Long-Term Repeatability and Reproducibility of Phosphene Characteristics in Chronically Implanted Argus II Retinal Prosthesis Subjects

Long-term Repeatability and Reproducibility of Phosphene Characteristics in Chronically Implanted Argus II Retinal Prosthesis Subjects

YVONNE H-L. LUO, JOE JIANGJIAN ZHONG, MONICA CLEMO, AND LYNDON DA CRUZ

PURPOSE: Previously published literatures of acute (Second Sight Medical Products Inc., Sylmar, CA, USA) studies on few subjects have shown contradictory evi- retinal prosthesis system has been implanted in an dence on the reproducibility and characteristics of the increasing number of patients.1 However, despite this elicited phosphenes, despite using the same stimulating growing clinical use, data describing the features of artifi- parameters with epiretinal electrode arrays. In this study, cial vision perceived by the users remain scarce in the we set out to investigate the long-term repeatilibity and published literature.2–5 The idea of developing useful reproducibility of phosphenes in subjects chronically vision by epiretinal electrical stimulation hinges on the implanted with the Argus II retinal prosthesis (Second premise that stimulation with a single electrode gives rise Sight Medical Products, Inc., Sylmar, CA, USA). to a discrete focal percept in a retinotopic manner. DESIGN: Retrospective interventional case series and Simultaneous stimulation with multiple electrodes reliability study. therefore theoretically leads to perception of a pattern in METHODS: Six Argus II subjects of >5 years implanta- concordance with the pattern defined by the stimulating tion from a single site participated. The 4-electrode clus- electrodes.3 ter (‘‘quad’’) closest to fovea was stimulated in each In earlier studies, Rizzo and associates have called into subject with a fixed biphasic current. Perceived phos- question the consistency and reproducibility of phosphenes phenes were depicted relative to subjective visual field elicited by patterned epiretinal microelectrode stimula- center. The stimulus was applied at reducing time inter- tion.4 In a study involving 5 end-stage RP patients and 1 vals from 20 minutes to 1 second. Two sets of stimula- patient with normal retina, only 48% of the single- tions were performed on the same day and 2 further sets electrode stimulations and 32% of the multielectrode stim- repeated on a separate visit >1 week apart. ulations elicited visual percepts that matched the electrical RESULTS: Each subject depicted phosphenes of consis- stimulation patterns. Of the single-electrode stimulations, tent shapes and sizes, and reported seeing the same colors 3 subjects reported ‘‘a line’’ on some occasions, while ‘‘clus- with the fixed stimulating parameters, irrespective of the ters of 2 or 3 images’’ were seen on other occasions. In interstimuli intervals. However, there is a wide intersubject particular, the authors reported that only 66% (out of 99 variation in the phosphene characteristics. Four subjects stimulations) of the elicited visual percepts were reproduc- drew phosphenes in the same visual field quadrant, as ible in 3 RP patients on 2 separate trials, despite using the predicted by the quad-fovea location. Two subjects depicted same stimulating parameters to activate the same elec- phosphenes in the same hemifield as the expected locations. trodes. Such inconsistencies in the form and reproduc- CONCLUSION: Phosphenes for each subject were ibility of phosphenes would seriously undermine the consistently reproducible in all our chronically implanted formation of pixelated vision. subjects. This has important implications in the develop- In this study, we set out to investigate the consistency ment of long-term pixelated prosthetic vision for future and reproducibility of phosphenes elicited in a cohort of devices. (Am J Ophthalmol 2016;170:100–109. Ó subjects chronically implanted with the Argus II system 2016 Elsevier Inc. All rights reserved.) at a single site. All of the subjects described have had the device implanted and functioning for more than 5 years.

INCE ENTERING THE COMMERCIAL MARKET AS A retinal prosthetic device for the treatment for end- Ò S stage retinitis pigmentosa (RP), the Argus II METHODS

Accepted for publication Jul 20, 2016. SUBJECT INCLUSION/EXCLUSION CRITERIA: This is a From the NIHR Biomedical Research Centre for Ophthalmology (Y.H- L.L., J.J.Z., M.C., L.d.C.) and Vitreoretinal Service (Y.H-L.L., L.d.C.), single-center prospective study. All but 1 subject from Moorfields Eye Hospital NHS Foundation Trust, London, United our center implanted with the Argus II retinal pros- Kingdom. thesis system as part of the phase I/II clinical trial Inquiries to Yvonne H-L. Luo, NIHR Biomedical Research Centre for Ophthalmology, Moorfields Eye Hospital NHS Foundation Trust, 162 (clinicaltrials.gov identifier: NCT00407602) took part City Road, London EC1V 2PD, UK; e-mail: [email protected] in the study (n ¼ 6). One subject was excluded, as his

100 © 2016 ELSEVIER INC.ALL RIGHTS RESERVED. 0002-9394/$36.00 http://dx.doi.org/10.1016/j.ajo.2016.07.021 TABLE 1. Demographics and Operation Dates of the Participating Chronically Implanted Argus II Retinal Prosthesis Subjects

Year of Age at Time of Subject ID Diagnosis Operation Operation (y)

001 Retinitis pigmentosa 2008 70 003 Retinitis pigmentosa 2008 72 005 Retinitis pigmentosa 2009 55 006 Choroideremia 2009 66 007 Retinitis pigmentosa 2009 63 009 Retinitis pigmentosa 2009 45 device ceased to function after he developed retinal detachment and thick macular pucker as a result of a FIGURE 1. Red-free fundus photograph of the microelectrode fall. The participating subjects’ demographics and oper- array of a chronically implanted Argus II retinal prosthesis sub- ation dates are shown in Table 1.Thestudywas ject (ID: 007). The designated quad for stimulation consisted of electrodes E07E08F07F08 (enclosed in white square). The approved by the institutional review boards and ethics fovea location is estimated to be 15.5 degrees temporal and committee, and adhered to the tenets of the Declara- 1.5 degrees inferior to the center of the optic disc. The quad- tion of Helsinki. fovea relation is calculated from the estimated fovea location, to the center of the stimulated quad. SELECTION OF STIMULATING ELECTRODES AND PA- RAMETERS: For each subject, a cluster of 4 electrodes (hereinafter referred to as a ‘‘quad’’) closest to the fovea, which were functioning with thresholds within the safety was performed, to assess the contact between the stimu- charge density limit, was selected for stimulation, and the lating electrodes and the retinal surface. The selected elicited phosphenes characterized for the purpose of this quad, quad-retina relation, quad threshold, and stimulating study (Figure 1). current for each subject were as shown in Table 2. An estimated location of the fovea was made on the fundus photograph (taken at the outset of the study) and PHOSPHENES DEPICTION: To record the phosphenes was used for each subject as a reference point, measuring perceived by each subject, we constructed a wall covered 15.5 6 1.1 degrees from the center of the optic disc hori- with smooth-surface black mats. The subjects were first zontally and 1.5 6 0.9 degrees vertically.6 The foveal po- asked to stand up and stretch out both arms fully to touch sition was estimated, as there were no remaining features of the black wall, so that their shoulders were square, facing the fovea on color photographs, fluorescein angiograms, or the wall. The standing position of each subject was then optical coherence tomography (OCT) scans owing to se- adjusted so that the distance between the front of their vere end-stage RP. eyes and the wall equaled 30 cm. Next, the subjects were Once the designated quad was chosen, the stimulating asked to point with the index finger of both hands simulta- current for each subject was arbitrarily set to be 100 mA neously on the black wall, to where they believed the cen- above the threshold (measured within the last 6 months) ter of their visual field was, while keeping their head and initially, and then adjusted according to the strength of eyes pointing straight ahead. A stack of white A4-size response and comfort level reported by the subject. We papers (in landscape layout) was then placed underneath aimed to elicit a clear, definite visual percept without the index fingers of each subject and pinned to the wall, causing any discomfort or physical ‘‘tingling’’ sensation so that the index fingers were pointing at the center of for each subject. Default settings for the Argus II retinal the top sheet of paper (ie, the center of the paper was prosthesis system Clinical Fitting System (CFS) employed approximating the proclaimed center of each subject’s for device fitting and standard testing were likewise used for visual field). A drawing pin with a protruding cylindrical this study, which generate cathodic-first, charge-balanced head was then inserted at the point where their index fin- biphasic square waves to avoid tissue damage from charge gers contacted with the wall, so as to mark the location of build-up. These default waveform parameters were as fol- the proclaimed visual field center. lows: phase width of 0.46 ms, interphase duration of 0 sec- During the experiment, the subjects were asked to posi- onds, and total stimulation duration of 250 ms at the tion themselves according to the setup above, hold onto frequency of 20 Hz.7 Swept-source OCT (DRI OCT-1 the preplaced drawing pin head, and adjust their head Atlantis; TOPCON, Topcon Medical Systems, Tokyo, and eye position until they felt the center of their visual Japan) imaging through the chosen quad for each subject field was in alignment with the drawing pin. With each

VOL. 170 LONG-TERM PHOSPHENE CHARACTERISTICS IN ARGUS II SUBJECTS 101 TABLE 2. Phosphene Features Described by Each Chronically Implanted ArgusÒ II Retinal Prosthesis Subject, From Stimulating the Designated Quad With the Above Parameters

Quad-Retina Threshold Stimulating Phosphene Subject ID Quad Relation (on OCT) (mA) Current (mA) Phosphene Features Duration, t (s)

001 C07C08 In contact no 137 277 White filled-in circle 0.5 < t < 1 D07D08 significant ERMa 003 A07A08 In contact no 250 350 Electric blue filled-in circle t < 0.5 B07B08 significant ERMa 005 E05E06 In contact no 137 237 Bluish-gray vertical line, with t < 0.5 F05F06 significant ERMa fizzy vertical edges 006 A07A08 Quad-retina 371 552 Yellow ‘‘7’’ shape 0.5 < t < 1 B07B08 separation ¼ 377 mm; no ERM visible 007 E07E08 In contact no 24 124 Orange filled-in ring that ripples out 0.5 < t < 1 F07F08 significant ERMa 009 E07E08 In contact no 97 124 Orange horizontal lines x2, with fizzy 0.5 < t < 1 F07F08 significant ERMa brightness in between the lines

ERM ¼ epiretinal membrane; OCT ¼ optical coherence tomography. aIn images where the electrode array is in direct contact with the retinal surface, owing to the artefact caused by the acoustic shadow of the array, it is difﬁcult to ascertain the presence, if any, of mild epiretinal membrane. quad stimulation, the subjects were instructed to keep their nondominant hand on the drawing pin as a point of reference, while drawing on the paper with a marker pen the outline of the phosphene they perceived in relation to their visual ﬁeld center. A fresh sheet of paper was used for each phosphene depiction (Figure 2).

EXPERIMENT DESIGN: We set out to test the consistency and reproducibility of phosphenes within each subject when the selected quad was stimulated using the same settings, but at different time intervals between the stimulations. The experiments were divided into those with long interstimuli intervals and short interstimuli intervals. For the long-interval experiments, the subjects were asked to draw the perceived phosphenes at baseline, and then at subsequent time points whereby the interstimuli intervals were 20 minutes, 10 minutes, 5 minutes, 2.5 minutes, and 1 minute. For the short-interval experiments, the phosphenes were depicted at baseline, as well as at FIGURE 2. Color photograph showing a chronically implanted the following stimulations with these interstimuli intervals: Argus II retinal prosthesis subject depicting the phosphene he 30 seconds, 20 seconds, 10 seconds, 5 seconds, 2 seconds, perceived from the selected quad stimulation. The subject was and 1 second. Owing to built-in safety features in the Argus positioned with his shoulders square facing the wall and with II system proprietary software and delay in the radiofre- the eye-to-wall distance of 30 cm. The protruding blue drawing quency link transmission between the external and pin was positioned at the visual field center as indicated by the internal coil of the device, we could not reduce the intersubject. Before commencing each quad stimulation, the subject stimuli interval below 1 second. Each set of long-interval was first asked to hold onto the drawing pin with his nondomi- experiments and short-interval experiments was repeated nant hand and adjust his head and eye position until he felt his visual field center was in alignment with the drawing pin. once on the same day (4 sets of experiments per visit). When the quad was stimulated, he was instructed to keep his This was then repeated on a separate visit between nondominant hand on the drawing pin as a point of reference, 1 week and 1 month later. In total, each subject yielded 4 while drawing on the paper with a marker pen the outline of sets of data for the long-interval experiments, as well as 4 the phosphene perceived in relation to the visual field center. sets of data for the short-interval experiments, obtained A fresh sheet of paper was used for each phosphene depiction. over 2 separate visits. The aim of long-interstimuli-

102 AMERICAN JOURNAL OF OPHTHALMOLOGY OCTOBER 2016 FIGURE 3. Composite image shows phosphenes depicted by a chronically implanted Argus II retinal prosthesis subject (ID: 009) during an experiment of long-interstimuli-interval stimulations. The colors of the drawings reflected the specific interstimuli intervals that elicited the phosphenes. The centroid of each phosphene drawing is shown as a solid circle of the same color. The visual field center is shown as the black dot. The set centroid (ie, the average location of all the phosphene centroid points) is shown as a white circle and marked with an arrow.

interval experiments was to assess the reproducibility of the By selecting individual phosphene drawings, we were phosphenes over different time periods, while the short- able to calculate and compare the changes in the interstimuli-interval experiments allowed us to evaluate diagonal length of the closest-ﬁt rectangles across the the temporal resolution of the phosphenes up to 1 Hz stim- different drawings, which indirectly reﬂected the ulations. changes in drawing size.

DATA ANALYSIS: All the phosphene drawings were Location analysis and comparison. Analyses on phos- scanned in full size and stored as digital images for further phene locations were performed in 2 ways. First, we set computer processing and analysis. out to show the variability of these locations within each To assess the consistency and reproducibility of the de- set of experiments. Second, we were interested to find out pictions, we compared the phosphene drawings for vari- whether the locations of depicted phosphenes matched ability in (1) shape, (2) size, and (3) location within each the retinotopic orientation of the stimulated quad for subject. each subject. To facilitate location analyses, we calculated the Shape comparison. Adobe Creative Suite 5 Photoshop centroid for each of the phosphene drawings using a Python (Adobe Systems Inc., San Jose, CA, USA) was employed plug-in in the GIMP software version 2.8.8 Within each set for the initial image processing. All the scanned phosphene of experiments (containing phosphene drawings of drawings for each set of stimulations were imported as indi- different interstimuli interval stimulations), the centroid vidual layers in the program and superimposed at the point of all the phosphene centroid points was then calculated. of subjective visual field center (ie, the point marked with This set centroid represented the average location of the drawing pin). The color of each phosphene drawing was the perceived phosphenes for that set of stimulations altered to reflect the interstimuli interval of the stimulation (Figure 3). that elicited the phosphene. Such superimposition allowed To assess the variability in phosphene locations, the us to display and qualitatively compare the shapes of the distance between each phosphene centroid and the set drawings across different stimulation intervals, as well as centroid point was measured for each set of experiments. across different sets of experiments simultaneously. These variations in the distances could then be compared across different sets of experiments within Size comparison. To compare the variations in size of each subject as well as among different subjects. The the phosphene drawings, we employed the ‘‘scale tool’’ location of all 8 set centroid points for each subject rela- in the GNU Image Manipulation Program (GIMP) tive to the subjective visual field center could also be software version 2.8 (Kimball S, Mattis P and the GIMP compared, to show intrasubject variability across the Development Team. GIMP version 2.8.14. Available at: different sets of experiments. http://www.gimp.org/downloads). The ‘‘scale tool’’ auto- To further simplify the analysis, a summary centroid matically forms a closest-fit rectangle with the maximum location for all 8 sets of experiments for each subject horizontal and vertical dimension of any selected image. (hereinafter referred to as ‘‘final centroid,’’ later explained

VOL. 170 LONG-TERM PHOSPHENE CHARACTERISTICS IN ARGUS II SUBJECTS 103 FIGURE 4. The phosphene drawings for the ﬁrst set (out of 4 sets) of long-interstimuli-interval stimulations (Left panels) and those of the short-interstimuli-interval stimulations (Right panels) are displayed for each of the chronically implanted Argus II retinal prosthesis subjects. The color of each phosphene drawing reﬂects the interstimuli interval of the stimulation that elicited the depicted phosphene, rather than the actual color perceived by the subject.

104 AMERICAN JOURNAL OF OPHTHALMOLOGY OCTOBER 2016 FIGURE 5. Diagrams showing the centroids of the phosphenes depicted by each of the chronically implanted Argus II retinal prosthesis subjects. The set centroids (white circles) and the ﬁnal centroid point (centroid of the set centroids, denoted as a red circle) are shown relative to each subjects visual ﬁeld center (black dot).

in Figure 5) was calculated. This final centroid location subjects (005, 006, and 009) drew phosphenes of significant was used to compare with the expected phosphene loca- difference in size during long-interstimuli-interval stimulation calculated from the quad-fovea relation of each sub- tions (P ¼ .005, P ¼ .005, and P ¼ .01, respectively, ject’s fundus photographs (Figure 1), to assess retinotopic Friedman test), while 1 subject (007) drew phosphenes agreement. significantly different in size during short-interval stimulations (P ¼ .01). For the remaining 2 subjects, phosphene depictions for neither long nor short interstimuli interval RESULTS stimulations showed statistically significant size variation. Irrespective of the drawings, all the subjects reported that SHAPE COMPARISON: Each subject reported seeing phos- they perceived phosphenes of similar sizes with their given phenes of the same color and shape irrespective of the quad stimulations. interstimuli intervals. However, across the subjects, there is great variability in the shapes and sizes of the phosphenes LOCATION ANALYSIS AND COMPARISON: Location perceived (Figure 4). The features of the phosphenes as variability. Variability in the phosphene locations was first described by each subject are summarized in Table 2. evaluated in terms of changes in the mean 6 standard deviation of the distance between the set centroid point and the SIZE COMPARISON: The mean 6 standard deviation of individual phosphene centroids within the set (Table 4). the diagonal distance of the closest-fit rectangle for each The changes in the position of the phosphenes across set of phosphene drawings is shown in Table 3. Three different sets of experiments are shown as set centroid

VOL. 170 LONG-TERM PHOSPHENE CHARACTERISTICS IN ARGUS II SUBJECTS 105 TABLE 3. Comparison of Phosphene Size Variation Depicted by Chronically Implanted Argus II retinal prosthesis subjects

Diagonal Distance of Closest-ﬁt Rectangle for Phosphene, mm

Subject ID 001 003 005 006 007 009

Long interval 1 27.98 6 6.28 84.24 6 14.94 58.53 6 25.34 58.53 6 25.34 102.20 6 21.73 36.89 6 9.11 Long interval 2 27.42 6 5.90 93.73 6 11.18 46.09 6 11.03 46.09 6 11.02 115.17 6 15.85 28.62 6 4.94 Long interval 3 25.90 6 7.81 103.69 6 5.51 29.92 6 9.81 29.92 6 9.81 105.47 6 19.52 51.99 6 13.10 Long interval 4 25.90 6 7.81 102.88 6 4.72 32.46 6 5.02 32.46 6 5.02 111.43 6 18.80 29.72 6 5.26 Friedman test, P ¼ .18 .22 .005* .005* .71 .01* Short interval 1 21.16 6 4.51 82.73 6 7.80 26.88 6 4.43 26.88 6 4.43 93.97 6 10.49 30.67 6 3.09 Short interval 2 18.03 6 7.70 84.12 6 6.56 24.83 6 4.60 24.83 6 4.60 96.42 6 9.41 26.81 6 4.01 Short interval 3 25.67 6 6.42 100.28 6 11.99 20.86 6 7.42 20.86 6 7.42 90.93 6 13.29 30.87 6 4.22 Short interval 4 20.68 6 3.66 84.95 6 5.54 23.85 6 4.06 23.85 6 4.64 76.30 6 12.58 30.49 6 3.28 Friedman Test, P ¼ .48 .17 .74 .07 .01* .18

Table shows the mean 6 standard deviation (in millimeters) of the diagonal distance of the closest-ﬁt rectangle for each set of phosphene drawings. Friedman’s analysis of variance showed statistically signiﬁcant difference in the drawing size for long-interstimuli-interval stimulations in 3 subjects and in short-interstimuli-interval stimulations in 1 subject (marked with asterisk, *).

TABLE 4. Intrasubject Variability of the Phosphene Locations (Represented by Phosphene Centroids) Depicted by Each of the Chronically Implanted Argus II Retinal Prosthesis Subjects

Mean 6 SD Phosphene Centroid–Set Centroid Distance, mm

Subject ID 001 003 005 006 007 009

Long interval 1 30.7 6 11.5 13.0 6 3.43 19.1 6 18.8 8.3 6 5.9 6.4 6 2.7 8.9 6 3.9 Long interval 2 27.0 6 20.6 9.0 6 7.0 25.2 6 17.4 9.4 6 7.2 6.3 6 2.1 14.2 6 6.4 Long interval 3 27.5 6 13.7 9.4 6 5.6 20.5 6 9.9 19.8 6 9.24 6.0 6 2.3 15.2 6 6.3 Long interval 4 27.5 6 6.4 8.3 6 3.9 13.3 6 5.1 10.2 6 6.5 9.0 6 1.5 12.6 6 5.6 Short interval 1 22.7 6 14.6 10.2 6 7.3 12.3 6 8.2 14.0 6 5.8 8.1 6 5.4 11.3 6 5.0 Short interval 2 28.4 6 17.4 8.6 6 2.7 16.8 6 5.9 12.7 6 8.4 11.2 6 6.8 10.5 6 6.5 Short interval 3 36.1 6 21.5 12.4 6 6.2 18.1 6 9.6 13.0 6 9.4 7.9 6 4.4 11.5 6 6.3 Short interval 4 39.8 6 15.5 10.5 6 8.0 14.6 6 7.3 14.8 6 8.1 16.8 6 7.5 11.6 6 7.5

The mean (6 standard deviation) distances of the individual phosphene centroids to the set centroid for all sets of experiments are shown. distribution relative to each subject’s visual field center larger numbers of pixels in the future,11 high-resolution (Figure 5 and Table 5). pixelated vision from electrical retinal stimulations may become an important avenue for vision restoration. Despite Retinotopic agreement. The quad-fovea location and the the encouraging outcomes showing improved visual perfor- expected phosphene location for each subject are as mance with the use of the device,1,10,12,13 understanding of shown in Table 6. Out of the 6 subjects, 4 subjects’ the interaction between the stimulating parameters and depicted phosphene locations fell in the same quadrant of the individual subjects’ visual percepts remains poor. the visual field as that of the expected phosphene Clearly, the ability to produce consistent, controllable, location. Two subjects (001 and 005) depicted and retinotopically defined phosphenes will remain phosphenes in the same hemifield (temporal and inferior, central to improving prosthetic vision in the future. respectively) as the expected locations. We chose the parafoveal quad as our focus for evaluation as physiologically, the photoreceptor-to–bipolar cells–to ganglion cells ratio in this region approaches 1:1:1. Direct electrical stimulation at this area would therefore theoret- DISCUSSION ically be most predictable, giving rise to focal, dot-like phosphenes, whether the bipolar cells or ganglion cells WITH THE ESTABLISHMENT OF GOOD LONG-TERM SAFETY were the main target of electrical stimulation. Greater den- profiles9,10 and advances in biotechnology promising sity of ganglion cells found in this region also allowed for

106 AMERICAN JOURNAL OF OPHTHALMOLOGY OCTOBER 2016 TABLE 5. Phosphene Locations (Represented by Set Centroid) Relative to Subjective Visual Field Center in Chronically Implanted Argus II Retinal Prosthesis Subjects

Subject ID

Location 001 003 005 006 007 009

Set centroid: long interval 1 36.6 mm 20.0 mm 25.9 mm 14.5 mm 9.3 mm 38.7 mm S(T) IT IT SN IN IT Set centroid: long interval 2 17.7 mm 9.2 mm 42.9 mm 22.3 mm 7.0 mm 28.2 mm (I)T ST IT SN IN IT Set centroid: long interval 3 25.7 mm 0.00 mm 35.5 mm 19.3 mm 9.6 mm 68.0 mm ST O IT ST IN (I)T Set centroid: long interval 4 23.9 mm 6.9 mm 24.4 mm 24.6 mm 6.1 mm 36.6 mm (S)T IT (I)T SN IN IT Set centroid: short interval 1 25.9 mm 15.1 mm 37.2 mm 33.0 mm 5.8 mm 46.6 mm (S)T IT IT ST I (I)T Set centroid: short interval 2 22.9 mm 11.9 mm 46.0 mm 55.3 mm 11.2 mm 46.9 mm ST IT IT ST IT IT Set centroid: short interval 3 41.2 mm 24.8 mm 30.3 mm 56.2 mm 2.0 mm 48.9 mm S(N) IT IT IT ST (I)T Set centroid: short interval 4 51.4 mm 10.6 mm 24.8 mm 55.5 mm 36.2 mm 54.5 mm ST IT ST IT T T Final centroid 24.8 mm 10.9 mm 31.5 mm 21.4 mm 6.1 mm 45.4 mm ST IT IT ST IT IT

IT ¼ inferotemporal to; O ¼ set centroid overlaps the visual field center; SN ¼ superonasal to; ST ¼ superotemporal to. Table shows the distance and relative location of the set centroid to the subject’s visual field center. When the angle of deviation of the set centroid point is less than 10 degrees from the horizontal or vertical line, the direction of deviation is bracketed. The final centroid was calculated to represent the summary position of all the set centroids for each subject.

TABLE 6. Quad Position Relative to the Estimated Fovea Location From Fundus Photograph, for Each Chronically Implanted Argus II Retinal Prosthesis Subject

Subject ID

Measurement 001 003 005 006 007 009

Quad C07C08 A07A08 E05E06 A07A08 E07E08 E07E08 D07D08 B07B08 F05F06 B07B08 F07F08 F07F08 Quad-fovea distance 1100 mm 1000 mm 1200 mm 1300 mm 1100 mm 2400 mm Quad position relative to fovea SN SN ST IN SN SN Expected phosphene location in VF 19.2 mm IT 17.5mm IT 21.0 mm IN 22.7 mm ST 19.2 mm IT 42.16 mm IT Final centroid location in VF 24.8 mm ST 10.9 mm IT 31.5 mm IT 21.4 mm ST 6.1 mm IT 45.4 mm IT

IN ¼ inferonasal; IT ¼ inferotemporal; SN ¼ superonasal; ST ¼ superotemporal; VF ¼ visual field. The expected phosphene location and the final centroid location relative to each subject’s visual field center are also shown.

lower stimulating currents to be used to elicit distinctive stimuli intervals (with temporal resolution down to 1 phosphenes reliably (ie, lower threshold). Quad stimula- second) and was reproducible on separate occasions ranging tion, rather than single electrode, was chosen because from 1 week to 1 month apart. However, in spite of the good only 2 out of the 6 subjects had functioning single elec- intrasubject consistency, each of the 6 subjects experienced trodes in the parafoveal region at the time of the study. phosphenes of totally different shapes and sizes. It is likely Interestingly, our study has shown that irrespective of the that the intrasubject variations reﬂect the variety of genetic variability in size and location of the drawn phosphenes, all diseases, the duration the degeneration has been present, our subjects reported perceiving phosphenes of the same the different proportions of surviving bipolar and ganglion shapes and similar sizes when using the same stimulating pa- cell types, and the variability in reconnections and rameters. This consistency was seen across different inter- remodeling within these severely diseased retinas.

VOL. 170 LONG-TERM PHOSPHENE CHARACTERISTICS IN ARGUS II SUBJECTS 107 The wide variation in phosphene shape and size may in their heads and eyes pointing straight ahead during the part explain the wide range of intersubject performance task. The largest deviation of the phosphene centroid loca- levels observed when subjects performed tasks involving tion from the set centroid location was 39.8 6 15.5 mm, visual form differentiation, for example letter, shape indicating a disparity distance of about 5 cm, viewed at (daCruz, L., et al. IOVS 2012;53:ARVO E-Abstract 30 cm away (z9.5 degrees). One subject (007) showed 5507), and object recognition.1 Indeed, 1 subject (009), greater localization consistency than the others who consistently outperformed other subjects in all the (Figure 4), and this was reflected in his superior perfor- vision form recognition tasks, depicted phosphene shapes mance in carrying out object prehension tasks.15 This is closest to the stimulation pattern (ie, 2 horizontal lines in keeping with Sabbah and associates’ conclusion that with filled-in center, simulating a box shape from the the Argus II subjects are able to develop strategies to mini- quad stimulation). It is also interesting to note that despite mize the impact of head-eye misalignment. Our study also the intersubject variability in perceived phosphenes, Argus showed that 4 out of the 6 subjects had phosphene loca- II subjects still performed statistically significantly better in tions in the expected visual field quadrant, as indicated visual form differentiation tasks with the device than by the relative quad-fovea location. These subjects may without. This may be attributable to the intrasubject con- be better at keeping their head-eye alignment, such that sistency and reproducibility of the phosphenes elicited. their subjective visual field center was in alignment with Such consistency and reproducibility may allow each sub- the estimated fovea location. ject to learn and adapt to the prosthetic visual information In general, our study showed that stimulation of the same (albeit crude and unlike natural vision) and use it to make quad with the same stimulating parameters gave rise to consistent simple decisions about form. This learning is consistently reproducible phosphenes for a given subject evident in the improvement of psychophysical and other in a cohort of chronically implanted subjects more than 5 visual task performance with training.11,13 years after the initial surgery. This consistency is an encour- Sabbah and associates reported that the perceived posi- aging basis for the construction of more complicated pixe- tion of the phosphene is dependent on the direction of lated images. Given the vastly different shapes and sizes of gaze (ie, the eye position) of the Argus II subject, and the phosphenes perceived by individual subjects, future that misalignment of head and eye position occurred in work into determining the suitable stimulating parameters all the subjects in their study.14 This would explain the var- for each electrode/quad stimulation may be required for iations in the location of the phosphene drawings in our each subject, to achieve the construction of useful pixe- study, even though the subjects were instructed to keep lated prosthesis-based vision.

FUNDING/SUPPORT: Y.H. LUO, J.J. ZHONG, M. CLEMO, AND L. DA CRUZ RECEIVED FINANCIAL SUPPORT FROM THE DEPARTMENT of Health, UK through the award made by the National Institute for Health Research (NIHR) to Moorﬁelds Eye Hospital National Health Service (NHS) Foundation Trust, and University College London (UCL) Institute of Ophthalmology, for a Specialist Biomedical Research Centre for Ophthalmology. Financial disclosures: Lyndon da Cruz attended a single paid advisory board meeting for Second Sight Medical Products. The following authors have no ﬁnancial disclosures: Yvonne H-L. Luo, Joe Jiangjian Zhong, and Monica Clemo. All authors attest that they meet the current ICMJE criteria for author- ship.

REFERENCES 6. Rohrschneider K. Determination of the location of the fovea on the fundus. Invest Ophthalmol Vis Sci 2004;45(9): 1. Luo YH-L, daCruz L. The ArgusÒ II Retinal Prosthesis Sys- 3257. tem. Prog Retin Eye Res 2016;50:89–107. 7. ArgusÒ II Retinal Prosthesis System. Available at: http://www. 2. Humayun MS, de Juan E, Dagnelie GG, Greenberg RJ, accessdata.fda.gov/cdrh_docs/pdf11/h110002c.pdf. Accessed Propst RH, Phillips DH. Visual perception elicited by electri- February 12, 2016. cal stimulation of retina in blind humans. Arch Ophthalmol 8. Light and shadow centroids j GIMP Plugin Registry. Available 1996;114(1):40–46. at: http://registry.gimp.org/node/27529. Accessed February 14, 3. Humayun MS, de Juan E, Weiland JD, et al. Pattern electrical 2016. stimulation of the human retina. Vision Res 1999;39(15): 9. Ho AC, Humayun MS, Dorn JD, et al. Long-term results from 2569–2576. an epiretinal prosthesis to restore sight to the blind. Ophthal- 4. Rizzo JF, Wyatt J, Loewenstein J, Kelly S, Shire D. Perceptual mology 2015;122(8):1547–1554. efﬁcacy of electrical stimulation of human retina with a 10. Humayun MS, Dorn JD, daCruz L, et al. Interim results from microelectrode array during short-term surgical trials. Invest the International Trial of Second Sight’s visual prosthesis. Ophthalmol Vis Sci 2003;44(12):5362–5369. Ophthalmology 2012;119(4):779–788. 5. Humayun MS, Weiland JDJ, Fujii GYG, et al. Visual percep- 11. Stronks HC, Dagnelie GG. The functional performance of tion in a blind subject with a chronic microelectronic retinal the Argus II retinal prosthesis. Expert Rev Med Dev 2014; prosthesis. Vision Res 2003;43(24):2573–2581. 11(1):23–30.

108 AMERICAN JOURNAL OF OPHTHALMOLOGY OCTOBER 2016 12. Garcia S, Petrini K, Rubin GS, daCruz L, Nardini M. Visual spatial localization in blind subjects wearing an Argus II and non-visual navigation in blind patients with a retinal retinal prosthesis. Invest Ophthalmol Vis Sci 2014;55(12): prosthesis. PLoS One 2015;10:e0134369. 8259–8266. 13. Chader GJ, Weiland J, Humayun MS. Artiﬁcial vision: needs, 15. Luo YH-L, Zhong JJ, daCruz L. The use of ArgusÒ functioning, and testing of a retinal electronic prosthesis. Prog II retinal prosthesis by blind subjects to achieve localisa- Brain Res 2009;175(175):317–332. tion and prehension of objects in 3-dimensional space. 14. Sabbah N, Authie CN, Sanda N, Mohand-Said S, Graefes Arch Clin Exp Ophthalmol 2015;253(11): Sahel J-A, Safran AB. Importance of eye position on 1907–1914.

VOL. 170 LONG-TERM PHOSPHENE CHARACTERISTICS IN ARGUS II SUBJECTS 109