IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 18, NO.
1, FEBRUARY 2010 1
Electric Field Stimulation of Bipolar Cells in a
Degenerated Retina—A Theoretical Study
Matthias Gerhardt, John Alderman, and Alfred Stett
Abstract—Subretinal implants are the subject of clinical inves-
tigation for their ability to evoke useful visual sensations in blind
individuals via electrical stimulation of the diseased retina. We in-
vestigated the spatial characteristic of the retinal polarization ob-
tained by electric field stimulation through a subretinally located
monopolar electrode array and bipolar electrode array. By com-
bining electric potential simulation through a boundary element
method with a segmented cell model, we computed the membrane
m
voltage at the axon terminal of the bipolar cells as a function of the
axon length (50–110 ) and the electrode diameter. We found
m
that short OFF bipolar cells are predominantly addressed by small
150 s
bipolar electrodes (diameter between 60 and 100 ) and by using
a short duration ( ) of the stimulating voltage pulse. Long
100 m 150 s
ON cells are best addressed by large monopolar electrodes (diam-
eter ) and a long pulse duration ( ). How-
ever, the low selectivity of the electric field stimulation with regard
to the cell length does not enable the individual depolarization of
long OFF cells and short ON cells. When the stimulation must take
place at multiple retinal sites simultaneously, the bipolar electrode
arrays allow for higher spatial modulation of the polarization of
the axon terminal than the monopolar arrays.
Index Terms—Bipolar cells, microelectrode array, retina im-
plant, simulation, subretinal, visual prosthesis.
I. INTRODUCTION
EGENERATIVE diseases of the retina, such as retinitis
D pigmentosa or age-related macular degeneration, are as-
sociated with atrophy of the photoreceptors (PRs), which in turn
is among the major factors that lead to blindness. The punc-
tiform electrical stimulation of the distal or proximal side of
the retina of normal-sighted and blind individuals is known to
cause the perception of so-called phosphenes [11], [30], [32]—a
dot-like visual sensation without any visual stimulation [3], [4].
Thus, a series of phosphenes that could code for a visual image
has become the basis of therapeutic approaches to help blind pa-
tients. Multilocal electrical stimulation of the retina is applied
Manuscript received January 14, 2009; revised July 15, 2009; accepted Oc-
tober 19, 2009. First published January 12, 2010; current version published Feb-
ruary 24, 2010. .This work was supported in part by the Federal Ministry of Ed-
ucation and Research (BMBF), Germany, under Grant 0315113 to Retina Im-
plant AG, Reutlingen, Germany and in part by the European Commission (Marie
Curie Transfer of Knowledge Program, Grant MTKD-CT-2005-029568)
M. Gerhardt was with NMI Natural and Medical Sciences Institute at the
University of Tuebingen, D-16348 Wandlitz, Germany. He is now with the
Tyndall National Institute, Lee Maltings, Cork, Ireland (e-mail: matthias.ger-
hardt@alumni.tu-berlin.de).
J. Alderman is with the Tyndall National Institute, Cork, Ireland (e-mail: john.
alderman@tyndall.ie).
A. Stett is with NMI Natural and Medical Sciences Institute at the University
of Tuebingen, Reutlingen, Germany (e-mail: stett@nmi.de).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TNSRE.2009.2037323
1534-4320/$26.00 © 2010 IEEE
by an implanted electrode array to evoke a multiphosphene per-
ception based on the pattern of the stimulating electrodes [30].
Two approaches are distinguished according to the position
of the electrode array. In the epiretinal approach, the electrode
array is in close contact with the ganglion cell layer to stimulate
the ganglion cells. In the subretinal approach, the electrode array
is located in the layer of degenerated PR cells, and bipolar and
horizontal cells are the targets of electrical stimulation.
Examination of blind patients revealed that both approaches
work with simple patterns of stimulating electrodes [19], [30],
[49]. However, the reliable perception of complex patterns with
sufficient spatial resolution useful for daily life has not yet been
established. The simultaneous stimulation of ganglion cells and
their axons, and the question of how to distinguish between them
may be a problem, as reported by Rizzo et al. [19], [20] for the
epiretinal approach. Furthermore, the signals of ganglion cells
are the outcome of the very complex retinal network activity,
which is transmitted by the ganglion cell axons to the brain [30].
Accordingly, Eckmiller et al. [8] devised a “retinal encoder” to
stimulate ganglion cells with a pattern similar to the ganglion
cell output generated by a native retina. Clinical results on the
success of this retinal encoder are not available. A challenge
may be the penetration of the electric field into the inner retina
and the polarization of bipolar cells via epiretinal stimulation.
Ahuja et al. [35] suggests different pulse frequency’s for selec-
tively stimulating ganglion cells over bipolar cells.
The subretinal approach is based on the assumption that the
inner retina of blind patients is still capable of processing infor-
mation, and that it can be electrically stimulated in a way that
the modulation of the ganglion cell activity yields meaningful
visual perception. The assumption that degenerated retinas still
retain processing capabilities is supported by evidence: a high
percentage of the neurons in the inner retina remained histo-
logically intact in patients with retinitis pigmentosa, even after
many years of blindness [11], [39], [40].
This study addresses the question of how to improve the spa-
tial resolution and cell selectivity of the electric field stimulation
with a subretinally implanted electrode array. To analyze the re-
quirements for subretinal stimulation, a brief description of the
basic radial signal pathway from the PR cells to the ganglion
cells in an intact retina is presented (Fig. 1).
In the PR layer, PR cells sample the image of visual scenes
projected by the eye lens on the back of the eye. Each PR “mea-
sures” the brightness of the corresponding point in the image
and then transforms this information into a neural signal via the
release of the neurotransmitter glutamate. This release of gluta-
mate takes place at the presynaptic terminals (PST) of the PR
cells located in the outer plexiform layer (OPL) of the retina.
2 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 18, NO. 1, FEBRUARY 2010
Fig. 1. Schematic representation of the radial components of the cone pathway
in an intact retina. GC: Ganglion cell layer. IPL: Inner plexiform layer. BC:
Bipolar cell layer. OPL: Outer plexiform layer. PR: Photoreceptor layer. The
arrow indicates the presynaptic terminals of the axons of the bipolar cells. Open
circle in the OPL indicates inverting synapses in the ON pathway.
The concentration of the released glutamate is inversely propor-
tional to the local brightness of the retinal image [28]. While glu-
tamate release is high in darkness, it is reduced by light. Each
PR is synaptically connected to a number of bipolar cells and
horizontal cells. Horizontal cells bundle the neural signals from
the PR in darkness but separate them as the level of brightness
increases [12]. Depending on the type of the glutamate receptors
located in the dendritic terminals of bipolar cells, PR signaling
either leads to depolarization (ON bipolar cells) or hyperpolar-
ization (OFF bipolar cells), when the light intensity is increased
[1], [6]. This change of membrane voltage propagates electro-
tonically along the soma and cable-like axon of the bipolar cells
from the OPL to the inner plexiform layer (IPL). If an incoming
depolarization reaches the threshold to trigger the opening of the
voltage-controlled channels in the axon terminal, the re-
sulting intracellular increase in initiates the release of glu-
tamate from the bipolar cell PSTs [13], [15], [21]. This causes
synaptic signaling from the bipolar cells to the amacrine and
ganglion cells. Interestingly, the axons of the ON bipolar cells
are longer than the axons of the OFF bipolar cells (Fig. 1). The
ON bipolar cell axons terminate in the inner half of the IPL,
whereas the OFF bipolar cell axons terminate in the outer half
of the IPL [28]. Ganglion cell spiking is modulated because of
the signal propagation in the retinal circuitry. Each ganglion cell
receives input from the PR layer in a certain region according to
its receptive field with center and antagonistic surround [7]. The
receptive fields are overlapping structures [22], and each PR is
simultaneously connected to several ON and OFF bipolar cells.
The voltage-gated channels located at the axon termi-
nals of the bipolar cells are the ideal target for the subretinal
electrical stimulation of the retina. The primary task is to mod-
ulate the membrane voltage at the PST to control the transmitter
release, and ultimately, the signalling to the amacrine and gan-
glion cells. Therefore, electrode configurations and stimulation
protocols are required, which enable the efficient stimulation of
the bipolar cell axons. If the subretinal electrical stimulation is
capable for controlled depolarization of the bipolar cell PSTs in
the IPL, then it could have the same effect as native excitation
by light.
The most efficient way to extracellularly stimulate the ex-
tended, cable-like bipolar cells is by electric fields directed par-
allel to the axon [53]. Thus, electrode configurations that pro-
duce transretinal currents have to be identified. A recent theo-
retical simulation on the interactions between bipolar cells and
electrical fields has shown that a positive electric current flowing
from the outer to the inner retina can depolarize the axon termi-
nals. An opposite directed current is observed to hyperpolarize it
[26]. Furthermore, in vitro experiments have verified the stimu-
lating effect of retinal inward-directed currents [26]. According
to [13], [15], [21], the phosphene perceptions evoked by sub-
retinal electrical stimulation may be attributed to the depolar-
ization of the PST. Thus, an ideal subretinal stimulation needs
to satisfy at least three major aspects: 1) the subretinal electrical
stimulation must control the depolarization of the PST of the
bipolar cells to modulate transmitter signaling, 2) a high spa-
tial resolution must be reached by the electrical stimulation to
imprint more complex stimulation patterns into the polarization
patterns at the PSTs in the INL, and 3) the ON and OFF chan-
nels of the visual pathway should be stimulated separately to
allow adequate coding of contrast information. To fulfill these
requirements, appropriate electrode configurations and stimu-
lating waveforms must be chosen. In current retinal implants,
the so-called monopolar stimulation is applied [2]. The term
“monopolar” indicates that the stimulation electrodes contact
the retina and that the electric potential of all the stimulating
electrodes is defined with respect to the distantly located return
electrode.
The disadvantage of this kind of electrical stimulation is the
superposition of electric fields between the simultaneously stim-
ulated neighboring electrodes [14] and the spread of the stimu-
lation current to distant tissue areas. These effects could limit
the spatial resolution obtainable by the subretinal stimulation.
In the bipolar electrode configuration, the return electrode
surrounding the stimulating electrodes [19] may circumvent
this limitation, because the field superposition effect between
the neighboring electrodes is minimized and the stimulation
focality is improved [36]. Palanker et al. [14] used similar
structures for implants, where the retina penetrated into the
electrode array. Loudin et al. [16] described in detail the issue
of the interference of multiple electrodes in monopolar arrays
and its improvement in bipolar arrays.
In this study, we compared the effects of the monopolar and
bipolar electrode array on the bipolar cell layer by simulating
the polarization of simplified bipolar cells in the external elec-
tric field. The results demonstrated the advantages and disad-
vantages of the monopolar and bipolar electrode configurations
with respect to the spatial contrast of the polarization pattern in
the IPL. It also indicated how to selectively stimulate bipolar
cells that vary in length. Some of these results have been re-
ported previously [26], [41].
II. METHODS
The effect of extracellular stimulation with an electrode array
on the axon terminal of bipolar cells was modeled in a two-step
procedure [44]. First, the electric potential produced by an
electrode array in the extracellular space was simulated by a
boundary element method. Second, the extracellular potential
served as an input of a segmented cable model [45] of the bipolar
cells (Fig. 2).
A. Electric Potential Simulation
For simplification, we replaced the retina with its various
retinal layers [10] by a homogeneous, isotropic, and purely re-
sistive electrolyte. The stationary electric potential inside the
GERHARDT et al.: ELECTRIC FIELD STIMULATION OF BIPOLAR CELLS IN A DEGENERATED RETINA—A THEORETICAL STUDY 3

Fig. 2. Model for extracellular stimulation of a bipolar cell. (a) The cable-like bipolar cell of length L is exposed to an electric potential 8 resulting from the
stimulating voltage, V , applied between a stimulation electrode of an electrode array and the grounded return electrode. The difference between the intracellular
potential, 8 (x; t), and the extracellular potential, 8 (x; t), causes an excursion of the membrane voltage, V (x; t), from the resting value. A: soma-dendritic
terminal, B: axon terminal. (b) Cable model with n segments, representing the passive membrane properties of the axon with R membrane resistance, C
membrane capacitance, and R intracellular resistance.

Fig. 4. Dimensions (in micrometers) of the units of the monopolar electrode

arrays (lower left) and the bipolar electrode arrays (lower right). The shaded
areas represent the electrodes.

Fig. 3. Schematic representation of monopolar and bipolar electrode concept

for stimulation. In the case of the monopolar electrodes, the distance of the re-
turn electrode from the electrode array does not influence the current spread pro-
duced by the stimulating electrodes (marked with +) and the stimulus spreads
throughout the tissue. For the bipolar electrodes, the return electrode is placed
in-plane with the stimulating electrodes, reducing the current spread.
electrolyte above a monopolar and a bipolar electrode array
(Fig. 3) was simulated using the boundary elements method
(Lorentz, Integrated Soft, Canada). The model array consisted
of 3 5 subunits with an area of 50 50 (Fig. 4). For the
monopolar electrode array simulation, the distance between the
return electrode and the stimulation electrodes was increased
until the shape of the electric potential function appeared to be
independent from the distance. For the bipolar electrodes, the
return electrode was in plane with the stimulation electrodes.
The potential of the return electrode was defined by , of membrane segments and were replaced by a linear and
and the stimulation electrodes were either at (Dirichlet
boundary condition) or floating (Neumann condition). The re-
sistivity of the substrate in the plane between the electrodes was
(glass).
The resistivity of all the electrodes was defined by
(platinum), and the resistivity of the
electrolyte above the electrodes was defined by .
To compare the electrode configurations, the same pattern of
the potential-free electrodes and electrodes at was
used (Fig. 4).
B. Bipolar Cell Model
Bipolar cells were modeled as hollow membrane cables of
the variable length, , and fixed diameter, (1.4 , [37]), sur-
rounded by the extracellular electrolyte. The cables consisted
time-invariant resistor and capacitor (RC) network, as shown
4 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 18, NO. 1, FEBRUARY 2010

in Fig. 2. Active components, i.e., voltage-sensitive ion chan- Segment :

nels, were neglected. Each segment contained the membrane
impedance, , with the complex frequency,
(11)

(1a) Using a matrix equation, the linear equation system (9)–(11)

(1b) for the membrane currents is solved:

with membrane capacitance, , membrane resistance, ,

and intracellular resistance, (12)

(2) where .
(3) is the load vector and

(4)

All the elements were defined by the segment length, , and

the specific membrane resistance, (24 ), membrane .. .. .. .. ..
. . . . .
capacity, (1.1 ), and the specific resistance of the
intracellular electrolyte, (130 ) [42].
The cell model was coupled to the extracellular potential (13)
at node , which was obtained by the potential simulation de- the stiffness matrix with
scribed in Section II-A. According to the range of the length of
the bipolar cells described in the literature [27], [37], [42], the
(14)
values 50, 70, 90, and 110 for cell length were chosen.

The inverse stiffness matrix was numerically calculated

C. Calculation of the Membrane Voltage (OriginC 7.5). With (5) and (8), the membrane potential at the
The membrane voltage axon terminal is given by

(5) (15)

at the axon terminal of the bipolar cell was calculated in the

Laplace domain as a function of the differences of the extracel- where are the elements of the inverse matrix . The
lular potential at two neighboring nodes response of the axon terminal to a stimulating voltage step

(6)

with (16)
( inverse Laplace transform) is given by

(7)
(17)

(8)
The step response of the membrane voltage in the time-do-
the following relations are obtained: main, ), was obtained by computing the inverse Laplace
Segment 1: transform of ) using the residuum theorem [29]

(9)
(18)
Segment :
The Jordan curve , which defines the way of integration,
includes all the poles of . The numerical integration was
performed using the trapeze method [17] containing 1400 sup-
(10) porting points. Following a convergence analysis, a cable con-
sisting of 100 segments was chosen for the simulations.
GERHARDT et al.: ELECTRIC FIELD STIMULATION OF BIPOLAR CELLS IN A DEGENERATED RETINA—A THEORETICAL STUDY
2
Fig. 5. Potential 8 in the bulk electrolyte (as seen above) of a 3 5 electrode
2
array with 50 50  m unit size. The electrodes marked by a star (3) are at
floating potentials, and the others at 1 V. (a), (b): monopolar array; (c), (d):
bipolar array. The side-views [(a), (c)] show the potential profile in the bulk
along the center line shown in the top-views (b, d). The top-views show the
m
potential profile in a plane 25  above the electrode array [horizontal line in
(a) and (c)].
III. RESULTS
A. Potential Pattern Produced by an Electrode Array
The most efficient way to stimulate extracellular axons is
through a potential gradient parallel to the axon [53]. Therefore,
in the case of subretinal stimulation with monopolar or bipolar
electrode arrays, a potential gradient perpendicular to the
surface of the electrode array is the appropriate stimulus for the
bipolar cells in the adjacent retina.
We first simulated the static potential function in the isotropic
electrolyte produced by a 3 5 electrode array. The 2-D spatial
voltage pattern simultaneously applied to a set of electrodes was
transferred into a 3-D spatial potential pattern in the bulk elec-
trolyte above the array, as shown in Fig. 5.
There were significant differences between the results ob-
tained for a monopolar and a bipolar array. The main feature of
5
Fig. 6. Localization of the bipolar cell model with respect to the electrode array
(the star indicates the axon terminal).
the potential pattern produced by the monopolar array was ob-
served at the surface of electrodes defined as “floating,” which
was determined by the superposition of the field generated by
the stimulating electrodes [Fig. 5(a)]. Consequently, there was
a remarkable potential gradient from the “floating” electrodes
toward the bulk electrolyte. The potential pattern at any plane
parallel to the electrodes was blurred compared with the poten-
tial pattern applied to the electrodes [Fig. 5(b)].
Using a bipolar array, we observed that the field superposition
do not influence the projection of the boundary condition in the
bulk above the electrode array [Fig. 5(c)].
The direction of the electrical field vector at the surface of
the stimulating electrode is opposite that of the direction at the
grounded return electrode in plane with the stimulated elec-
trodes. As a result, the potential in the bulk electrolyte distant
to the surface is nonzero. The potential at the surface of elec-
trodes defined as “floating” is determined by the potential in the
vicinity of the ground electrode. Thus, there is a potential drop
from the surface of the stimulating electrodes toward the bulk,
and from the bulk toward the “floating” electrodes.
B. Polarization of the Bipolar Cells of Different Lengths
We then simulated the voltage-step response of the axon ter-
minal of a bipolar cell located at variable positions along the
centerline of the array, as shown in Fig. 6. Owing to the presence
of a debris layer at the distal retina side after PR degeneration
[9], we included a 5- gap between the surface of the electrode
array and the soma-dendritic terminal of the cell. As explained
in the Methods section, the potential drop perpendicular to the
surface of the array served as an input to bipolar cells of vari-
able length. Fig. 7 shows the voltage step-response of the mem-
brane at the bipolar cell’s axon terminal above a stimulating
electrode. An increase in the membrane voltage, which corre-
sponds to the depolarization of the axon terminal, was found in
the cells on top of the stimulating electrodes. In both monopolar
stimulation and bipolar stimulation, the time to reach maximum
membrane depolarization is shorter when the axon is shorter.
The main difference between the monopolar and bipolar stim-
ulation is the steady-state membrane voltage at the axon ter-
minal of the short cells and long cells. Subsequently, the max-
imum membrane voltages found in the monopolar stimulation
were observed to be higher than in the bipolar stimulation. With
monopolar electrode configuration, stronger depolarization was
observed in the long bipolar cells [Fig. 7(a)]. With the bipolar
6 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 18, NO. 1, FEBRUARY 2010
Fig. 7. Voltage step-response of the axon terminal of the bipolar cells of dif-
ferent cell lengths for (a) monopolar and (b) bipolar stimulation. The cell was
positioned above the center electrode of the array.
electrode configuration, however, the long bipolar cells became
less depolarized than the short cells [Fig. 7(b)].
Fig. 8 shows the dependence of the polarization of the axon
terminal on the localization of the bipolar cell above the array.
As expected, the cells above the “floating” electrodes were also
depolarized due to the superposition effects above a monopolar
electrode array. The difference in the steady-state membrane
voltage of the cells above the stimulating and “floating” elec-
trodes was found to be dependent on the length of the axon. For
the shorter cells, the spatial modulation of the membrane voltage
was found to be higher than that for the longer cells [Fig. 9(a)].
Owing to the field superposition effect, as shown in Fig. 5(a)
and (b), the cells above the center of the array are more strongly
depolarized than those located above the outer electrodes.
However, with the bipolar electrode array, a negative shift in
the membrane voltage was observed in the cells located above
the “floating” electrodes [Fig. 9(c)]. In physiological terms, this
is referred to as hyperpolarization. The amplitudes of the depo-
larization and the hyperpolarization were found to be higher in
short cells than in long cells.
The localization of the transition from the depolarization
to the hyperpolarization was independent of the stimulation
strength [Fig. 9(d)]. Furthermore, the membrane polarization
above a stimulating electrode was not dependent on the local-
ization of the electrode within the array.
The transmembrane voltage at the axon-terminal will also be
strongly affected by the variation of the height of the gap be-
tween the electrode array and the soma-dendritic terminal of the
cell. This variable was neglected in this study.
C. Influence of Electrode Size
Finally, we investigated the dependence of the depolariza-
tion of the axon terminal on the electrode size. We simulated
the polarization of a bipolar cell located above the center of
a monopolar and a bipolar electrode without the surrounding
array. In case of the bipolar array, the square pixel with the re-
turn electrode was scaled proportionally.
The polarization of the bipolar cell’s axon terminal increased
up to maximum, followed by a decrease, with an increase in
the electrode diameter in both the cases (Fig. 10). The longer
the cell, the higher the maximum polarization observed. In the
monopolar case, the maximum polarization was obtained with
an electrode diameter that corresponded to the length of the cell.
Fig. 10 shows the polarization of a cell with a length of 50 ,
which was approximately 60% of the polarization of a cell with
a length of 110 at an electrode diameter of 160 . For a
given cell length, the size of the monopolar electrodes necessary
to reach the maximum membrane polarization was found to be
smaller than that of the bipolar electrodes. The difference of the
optimum electrode diameters of two different cell lengths was
greater with bipolar electrodes than with monopolar electrodes.
IV. DISCUSSION
Electrical stimulation of the retinal cells by means of an elec-
trode array implanted on or into the retina is currently being
investigated as an option to provide artificial vision in patients
with photoreceptor degeneration [30]. With subretinal implants,
an electrode array is positioned adjacent to the bipolar cells
that are known to survive after years of blindness [47], [48].
In a recent clinical pilot study, it has been shown that blind re-
tinitis pigmentosa patients are able to discriminate individual
letters by means of an electrode array with 1500 50 50
monopolar electrodes [49]. However, the successful discrimi-
nation required large letters with a retinal image height of ap-
proximately 2 mm. In an earlier ex vivo study, it was shown
that the monopolar stimulation with two needle-type electrodes
separated by more than 100 resulted in two separate retinal
patches with excited ganglion cells [46]. In this study, we simu-
lated the membrane polarization at the PST of the bipolar cells
by subretinal stimulation with monopolar and bipolar electrode
arrays. Using a simplified cell model, we focused on the depen-
dence of the membrane polarization on the length of the bipolar
cells, on the localization of the cells above the monopolar and
bipolar electrodes within a planar array, and on the size of the
electrodes.
A. Spatial Discrimination
The membrane potential at the axon terminal of the bipolar
cells, and thus the synaptic transmission to the successive cells,
can be controlled with both electrode configurations. A weaker
membrane depolarization is achieved with the bipolar electrodes
than with the monopolar electrodes, with the same stimulation
strength and the same electrode size (Fig. 7). This is the result
GERHARDT et al.: ELECTRIC FIELD STIMULATION OF BIPOLAR CELLS IN A DEGENERATED RETINA—A THEORETICAL STUDY
Fig. 8. Step response of the axon terminal as a function of the location
of the fast decay of the potential above the bipolar electrodes.
With regard to the spatial resolution, the bipolar electrode ar-
rays are observed to be superior to the monopolar electrode ar-
rays. As described earlier, a punctiform monopolar stimulation
of the distal retina site leads to ganglion cell activity within a
small retinal patch whose diameter increases with increasing
stimulation strength. By using an array of monopolar electrodes,
the superposition of the electric field leads to the depolarized
bipolar cells above the stimulating and non stimulating elec-
trodes (Figs. 8 and 9). Consequently, the modulation depth of
the voltage pattern in the bipolar cell axon terminal (PST) be-
comes much lower than that of the original stimulation voltage
pattern applied to the electrodes. When depolarization reaches
the threshold for gating the channels located at the axon
terminal, a blurred pattern of ganglion cell activity is evoked.
Thus, spatial discrimination is limited when the monopolar elec-
trodes are stimulated simultaneously.
With respect to the bipolar electrodes, the depolarized retinal
patch is defined by the electrode geometry and not by the super-
position effects and stimulation strength. Each stimulating elec-
trode is surrounded by bipolar cells that are polarized opposite
to the cells above the stimulating electrodes. Thus, a unipolar
voltage pattern applied to a bipolar electrode array is transferred
onto a dipolar pattern of membrane voltages. It has been shown
that a retinal inward-directed positive current (flowing from the
photoreceptor side to the ganglion cell side), which causes depo-
larization of the terminals, has a significantly lower stimulation
threshold than an outward-directed current, which leads to hy-
perpolarization [26], [46]. Thus, it is reasonable to conclude that
with an array of simultaneously stimulating bipolar electrodes,
7
X and length L of the bipolar cell.
complex spatial stimulation patterns can be transferred onto the
polarization patterns at PSTs with higher spatial discrimination
than with a monopolar array. Lovell et al. [50] have drawn a
similar conclusion for epiretinal ganglion cell stimulation.
We found a range of optimum electrode diameters (from 50 to
110 for monopolar and from 70 to 180 for bipolar elec-
trodes) to produce maximum depolarization of the bipolar cells
(Fig. 10). Operating outside this range requires higher ampli-
tudes of the stimulating voltage which is, however, limited for
electrochemical reasons. Owing to the limited charge transfer
capabilities of microelectrodes [51], the minimum electrode di-
ameter is limited and consequently, so is the minimum electrode
distance. The spatial resolution is then limited to a certain spa-
tial frequency according to the electrode size and distance, both
for the monopolar array and the bipolar electrode array.
B. Cell-Length Selectivity of the Stimulation
In the intact retina, the processing of the contrast informa-
tion is afforded by the separation of the visual input into the
ON and OFF channels. This separation is achieved in the OPL,
where PR signaling leads either to a depolarization or hyperpo-
larization of the bipolar cells. We investigated whether the ON
and OFF bipolar cells can be selectively electrically stimulated
by using their different axon lengths as a filter. Axons of dif-
ferent lengths are unequally polarized by the same field: Long
axons (ON cells) are more polarized by the monopolar elec-
trodes than the short axons, whereas the bipolar electrodes affect
the short axons (OFF cells) stronger than the long axons [Fig. 7,
Fig. 9(a) and (b)] The most selective stimulation is achieved ei-
ther by short stimulation pulses applied by a bipolar array that
8 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 18, NO. 1, FEBRUARY 2010
Fig. 9. Polarization (t = 300 s
 ) of the axon terminal as a function of the cell
localization for (a) a monopolar and (c) a bipolar electrode array. (a) and (c):
membrane voltage of cells of different lengths, stimulated with 1 V. (b) and (c):
m
membrane voltage (cell length: 70  ) as a function of stimulation amplitude.
predominantly polarizes the short OFF cells (Fig. 7) or by the
large monopolar electrodes that address the long ON cells with
a higher efficacy than the short OFF cells (Fig. 10). However,
the axon length is not a highly selective filter and thus not ap-
propriate for the selective activation of the bipolar cells.
Fig. 10. Dependence of the steady-state membrane voltage on the electrode
size and cell length. The cell model was placed above a monopolar (top) and a
bipolar (bottom) electrode.
C. Conclusion and Limitations
To increase the spatial modulation of the polarization of
the axon terminals of the bipolar cells, bipolar arrays offer a
promising approach when the stimulation must simultaneously
take place at multiple retinal sites. We found that the short OFF
cells are predominantly addressed by the bipolar electrodes
(diameter between 60 and 100 ) and by using a short dura-
tion ( ) of the stimulating voltage impulse. The long
ON cells are best addressed by the large monopolar electrodes
(diameter ) and long pulse duration ( ).
However, we also are aware of the limitations of our com-
putational model. First, we have assumed a homogeneous and
isotropic resistivity of the retinal tissue. In practice, the laminar
profile of resistivity in various layers of the retina [10] may af-
fect the intraretinal potential profile and thus the polarization of
the bipolar cells. Second, the optimal pulse durations strongly
depend on the geometry of the cell. Therefore, the inclusion of
a cell soma or a variation in the diameter and stratification of
the axon may also affect the outcome. Third, the voltage-sensi-
tive ion channels activated by the changes in the transmembrane
GERHARDT et al.: ELECTRIC FIELD STIMULATION OF BIPOLAR CELLS IN A DEGENERATED RETINA—A THEORETICAL STUDY
voltage may affect the dynamics of the PST depolarization. Fi-
nally, the dependence of the membrane polarization on the dis-
tance between the electrode array and the soma-dendritic ter-
minal of the bipolar cells was neglected in this study, although
it is likely to be highly variable in practice. This additionally re-
duces the selectivity of the electric field stimulation with regard
to the length of the bipolar cells, especially when remodeling
occurs in the degenerated retinas [47], [48].
ACKNOWLEDGMENT
The authors would like to thank Prof. E. Zrenner, Prof.
A. Rothermel and the Retina Implant AG for inspiration and
fruitful discussions, and Dr. D. W. M. Arrigan for proofreading
the manuscript.
REFERENCES
[1] G. B. Awatramani and M. M. Slaughter, “Origin of transient and sus-
tained responses in ganglion cells of the retina,” J. Neurosci., vol. 20,
pp. 7087–7095, Sep. 2000.
[2] J. N. Burghartz, T. Engelhardt, H. G. Graf, C. Harendt, H. Richter, C.
Scherjon, and K. Warkentin, “CMOS imager technologies for biomed-
ical applications,” in 2008 ISSCC Proc., pp. 142–602.
[3] R. H. Carpenter, “Electrical stimulation of the human eye in different
adaptational states,” J. Physiol., vol. 221, pp. 137–148, Feb. 1972.
[4] D. R. Crapper and W. K. Noell, “Retinal excitation and inhibition from
direct elctrical stimulation,” J. Neurophysiol., vol. 26, pp. 924–947,
Nov. 1963.
[5] A. Y. Chow, M. T. Pardue, J. I. Perlman, S. L. Ball, V. Y. Chow, J.
R. Hetling, G. A. Peyman, C. Liang, E. B. Stubbs, and N. S. Peachey,
“Subretinal implantation of semiconductor-based photodiodes: dura-
bility of novel implant designs,” J. Rehabil. Res. Dev., vol. 39, pp.
313–321, May 2002.
[6] F. Werblin, “Synaptic connections, receptive fields, and patterns of ac-
tivity in the tiger salamander retina. A simulation of patterns of ac-
tivity formed at each cellular level from photoreceptors to ganglion
cells [the Friendenwald lecture],” Invest. Ophthalmol. Vis. Sci., vol. 32,
pp. 459–483, Mar. 1991.
[7] J. E. Dowling, The Retina. Cambridge, MA: Havard Univ. Press,
1987.
[8] R. Eckmiller, “Learning retina implants with epiretinal contacts,” Oph-
thalmic. Res., vol. 29, pp. 281–289, May 1997.
[9] R. N. Fariss, S. S. Apte, P. J. Luthert, A. C. Bird, and A. H. Milam,
“Accumulation of tissue inhibitor of metalloproteinases-3 in human
eyes with Sorsby’s fundus dystrophy or retinitis pigmentosa,” Br. J.
Ophthalmol., vol. 82, pp. 1329–1334, Nov. 1998.
[10] C. J. Karwoski and X. Xu, “Current source-density analysis of light-
evoked field potentials in rabbit retina,” Vis. Neurosci., vol. 16, pp.
369–377, 1999.
[11] M. S. Humayun, E. de Juan, G. Dagnelie, R. J. Greenberg, R. H. Propst,
and D. H. Phillips, “Visual perception elicited by electrical stimulation
of retina in blind humans,” Arch. Ophthalmol., vol. 114, pp. 40–46, Jan.
1996.
[12] M. J. Lankheet, M. H. Rowe, R. J. V. Wezel, and W. A. van de Grind,
“Spatial and temporal properties of cat horizontal cells after prolonged
dark adaptation,” Vis. Res., vol. 36, pp. 3955–3967, Dec. 1996.
[13] Y.-P. Ma and Z.-H. Pan, “Spontaneous regenerative activity in mam-
malian retinal bipolar cells: roles of multiple subtypes of voltage-de-
pendent Ca channels,” Vis. Neurosci., vol. 20, pp. 131–139, 2003.
[14] D. Palanker, A. Vankov, P. Huie, and S. Baccus, “Design of a high-
resolution optoelectronic retinal prosthesis,” J. Neural. Eng., vol. 2, pp.
S105–S120, Mar. 2005.
[15] Z. H. Pan, H. J. Hu, P. Perring, and R. Andrade, “T-type Ca(2+) chan-
nels mediate neurotransmitter release in retinal bipolar cells,” Neuron,
vol. 32, pp. 89–98, Oct. 2001.
[16] J. D. Loudin, D. M. Simanovskii, K. Vijayraghavan, C. K. Sramek, A.
F. Butterwick, P. Huie, G. Y. McLean, and D. V. Palanker, “Optoelec-
tronic retinal prosthesis: system design and performance,” J. Neural.
Eng., vol. 4, pp. S72–S84, Mar. 2007.
[17] I. Press and H. William, Recipes in C. Cambridge, U.K.: Cambridge
Univ. Press, 1988.
9
[18] F. Rattay, Electrical Nerve Stimulation. New York: Springer Verlag,
1990.
[19] J. F. Rizzo, J. Wyatt, J. Loewenstein, S. Kelly, and D. Shire, “Perceptual
efficacy of electrical stimulation of human retina with a microelectrode
array during short-term surgical trials,” Invest. Ophthalmol. Vis. Sci.,
vol. 44, pp. 5362–5369, Dec. 2003.
[20] J. F. Rizzo, J. Wyatt, J. Loewenstein, S. Kelly, and D. Shire, “Methods
and perceptual thresholds for short-term electrical stimulation of
human retina with microelectrode arrays,” Invest. Ophthalmol. Vis.
Sci., vol. 44, pp. 5355–5361, Dec. 2003.
[21] H. Satoh, K. Aoki, S. I. Watanabe, and A. Kaneko, “L-type calcium
channels in the axon terminal of mouse bipolar cells,” Neuroreport,
vol. 9, pp. 2161–2165, Jul. 1998.
[22] R. Segev, J. Goodhouse, J. Puchalla, and M. J. Berry, “Recording spikes
from a large fraction of the ganglion cells in a retinal patch,” Nat. Neu-
rosci., vol. 7, pp. 1154–1161, Oct. 2004.
[23] R. Siminoff, “Dynamics of chromaticity horizontal cells in the fresh-
water turtle retina,” Biol. Cybern., vol. 54, pp. 269–279, 1986.
[24] A. Stett, W. Barth, S. Weiss, H. Haemmerle, and E. Zrenner, “Electrical
multisite stimulation of the isolated chicken retina,” Vision. Res., vol.
40, pp. 1785–1795, 2000.
[25] H. C. Tuckwell, Introduction to Theoretical Neurobiology. Cam-
bridge, U.K.: Cambridge Univ. Press, 1988, vol. 1.
[26] M. Gerhardt and A. Stett, “Subretinal stimulation with hyperpolarising
and depolarising voltage steps,” in Proc. MEA Meeting 2008. BIOPRO
Edition, vol. 5, pp. 144–147.
[27] B. B. Boycott and H. Wässle, “Morphological classification of bipolar
cells of the primate retina,” Eur .J. Neurosci., vol. 3, pp. 1069–1088,
Jun. 1991.
[28] H. Wässle, “Parallel processing in the mammalian retina,” Nat. Rev.
Neurosci., vol. 5, pp. 747–757, Oct. 2004.
[29] H. Weber, Laplace-Transformation für Ingenieure der Elek-
trotechnik. Stuttgart, Germany: B. G. Teubner, 1990.
[30] E. Zrenner, “The subretinal implant: Can microphotodiode arrays re-
place degenerated retinal photoreceptors to restore vision?,” Ophthal-
mologica., vol. 216, no. Suppl 1, pp. 8–20, 2002.
[31] E. Zrenner, D. Besch, K. U. Bartz-Schmidt, F. Gekeler, V. P. Gabel, G.
Kuttenkeuler, W. Nisch, H. Sachs, H. Sailer, A. Stett, B. Wilhelm, and
R. Wilke, “Subretinal chronic multi-electrode arrays in blind patients:
function testing and pattern recognition,” in Proc. MEA Meeting 2006
BIOPRO, vol. 3, p. 90.
[32] E. Zrenner, A. Stett, S. Weiss, R. B. Aramant, E. Guenther, K. Kohler,
K. D. Miliczek, M. J. Seiler, and H. Haemmerle, “Can subretinal
microphotodiodes successfully replace degenerated photoreceptors?,”
Vis. Res., vol. 39, pp. 2555–2567, Jul. 1999.
[33] E. Zrenner, R. Wilke, T. Zabel1, H. Sachs, K. Bartz-Schmidt, F.
Gekeler, B. Wilhelm, U. Greppmaier, and A. Stett, “Psychometric
analysis of visual sensations mediated by subretinal microelectrode
arrays implanted into blind retinitis pigmentosa patients,” in ARVO
Proc. Invest. Ophthalmol. Vis. Sci., 2007, p. 48.
[34] H. Kolb, E. Fernandez, and R. J. Nelson, Webvison 2008 [Online].
Available: http://www.atm.com http://webvision.med.utah.edu
[35] A. K. Ahuja, M. R. Behrend, M. Koroda, M. S. Humayun, and J.
D. Weiland, “An in vitro model of a retinal prosthesis,” IEEE Trans
Biomed., vol. 55, no. 6, pp. 1744–1753, Jun. 2008.
[36] S. Joucla and B. Yvert, “Improved focalization of electrical micro-stim-
ulation using microelectrode arrays: A modeling study,” Plos ONE.,
vol. 4, p. e4828, Mar. 2009.
[37] K. K. Gosh, S. Bujan, S. Haverkamp, A. Feigenspan, and H. Wässle,
“Types of bipolar cells in the mouse retina,” J. Comp. Neurol., vol. 469,
pp. 70–82, Jun. 2004.
[38] N. Cuenca, I. Pinilla, Y. Sauvé, and R. Lund, “Early changes in synaptic
connectivity following progressive photoreceptor degeneration in RCS
rats,” Eur. J. Neurosci., vol. 22, pp. 1057–72, Jun. 2005.
[39] K. Kohler, J. A. Hartmann, D. Werts, and E. Zrenner, “Histological
studies of retinal degeneration and biocompatibility of subretinal im-
plants,” Ophthalmologe, vol. 98, pp. 364–368, Apr. 2001.
[40] A. Santos, M. S. Humayun, E. J. de Juan, R. J. Greenburg, M. J. Marsh,
I. B. Klock, and A. H. Milam, “Preservation of the inner retina in re-
tinitis pigmentosa. A morphometric analysis,” Arch. Ophthalmol., vol.
115, p. 511, 1997.
[41] M. Gerhardt and A. Stett, “Simulation of extracellular stimulation of
bipolar cells with monopolar and dipolar electrode configuration,” in
Proc. MEA Meeting 2006. BIOPRO Edition, vol. 5, pp. 93–94.
[42] L. Oldetal, M. L. Veruki, and E. Hartveit, “Passive membrane proper-
ties and electrotonic signal processing in retinal rod bipolar cells,” J.
Physiol., vol. 587, pp. 829–849, Jan. 2009.
10 IEEE TRANSACTIONS ON NEURAL SYSTEMS AND REHABILITATION ENGINEERING, VOL. 18, NO. 1, FEBRUARY 2010
[43] A. Prinz and P. Fromherz, “Effect of neuritic cables on conductance
estimates for remote electrical synapses,” J. Neurophysiol., vol. 89, pp.
2215–24, Dec. 2003.
[44] F. Rattay et al., “Mechanisms of Electrical Stimulation with Neural
Prostheses,” Neuromodulation, vol. 6, no. 1, pp. 42–56, 2003.
[45] J. T. Rubinstein and F. A. Spelman, “Analytical theory for extracellular
electrical stimulation of nerve with focal electrodes. I. Passive unmyeli-
nated axon,” Biophys. J., vol. 54, no. 6, pp. 975–81, Dec. 1988.
[46] A. Stett, A. Mai, and T. Herrmann, “Retinal charge sensitivity and
spatial discrimination obtainable by subretinal implants: Key lessons
learned from isolated chicken retina,” J. Neural. Eng., vol. 4, pp. 7–16,
Mar. 2007.
[47] B. W. Jones and R. E. Marc, “Retinal remodeling during retinal degen-
eration,” Exp. Eye Res., vol. 81, no. 2, pp. 123–37, Aug. 2005.
[48] R. E. Marc et al., “Neural remodeling in retinal degeneration,” Prog.
Retin Eye Res., vol. 22, no. 5, pp. 607–55, Sep. 2003.
[49] E. Zrenner et al., “Blind retinitis pigmentosa patients can read letters
and recognize the direction of fine stripe patterns with subretinal elec-
tronic implants,” Invest. Ophthalmol. Vis. Sci., vol. 50, p. E-Abstract
4581, 2009.
[50] N. Lovell et al., “Advances in retinal neuroprosthetics,” in Handbook
of Neural Engineering. New York: Wiley–IEEE, 2007, p. 337.
[51] R. V. Shannon, “A model of safe levels for electrical stimulation,” IEEE
Trans. Biomed. Eng., vol. 39, pp. 424–426, 1992.
[52] C. Gargini et al., “Retinal organization in the retinal degeneration 10
(rd10) mutant mouse: A morphological and ERG study,” J. Comp.
Neurol., vol. 500, no. 2, pp. 222–38, Jan. 2007.
[53] J. Reilly, Applied Bioelectricity: From Electrical Stimulation to Elec-
tropathology. New York: Springer, 1998, p. 115.
Matthias Gerhardt was born in Bernau, Germany,
on November 16, 1977. He received the Dipl.Ing
degree from Technische Universitaet Berlin, Berlin,
Germany, in 2004, and the Dr.-Ing. degree from the
University of Ulm, Ulm, Germany, in 2009.
Currently, he is a Research Fellow at the Tyndall
National Institute Cork, Ireland. His research is
mainly focused on possible interfaces between
electronic devices and nerve tissues like the retina.
John Alderman, photograph and biography not available at the time of publi-
cation.
Alfred Stett received the M.S. degree (Diplom) in
physics and the doctorate (Dr.rer.nat.) in 1995, both
from the University of Ulm, Ulm, Germany.
He is the Deputy Managing Director of the Natural
and Medical Sciences Institute (NMI), Reutlingen,
Germany. From 1994 to 1996 he worked on bi-di-
rectional interfacing of neurons with semiconductor
devices at the Max-Plank-Institute for Biochemistry.
In 1996 he joined the University Eye Hospital, Tue-
bingen, and started his research on electrical retina
stimulation. Since 1998 he is at the NMI in Reut-
lingen, where he led several research and development projects in the field of
neuroprosthetics and development of tools for electrophysiology.