electrode stimulation was found tobe on average 1.89 ± 0.18 times greater than for six-return elec-trode stimulation (P < 0.0001). Single-return electrode stimulationwas on average 2.18 ± 0.19 times greater than six-return electrodestimulation (P < 0.0001). However, the average response to single-return stimulation was not significantly greater than the averageresponse to two-return electrode stimulation (P = 0.28).

4. Discussion

4.1. Cortical evoked potentials

This study showed that biphasic, bipolar electrical stimulationfrom the suprachoroidal space can elicit EEPs. The thresholdsfound for cats in this paper (76.47 ± 8.76 nC) are in a similar rangeto that of suprachoroidal monopolar stimulation reported for rab-bits by Yamauchi et al. (2005) (150 ± 122 nC, biphasic pulses with1 ms pulse durations), by Sakaguchi et al. (2004) (33 ± 16.1 nC, bi-phasic pulses with 500 ls pulse durations), by Nakauchi et al.(2005) (27.5 ± 5 nC, biphasic pulses with 500 ls pulse durations),and by Zhou et al. (2008) (varied from 128 nC on the day of implan-tation to 80 nC six weeks after implantation, found with biphasicpulses with 1 ms pulse durations). The differences in reportedthresholds can be attributed to the variability in the stimulatingand recording electrode placements and the different animalsused. Cats were used in this study as their ocular circulation andanatomy make them a closer model to humans than rabbits (Bill,1975; Henkind, 1966), and their visual system is a well used modelof the mammalian visual system.

The threshold value we report here equates to a charge densityof 184.05 lC cm�2 for an electrode with a flat surface area. How-ever, due to the roughening that occurs in the manufacture of theseelectrodes (roughening of approximately 2.08 times reported bySchuettler, 2007) the charge density is closer to 88.49 lC cm�2.This value is close to the safe charge limitations that have been re-ported for platinum (100–150 lC cm�2 for pulse durations of0.2 ms (Rose & Robblee, 1990), and 300–350 lC cm�2 for pulsedurations greater than 0.6 ms (Brummer & Turner, 1977). Futureelectrode designs may require an increase in the electrode size toensure stimuli remain under the acceptable charge density values.

Y.T. Wong et al. / Vision Research 49 (2009) 825–833 831

The average latency to the first peak of the electrically evokedresponse was 15.6 ± 0.94 ms, which is the same latency as reportedby Nakauchi et al. (2005) (15.7 ± 2 ms) but longer than reported bySakaguchi et al. (2004) (9 ± 1 ms). It has been shown that pulsewidths longer than 1 ms preferentially activate the bipolar andamacrine cells while shorter pulses activate the ganglion cells(Fried, Hsueh, and Werblin, 2006). This however does not explainthe discrepancies in latencies as all three were recorded inresponse to similar stimulus pulse durations, 400 ls in this studyand 500 ls in the two other reported studies. The discrepancymay in fact be due to the different retinal cell densities of theanimals tested. The cats used in this study and the pigmentedrabbits in the Nakauchi et al. study have a greater retinal celldensity than the albino rabbits used in the Sakaguchi et al. study(Donatien, Aigner, & Jeffery, 2002).

The EEP latency reported here was less than that measured forvisually evoked potentials (20–25 ms), which leads the authors tobelieve that electrical stimulation bypasses the photoreceptors,and stimulates the bipolar cells. With increasing charge injection,this latency decreases to a similar range (10.5 ± 0.62 ms) as re-ported for epiretinal stimulation responses (Eckhorn et al.(2006)), indicating a likely activation of ganglion cells.

4.2. Analysis of the bipolar, six-return suprachoroidal approach

One of the significant drawbacks of the suprachoroidal place-ment of electrodes is the need to stimulate through the choroid,the retinal pigment epithelium, and Bruch’s membrane (reportedto be a resistive and capacitive layer (Brindley, 1956; Brown &Tasaki, 1961)), and potentially through a fibrotic glial seal due toretinal remodeling (Jones & Marc, 2005). This may result in morediffuse responses and therefore a more limited maximum spatialresolution when compared to that theoretically achievable throughepiretinal and subretinal stimulation (Palanker, Vankov, Huie, &Baccus, 2005). The more limited spatial resolution will still be ofconsiderable value to the implant recipient in enabling them toperform tasks such as light and object localization and motiontracking (Yanai et al., 2007). To examine the effect of this remodel-ing, further experiments should be undertaken in retinal degener-ated animals such as those reported by Rah, Maggs, Blankenship,Narfstrom, and Lyons (2005) and Menotti-Raymond et al. (2007).The authors however, have shown that biphasic, bipolar stimula-tion in the suprachoroidal space can penetrate the choroid andBruch’s membrane, and elicit focused responses, removing theneed to breach the choroid and retina to place a monopolar return(Fig. 5a and b).

The thresholds for biphasic, bipolar stimulation were found tobe higher than that for biphasic, monopolar stimulation indicatingthat charge is being spread throughout the choroid, retinal pig-ment epithelium and the outer retinal structures. Both the choroidand retinal pigment epithelium have been reported to be highimpedance structures (Miller & Steinberg, 1977), with the tightjunctions of the retinal pigment epithelium the primary contribu-tor to this impedance (Cohen, 1965; Hudspeth & Yee, 1973). Formonopolar stimulation, a return electrode placed in the vitreousyields transretinal currents that effectively stimulate the survivingretinal cells. Bipolar stimulation with single or multiple returnelectrodes in the suprachoroidal space are likely to yield morearc-shaped shaped current distributions with a portion of the cur-rent flowing transretinally and the remaining flowing through theouter retinal structures. However, responses can still be elicitedwith bipolar stimulation, removing the need to breach the choroidor retina to place a return electrode in the vitreous body. This in-crease in threshold accounts for the larger extent of activity forthe monopolar stimulation when compared to bipolar stimulationof the same charge injection.

For bipolar stimulation, when six-return electrodes were used,responses were elicited from a more localized area than for singleand two-return electrode stimulation. This effect could result froma change in threshold value for single, and two-return electrodeconfigurations compared to six-return electrodes; however as thestimulating electrode did not change for these electrode configura-tions, the thresholds for the three stimulation paradigms were con-sistent. This leads to the conclusion that the use of biphasic, bipolarstimulation and six-return electrodes to capture charge was shownto elicit activity from spatially restricted populations of neurons,offering the ability to evoke smaller, more focused cortical activa-tion (Fig. 8). The six-return electrode configuration will be ofincreasing benefit when many stimulation sites are used simulta-neously, as the six-return electrodes effectively guard each stimu-lation site from electrical cross-talk. This electrical cross-talk hasbeen problematic in restoring multiple discrete phosphenes in hu-man trials (Horsager et al., 2008; Loudin et al., 2007; Rizzo et al.,2003). The guard effect of the six-return electrode configurationhas been shown to reduce cross-talk in saline bath experimentsand modeling studies (Lovell et al., 2005; Wong et al., 2007).

4.3. Comparison to other retinal approaches

The thresholds for suprachoroidal stimulation were higher thanthose reported for subretinal placement of electrodes; approxi-mately 20 nC for 50 lm diameter planar electrodes reported bySachs et al. (2005). This is likely due to the increased distance tothe neural tissue and will therefore result in increased power con-sumption of the implant. However, this may be offset by the re-duced long-term risk compared to the risk posed by electrodes inintimate contact with the retinal tissue, and also perhaps the rela-tive ease of implantation and fixation which offers a distinctadvantage.

For further comparison, epiretinal placement of electrodes incats have reported thresholds of 2–6 nC when electrodes are inintimate contact with the retinal surface (Schanze, Wilms, Eger,Lutz, & Eckhorn, 2002). However, poor electrode contact in thesame study resulted in thresholds greater than 100 nC. In anotherstudy by Hesse, Schanze, Wilms, & Eger (2000) no responses wererecorded with stimulations even at 200 nC. De Balthasar et al.(2008) also showed that stimulation thresholds in humans werecorrelated with the electrode distance to the retinal surface. Epire-tinal placement of electrodes, when intimate contact is achieved,will result in lower thresholds than for suprachoroidal placement,however as reported by other groups, this intimate contact and fix-ation of electrodes to the retina has been problematic, and is likelyto be even more difficult as electrode numbers increase. Intimatecontact and fixation of electrodes for suprachoroidal placement isconsiderably more straightforward with the electrode pushed to-wards the choroid by the scleral tissue.

While there are many advantages to epiretinal and subretinalelectrode placements, the prospect of multiple arrays being in-serted in the suprachoroidal space, covering a substantially-largerarea of the visual field with no apparent consequences in choroidalblood flow (Yamauchi et al., 2005), bulk transfer within the vitre-ous, cellular nourishment, and retina/electrode attachment, andfixation issues may prove suprachoroidal placement a viablealternative.

5. Conclusion

This study has shown that a retinal implant utilizing bipolarstimulation in the suprachoroidal space is feasible, and offers anumber of distinct advantages over other placement sites andmonopolar stimulation. These advantages arise due to the distance

832 Y.T. Wong et al. / Vision Research 49 (2009) 825–833

from the retinal tissue, which reduces the visual acuity that a ret-inal implant can provide. This may limit the implant to only pro-viding the ability to perform gross tasks such as objectlocalization and motion tracking. Thresholds to elicit EEPs in thecat were found to be similar to that reported in literature for supra-choroidal monopolar stimulation of the rabbit. In addition, the useof six-return electrodes surrounding each stimulating electrode,was shown to elicit a smaller amplitude and area of activation onthe cortex than stimulations with only a single-return electrode.While this is a promising step towards the realization of a supra-choroidal vision prosthesis, more work is needed to quantify thelocalization of retinal activation and the degree of interactionwhen stimulating multiple electrodes simultaneously.

Acknowledgments

The authors would like to acknowledge the help of Phil Preston,Chris Dodds, David Tsai, and Louis Jung in the running of the exper-iments, and manufacturing of electrodes. The authors would alsolike to acknowledge the input of Yves Kerdraon, Penny Allen, andMark McCombe on surgical procedures. This work was partiallysupported by the Australian Federal Government (NH & MRC andARC), Retina Australia, Korea Health 21 R&D Project A050251 andNBS-ERC.

References

Bill, A. (1975). Blood circulation and fluid dynamics in the eye. Physiological Reviews,55, 383–417.

Brelén, M. E., De Potter, P., Gersdorff, M., Cosnard, G., Veraart, C., & Delbeke, J.(2006). Intraorbital implantation of a stimulating electrode for an optic nervevisual prosthesis. Journal of Neurosurgery, 104, 593–597.

Brindley, G. (1956). The passive electrical properties of the frog’s retina and scleralfor radial fields and currents. Journal of Physiology, 134, 339–352.

Brindley, G. S., & Lewin, W. (1968). The sensations produced by electricalstimulation of the visual cortex. Journal of Physiology (London), 196, 479–493.

Brown, K., & Tasaki, K. (1961). Localization of electrical activity in the cat retina byan electrode marking method. Journal of Physiology, 158, 281–295.

Brummer, S. B., & Turner, M. J. (1977). Electrical stimulation with Pt electrodes: II –Estimation of maximum surface redox (theoretical nongassing) limits. IEEETransactions on Biomedical Engineering, 24, 440–443.

Cohen, A. I. (1965). A possible cytological basis for the R membrane in thevertebrate eye. Nature, 205, 1222.

De Balthasar, C., Patel, S., Roy, A., Freda, R., Greenwald, S., Horsager, A., et al. (2008).Factors affecting perceptual thresholds in epiretinal prostheses. InvestigativeOphthalmology and Vision Science, 49, 2303–2314.

Dobelle, W. H., & Mladejovsky, M. G. (1974). Phosphenes produced by electricalstimulation of human occipital cortex, and their application to the developmentof a prosthesis for the blind. Journal of Physiology, 243, 553–576.

Donatien, P., Aigner, B., & Jeffery, G. (2002). Variations in cell density in the ganglioncell layer of the retina as a function of ocular pigmentation. European Journal ofNeuroscience, 15, 1597–1602.

Dowling, J. (2005). Artificial human vision. Expert Review of Medical Devices, 2,73–85.

Eckhorn, R., Wilms, M., Schanze, T., Eger, M., Hesse, L., Eysel, U. T., et al. (2006).Visual resolution with retinal implants estimated from recordings in cat visualcortex. Vision Research, 46, 2675–2690.

Fried, S. I., Hsueh, H. A., & Werblin, F. S. (2006). A method for generating precisetemporal patterns of retinal spiking using prosthetic stimulation. Journal ofNeurophysiology, 95, 970–978.

Gerding, H., Benner, F. P., & Taneri, S. (2007). Experimental implantation ofepiretinal retina implants (EPI-RET) with an IOL-type receiver unit. Journal ofNeural Engineering, 4, S38–S49.

Guven, D., Weiland, J. D., Fujii, G., Mech, B. V., Mahadevappa, M., Greenberg, R., et al.(2005). Long-term stimulation by active epiretinal implants in normal andRCD1 dogs. Journal of Neural Engineering, 2, S65–S73.

Hallum, L. E., Chen, S. C., Preston, P. J., Suaning, G. J., & Lovell, N. H. (2005).Simulating prosthetic vision. Investigative Ophthalmology and Visual Science, 46,1522 (ARVO E-Abstract).

Henkind, P. (1966). The retinal vascular system of the domestic cat. ExperimentalEye Research, 5, 10–20.

Hesse, L., Schanze, T., Wilms, M., & Eger, M. (2000). Implantation of retinastimulation electrodes and recording of electrical stimulation responses in thevisual cortex of the cat. Graefe’s Archive for Clinical and ExperimentalOphthalmology, 238, 840–845.

Horsager, A., Weiland, J. D., Greenberg, R. J., Humayun, M. S., & Fine, I. (2008).Spatiotemporal integration of perceptual brightness in retinal prosthesis

patients. Investigative Ophthalmology and Visual Science, 49, 3011 (ARVO E-Abstract).

Hudspeth, A. J., & Yee, A. G. (1973). The intercellular junctional complexes of retinalpigment epithelia. Investigative Ophthalmology and Visual Science, 12, 354–365.

Humayun, M. S., Weiland, J. D., Fujii, G. Y., Greenberg, R., Williamson, R., Little, J.,et al. (2003). Visual perception in a blind subject with a chronic microelectronicretinal prosthesis. Vision Research, 43, 2573–2581.

Jones, B. W., & Marc, R. E. (2005). Retinal remodeling during retinal degeneration.Experimental Eye Research, 81, 123–137.

Kanda, H., Morimoto, T., Fujikado, T., Tano, Y., Fukuda, Y., & Sawai, H. (2004).Electrophysiological studies of the feasibility of suprachoroidal transretinalstimulation for artificial vision in normal and RCS rats. InvestigativeOphthalmology and Visual Science, 45, 560–566.

Loudin, J. D., Simanovskii, D. M., Vijayraghavan, K., Sramek, C. K., Butterwick, A. F.,Huie, P., et al. (2007). Optoelectronic retinal prosthesis: System design andperformance. Journal of Neural Engineering, 4, S72–S84.

Lovell, N. H., Dokos, S., Cloherty, S. C., Preston, P., & Suaning, G. J. (2005). Currentdistribution during parallel stimulation: Implications for an epiretinalneuroprosthesis. Proceedings of 27th annual international conference of the IEEEEMBS, Shanghai, China (pp. 5242–5245).

Majji, A. B., Humayun, M. S., Weiland, J. D., Suzuki, S., D’Anna, S. A., & de Juan, E. Jr,(1999). Long-term histological and electrophysiological results of an inactiveepiretinal electrode array implantation in dogs. Investigative Ophthalmology andVisual Science, 40, 2073–2081.

Marg, E., & Dierssen, G. (1965). Reported visual percepts from stimulation of thehuman brain with microelectrodes during therapeutic surgery. ConfiniaNeurologica, 26, 57–75.

Menotti-Raymond, M., David, V. A., Schäffer, A. A., Stephens, R., Wells, D., Kumar-Singh, R., et al. (2007). Mutation in CEP290 discovered for cat model of humanretinal degeneration. Journal of Heredity, 98, 211–220.

Miller, S. S., & Steinberg, R. H. (1977). Passive ionic properties of frog retinal pigmentepithelium. Journal of Membrane Biology, 36, 337–372.

Nakauchi, K., Fujikado, T., Kanda, H., Morimoto, T., Choi, J., Ikuno, Y., et al. (2005).Transretinal electrical stimulation by an intrascleral multichannel electrodearray in rabbit eyes. Graefe’s Archive for Clinical and Experimental Ophthalmology,243, 169–174.

Nashold, B. S. Jr, (1970). Phosphenes resulting from stimulation of the midbrain inman. Archives of Ophthalmology, 84, 433–435.

Palanker, D., Vankov, A., Huie, P., & Baccus, S. (2005). Design of a high resolutionoptoelectronic retinal prosthesis. Journal of Neural Engineering, 2, S105–S120.

Pezaris, J., & Reid, R. (2007). Demonstration of artificial visual percepts generatedthrough thalamic microstimulation. Proceedings of the National Academy ofSciences, 104, 7670–7675.

Rah, H. C., Maggs, D. J., Blankenship, T. N., Narfstrom, K., & Lyons, L. A. (2005). Early-onset, autosomal recessive, progressive retinal atrophy in persian cats.Investigative Ophthalmology and Visual Science, 46, 1742–1747.

Rizzo, J. F., Wyatt, J., Loewenstein, J. K., Kelly, S., & Shire, D. (2003). Perceptualefficacy of electrical stimulation of human retina with a microelectrode arrayduring short-term surgical trials. Investigative Ophthalmology and Visual Science,44, 5362–5369.

Rose, T., & Robblee, L. (1990). Electrical stimulation with Pt electrodes. VIII.Electrochemically safe charge injection limits with 0.2 ms pulses. IEEETransactions on Biomedical Engineering, 37, 1118–1120.

Sachs, H. G., Schanze, T., Wilms, M., Rentzos, A., Brunner, U., Gekeler, F., et al. (2005).Subretinal implantation and testing of polyimide film electrodes in cats. Graefe’sArchive for Clinical and Experimental Ophthalmology, 243(5), 464–468.

Sakaguchi, H., Fujikado, T., Fang, X., Kanda, H., Osanai, M., Nakauchi, K., et al. (2004).Transretinal electrical stimulation with a suprachoroidal multichannelelectrode in rabbit eyes. Japanese Journal of Ophthalmology, 48, 256–261.

Santos, A., Humayun, M. S., de Juan, E., Jr, Greenburg, R. J., Marsh, M. J., Klock, I. B.,et al. (1997). Preservation of the inner retina in retinitis pigmentosa. Amorphometric analysis. Archives of Ophthalmology, 115, 511–515.

Schanze, T., Wilms, M., Eger, M., Lutz, H., & Eckhorn, R. (2002). Activation in catvisual cortex evoked by electrical retina stimulation. Graefe’s Archive for Clinicaland Experimental Ophthalmology, 240, 947–954.

Schuettler, M. (2007). Electrochemical properties of platinum electrodes in vitro:Comparison of six different surface qualities. Proceeding of 29th annualinternational conference of the IEEE EMBS, 186–198.

Schuettler, M., Stiess, S., King, B. V., & Suaning, G. J. (2005). Fabrication ofimplantable microelectrode arrays by laser-cutting of silicone rubber andplatinum foil. Journal of Neural Engineering, S121–S128.

Suaning, G., Schuettler, M., Ordonez, J. S., & Lovell, N. (2007). Fabrication of multi-layer, high-density micro-electrode arrays for neural stimulation and bio-signalrecording. In 3rd international IEEE/EMBS conference on neural engineering (pp.5–8).

Wong, Y. T., Dommel, N. B., Byrnes-Preston, P. J., Lehmann, T., Hallum, L. E., Lovell, N.H., et al. (2007). Retinal neurostimulator for a multi-focal vision prosthesis. IEEETransactions on Neural Systems and Rehabilitation Engineering, 15, 425–434.

Yamauchi, Y., Franco, L. M., Jackson, D. J., Naber, J. F., Ziv, R. O., Rizzo, J. F. III,, et al.(2005). Comparison of electrically evoked cortical potential thresholdsgenerated with subretinal or suprachoroidal placement of a microelectrodearray in the rabbit. Journal of Neural Engineering, S48–S56.

Yanai, D., Weiland, J., Mahadevappa, M., Greenberg, R., Fine, I., & Humayun, M.(2007). Visual performance using a retinal prosthesis in three subjects withretinitis pigmentosa. American Journal of Ophthalmology, 143, 820–827.

Y.T. Wong et al. / Vision Research 49 (2009) 825–833 833

Zhou, J. A., Woo, S. J., Park, S. I., Kim, E. T., Seo, J. M., Chung, H., et al. (2008). Asuprachoroidal electrical retinal stimulator design for long-term animalexperiments and in vivo assessment of its feasibility and biocompatibility inrabbits. Journal of Biomedicine and Biotechnology, 2008.

Zrenner, E., Wilke, R., Zabel, T., Sachs, H., Bartz-Schmidt, K., Gekeler, F., et al. (2007).Psychometric analysis of visual sensations mediated by subretinalmicroelectrode arrays implanted into blind retinitis pigmentosa patients.Investigative Ophthalmology and Visual Science, 48, 659 (ARVO E-Abstract).