This research is funded by The National Health and Medical Research Council Project Grant (Grant Number 1109056).

<ol>
	<li>Greenberg, R. J., Velte, T. J., Humayun, M. S., Scarlatis, G. N., de Juan, E. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+computational+model+of+electrical+stimulation+of+the+retinal+ganglion+cell.">A computational model of electrical stimulation of the retinal ganglion cell.</a> IEEE Transactions on Bio-medical Engineering. 46 (5), 505-514 (1999).</li><li>Guo, T., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Mediating+retinal+ganglion+cell+spike+rates+using+high-frequency+electrical+stimulation.">Mediating retinal ganglion cell spike rates using high-frequency electrical stimulation.</a> Frontiers in Neuroscience. 13, 413 (2019).</li><li>Loizos, K., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Increasing+electrical+stimulation+efficacy+in+degenerated+retina:+Stimulus+waveform+design+in+a+multiscale+computational+model.">Increasing electrical stimulation efficacy in degenerated retina: Stimulus waveform design in a multiscale computational model.</a> IEEE Transactions on Neural Systems and Rehabilitation Engineering. 26 (6), 1111-1120 (2018).</li><li>Cao, X., Sui, X., Lyu, Q., Li, L., Chai, X. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Effects+of+different+three-dimensional+electrodes+on+epiretinal+electrical+stimulation+by+modeling+analysis.">Effects of different three-dimensional electrodes on epiretinal electrical stimulation by modeling analysis.</a> Journal of Neuroengineering and Rehabilitation. 12 (1), 73 (2015).</li><li>Wilke, R. G. H., Moghadam, G. K., Lovell, N. H., Suaning, G. J., Dokos, S. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Electric+crosstalk+impairs+spatial+resolution+of+multi-electrode+arrays+in+retinal+implants.">Electric crosstalk impairs spatial resolution of multi-electrode arrays in retinal implants.</a> Journal of Neural Engineering. 8 (4), 046016 (2011).</li><li>AC/DC module user's guide. COMSOL AB Available from: <a target="_blank" href="https://doc.comsol.com/5.4/doc/com.comsol.help.acdc/ACDCModuleUsersGuide.pdf">https://doc.comsol.com/5.4/doc/com.comsol.help.acdc/ACDCModuleUsersGuide.pdf</a> (2018)</li><li>Malmivuo, P., Malmivuo, J., Plonsey, R. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=.">.</a> Bioelectromagnetism: Principles and Applications of Bioelectric and Biomagnetic Fields. , (1995).</li><li>Rall, W. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Electrophysiology+of+a+dendritic+neuron+model.">Electrophysiology of a dendritic neuron model.</a> Biophysical Journal. 2, 145-167 (1962).</li><li>Carnevale, N. T., Hines, M. L. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=.">.</a> The Neuron Book. , (2006).</li><li>Rattay, F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+basic+mechanism+for+the+electrical+stimulation+of+the+nervous+system.">The basic mechanism for the electrical stimulation of the nervous system.</a> Neuroscience. 89 (2), 335-346 (1999).</li><li>Hodgkin, A. L., Huxley, A. F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+quantitative+description+of+membrane+current+and+its+application+to+conduction+and+excitation+in+nerve.">A quantitative description of membrane current and its application to conduction and excitation in nerve.</a> The Journal of Physiology. 117 (4), 500-544 (1952).</li><li>Liang, T., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Threshold+suprachoroidal-transretinal+stimulation+current+required+by+different-size+electrodes+in+rabbit+eyes.">Threshold suprachoroidal-transretinal stimulation current required by different-size electrodes in rabbit eyes.</a> Ophthalmic Research. 45 (3), 113-121 (2011).</li><li>Jensen, R. J., Rizzo, J. F., Ziv, O. R., Grumet, A., Wyatt, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Thresholds+for+activation+of+rabbit+retinal+ganglion+cells+with+an+ultrafine,+extracellular+microelectrode.">Thresholds for activation of rabbit retinal ganglion cells with an ultrafine, extracellular microelectrode.</a> Investigative Ophthalmology and Visual Science. 44 (8), 3533-3543 (2003).</li><li>Kim, W., Choi, M., Kim, S. -. W. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+normative+retinal+and+choroidal+thicknesses+of+the+rabbit+as+revealed+by+spectral+domain+optical+coherence+tomography.">The normative retinal and choroidal thicknesses of the rabbit as revealed by spectral domain optical coherence tomography.</a> Journal of the Korean Ophthalmological Society. 62 (3), 354-361 (2021).</li><li>Guo, T., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Influence+of+cell+morphology+in+a+computational+model+of+ON+and+OFF+retinal+ganglion+cells.">Influence of cell morphology in a computational model of ON and OFF retinal ganglion cells.</a> 35th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). 2013, 4553-4556 (2013).</li><li>Haberbosch, L., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Safety+aspects,+tolerability+and+modeling+of+retinofugal+alternating+current+stimulation.">Safety aspects, tolerability and modeling of retinofugal alternating current stimulation.</a> Frontiers in Neuroscience. 13, 783 (2019).</li><li>Sheasby, B. W., Fohlmeister, J. F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Impulse+encoding+across+the+dendritic+morphologies+of+retinal+ganglion+cells.">Impulse encoding across the dendritic morphologies of retinal ganglion cells.</a> Journal of Neurophysiology. 81 (4), 1685-1698 (1999).</li><li>Rockhill, R. L., Daly, F. J., MacNeil, M. A., Brown, S. P., Masland, R. H. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+diversity+of+ganglion+cells+in+a+mammalian+retina.">The diversity of ganglion cells in a mammalian retina.</a> Journal of Neuroscience. 22 (9), 3831-3843 (2002).</li><li>Lukasiewicz, P., Werblin, F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+slowly+inactivating+potassium+current+truncates+spike+activity+in+ganglion+cells+of+the+tiger+salamander+retina.">A slowly inactivating potassium current truncates spike activity in ganglion cells of the tiger salamander retina.</a> The Journal of Neuroscience: The Official Journal of the Society for Neuroscience. 8 (12), 4470-4481 (1988).</li><li>Van Rossum, G. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=.">.</a> Python Reference Manual. , (1995).</li><li>. Welcome to Spyder's Documentation - Spyder 5 documentation Available from: <a target="_blank" href="https://docs.spyder-idle.org/current/index.html">https://docs.spyder-idle.org/current/index.html</a> (2022)</li><li>Rattay, F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Ways+to+approximate+current-distance+relations+for+electrically+stimulated+fibers.">Ways to approximate current-distance relations for electrically stimulated fibers.</a> Journal of Theoretical Biology. 125 (3), 339-349 (1987).</li><li>Tsai, D., Morley, J. W., Suaning, G. J., Lovell, N. H. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Direct+activation+and+temporal+response+properties+of+rabbit+retinal+ganglion+cells+following+subretinal+stimulation.">Direct activation and temporal response properties of rabbit retinal ganglion cells following subretinal stimulation.</a> Journal of Neurophysiology. 102 (5), 2982-2993 (2009).</li><li>Tsai, D., Morley, J. W., Suaning, G. J., Lovell, N. H. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Frequency-dependent+reduction+of+voltage-gated+sodium+current+modulates+retinal+ganglion+cell+response+rate+to+electrical+stimulation.">Frequency-dependent reduction of voltage-gated sodium current modulates retinal ganglion cell response rate to electrical stimulation.</a> Journal of Neural Engineering. 8 (6), 066007 (2011).</li><li>Joucla, S., Gli&#232;re, A., Yvert, B. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Current+approaches+to+model+extracellular+electrical+neural+microstimulation.">Current approaches to model extracellular electrical neural microstimulation.</a> Frontiers in Computational Neuroscience. 8, 13 (2014).</li><li>. OpenFOAM Available from: <a target="_blank" href="https://www.openfoam.com/">https://www.openfoam.com/</a> (2022)</li><li>Barba, L., Forsyth, G. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=CFD+Python:+The+12+steps+to+Navier-Stokes+equations.">CFD Python: The 12 steps to Navier-Stokes equations.</a> Journal of Open Source Education. 1 (9), 21 (2018).</li></ol>

The authors declare no competing interests.

In this paper, we have demonstrated a modeling workflow that combined finite element and biophysical neuron modeling. The model is highly flexible, as it can be modified in its complexity to fit different purposes, and it provides a way to validate the results against empirical findings. We also demonstrated how we parameterized the model to enable automation.
The two-step modeling method combines the advantages of using FEM and neuron computational suite to solve the neuron's cable equation in the presence of an extracellular stimulation. An FEM is useful in accurately computing the extracellular field across the volume conductor, which often is impractical to solve analytically in the case of complex geometry or inhomogeneity of conductivity. The computational cost of this model is also relatively low, as a static condition is assumed.
While the described modeling method is advantageous in its ease of use and flexibility, there are limitations to this modeling workflow. Firstly, this method did not allow the presence of a neural membrane in calculating the electric field. Joucla et al.25 compared the two-step method to the whole FEM method, where the neural geometry and membrane properties were included in the FEM model. They showed that including the neuron in the electric field calculation would change the transmembrane potential calculation when a larger cell structure, such as a cell body, was included in the geometry. Specifically, the simplification of the neuron geometry in the two-step method means that the transmembrane potential of any point in a compartment is represented by the transmembrane potential at the node or the center point of the compartment. In contrast, the whole-FEM model proposed by Joucla included an explicit representation of the 3D geometry of the neuron, which enabled the individual evaluation of transmembrane potential on any point inside the compartment. Thus, the whole-FEM model might be more suitable if the exact shape and location of the transmembrane potential are needed. However, this method is computationally more expensive than the two-step method.
The second limitation of the modeling method concerns the availability of morphology and ion kinetics data. The model used here was based on the tiger salamander data, which has been used to model RGCs from other species, but there might have been differences in the types of ion channels present that have not been elucidated. Hence, it might be necessary in some cases to perform in vitro works to adjust the ion channel parameters.
Thirdly, the cost of the FEM software might be a constraint. In this case, an open-source FEM programme26 that has a built-in Poisson equations solver might be an alternative. Apart from the FEM software used, the software used in this workflow is free. While the FEM software used offers an intuitive GUI and a ready-to-use electric current modeling, it is possible to perform the extracellular value calculations in a general purpose programming software. However, this would necessitate manually defining the physical equations and the numerical methods to solve the equations27. Moreover, this method might be tedious when a complex tissue or electrode array geometry is to be used.

Visual neuroprostheses are a group of devices that deliver stimulations (electrical, light, etc.) to the neural cells in the visual pathway to create phosphenes or sensation of seeing the light. It is a treatment strategy that has been in clinical use for almost a decade for people with permanent blindness caused by degenerative retinal diseases. Typically, a complete system would include an external camera that captures the visual information around the user, a power supply and computing unit to process and translate the image to a series of electrical pulses, and an implanted electrode array that interfaces the neural tissue and deliver the electrical pulses to the neural cells. The working principle allows a visual neuroprosthesis to be placed in different sites along the visual pathway from the retina to the visual cortex, as long as it is downstream from the damaged tissue. A majority of current research in visual neuroprostheses focuses on increasing the efficacy of the stimulation and improving the spatial acuity to provide a more natural vision.
In the efforts to improve the efficacy of the stimulation, computational modeling has been a cost- and time-effective method to validate a prosthesis design and simulate its visual outcome. Computational modeling in this field gained popularity since 1999 as Greenberg1 modeled the response of a retinal ganglion cell to extracellular electrical stimuli. Since then, computational modeling has been used to optimize the parameters of the electrical pulse2,3 or the geometrical design of the electrode4,5. Despite the variation in complexity and research questions, these models work by determining the electrical voltage distribution in the medium (e.g., neural tissue) and estimating the electrical response that the neurons in the vicinity will produce due to the electrical voltage.
The electrical voltage distribution in a conductor can be found by solving the Poisson equations6 at all locations:
<img alt="Equation 1" src="/files/ftp_upload/63792/63792eq01.jpg" style="margin: auto;" />
<img alt="Equation 2" src="/files/ftp_upload/63792/63792eq02.jpg" style="margin: auto;" />
where E is the electric field, V the electric potential, J the current density, and&#160;&#963; is the electrical conductivity. The <img alt="Equation 12" src="/files/ftp_upload/63792/63792eq12.jpg" style="margin: auto;" /> in the equation indicates a gradient operator. In the case of stationary current, the following boundary conditions are imposed on the model:
<img alt="Equation 3" src="/files/ftp_upload/63792/63792eq3.jpg" style="margin: auto;" />
<img alt="Equation 4" src="/files/ftp_upload/63792/63792eq04.jpg" style="margin: auto;" />
where n is the normal to the surface, &#937; represents the boundary, and I0 represents the specific current. Together, they create electrical insulation at the external boundaries and create a current source for a selected boundary. If we assume a monopolar point source in a homogeneous medium with an isotropic conductivity, the extracellular electric potential at an arbitrary location can be calculated by7:
<img alt="Equation 5" src="/files/ftp_upload/63792/63792eq05.jpg" style="margin: auto;" />
where Ie&#160;is the current and is the distance between the electrode and the point of measurement. When the medium is inhomogeneous or anisotropic, or the electrode array has multiple electrodes, a computational suite to numerically solve the equations can be convenient. A finite-element modeling software6 breaks up the volume conductor into small sections known as &#39;elements&#39;. The elements are interconnected with each other such that the effects of change in one element influence change in others, and it solves the physical equations that serve to describe these elements. With the increasing computational speed of modern computers, this process can be completed within seconds. Once the electric potential is calculated, one can then estimate the electrical response of the neuron.
A neuron sends and receives information in the form of electrical signals. Such signals come in two forms - graded potentials and action potentials. Graded potentials are temporary changes to the membrane potential wherein the voltage across the membrane becomes more positive (depolarization) or negative (hyperpolarization). Graded potentials typically have localized effects. In cells that produce them, action potentials are all-or-nothing responses that can travel long distances along the length of an axon. Both graded and action potentials are sensitive to the electrical as well as the chemical environment. An action potential spike can be produced by various neuronal cell types, including the retinal ganglion cells, when a threshold transmembrane potential is crossed. The action potential spiking and propagation then trigger synaptic transmission of signals to downstream neurons. A neuron can be modeled as a cable that is divided into cylindrical segments, where each segment has capacitance and resistance due to the lipid bilayer membrane8. A neuron computational programme9 can estimate the electrical activity of an electrically-excitable cell by discretizing the cell into multiple compartments and solving the mathematical model10:
<img alt="Equation 6" src="/files/ftp_upload/63792/63792eq06.jpg" style="margin: auto;" />
In this equation, Cm is the membrane capacitance, Ve,n is the extracellular potential at node n, Vi,n the intracellular potential at node n, Rn the intracellular (longitudinal) resistance at node n, and Iion is the ionic current going through the ion channels at node n. The values of V from the FEM model are implemented as Ve,n for all nodes in the neuron when the stimulation is active.
The transmembrane currents from ion channels can be modeled by using Hodgkin-Huxley formulations11:
<img alt="Equation 7" src="/files/ftp_upload/63792/63792eq07.jpg" style="margin: auto;" />
where gi is the specific conductance of the channel, Vm the transmembrane potential (Vi,n - Ve,n) and Eion the reversal potential of the ion channel. For voltage-gated channels, such as Na channel, dimensionless parameters, m, and h, that describe the probability of opening or closing of the channels are introduced:
<img alt="Equation 8" src="/files/ftp_upload/63792/63792eq08.jpg" style="margin: auto;" />
where <img alt="Equation 9" src="/files/ftp_upload/63792/63792eq09.jpg" style="margin: auto;" /> is the maximum membrane conductance for the particular ion channel, and the values of parameters m and h are defined by differential equations:
<img alt="Equation 10" src="/files/ftp_upload/63792/63792eq10.jpg" style="margin: auto;" />
where &#945;x and&#160;&#946;x are voltage-dependent functions that define the rate constants of the ion channel. They generally take the form:
<img alt="Equation 11" src="/files/ftp_upload/63792/63792eq11v2.jpg" style="margin: auto;" />
The values of the parameters in these equations, including maximal conductance, as well as the constants A, B, C, and D, were typically found from empirical measurements.
With these building blocks, models of different complexities can be built by following the steps described. An FEM software is useful when the Poisson equation cannot be solved analytically, such as in the case of inhomogeneous or anisotropic conductance in the volume conductor or when the geometry of the electrode array is complex. After the extracellular potential values have been solved, the neuron cable model can then be numerically solved in the neuron computational software. Combining the two software enables the computation of a complex neuron cell or network to a non-uniform electric field.
A simple two-step model of a retinal ganglion cell under a suprachoroidal stimulation will be built using the aforementioned programs. In this study, the retinal ganglion cell will be subjected to a range of magnitudes of electrical current pulses. The location of the cell relative to the stimulus is also varied to show the distance-threshold relationship. Moreover, the study includes a validation of the computational result against an in vivo study of the cortical activation threshold using different sizes of stimulation electrode12, as well as an in vitro study showing the relationship between the electrode-neuron distance and the activation threshold13.

<table><tbody><tr><td>Computer workstation</td><td>N/A</td><td>N/A</td><td>Windows 64-bit operating system, at least 4GB of RAM, at least 3 GB of disk space</td></tr><tr><td>Anaconda Python</td><td>Anaconda Inc.</td><td>Version 3.9</td><td>The open source Individual Edition containing Python 3.9 and preinstalled packages to perform data manipulation, as well as Spyder Integrated Development Environment. It could be used to control the simulation, as well as to display and analyse the simulation data.</td></tr><tr><td>COMSOL Multiphysics</td><td>COMSOL</td><td>Version 5.6</td><td>The simulation suite to perform finite element modelling. The licence for the AC/DC module should be purchased. The Application Builder capability should be included in the licence to follow the automation tutorial.</td></tr><tr><td>NEURON</td><td>NEURON</td><td>Version 8.0</td><td>A freely-distributed software to perform the computation of neuronal cells and/or neural networks.</td></tr></tbody></table>

computational modeling of retinal neurons for visual prosthesis research - fundamental approaches

Computational modeling has become an increasingly important method in neural engineering due to its capacity to predict behaviors of in vivo and in vitro systems. This has the key advantage of minimizing the number of animals required in a given study by providing an often very precise prediction of physiological outcomes. In the field of visual prosthesis, computational modeling has an array of practical applications, including informing the design of an implantable electrode array and prediction of visual percepts that may be elicited through the delivery of electrical impulses from the said array. Some models described in the literature combine a three-dimensional (3D) morphology to compute the electric field and a cable model of the neuron or neural network of interest. To increase the accessibility of this two-step method to researchers who may have limited prior experience in computational modeling, we provide a video of the fundamental approaches to be taken in order to construct a computational model and utilize it in predicting the physiological and psychophysical outcomes of stimulation protocols deployed via a visual prosthesis. The guide comprises the steps to build a 3D model in a finite element modeling (FEM) software, the construction of a retinal ganglion cell model in a multi-compartmental neuron computational software, followed by the amalgamation of the two. A finite element modeling software to numerically solve physical equations would be used to solve electric field distribution in the electrical stimulations of tissue. Then, specialized software to simulate the electrical activities of a neural cell or network was used. To follow this tutorial, familiarity with the working principle of a neuroprosthesis, as well as neurophysiological concepts (e.g., action potential mechanism and an understanding of the Hodgkin-Huxley model), would be required.

We conducted two simulation protocols to demonstrate the use of the model. The first protocol involved varying the electrode size while keeping the location of the neuron and the electrical pulse parameters the same. The second protocol involved shifting the neuron in the x-direction in 100 &#181;m steps, while the size of the electrode remained constant. For both protocols, the pulse used was a single cathodic-first biphasic pulse of 0.25 ms width with a 0.05 ms interphase gap. For the first protocol, the radius of the electrode was varied to be 50, 150, 350, and 500 &#181;m, while for the second protocol, the radius of the electrode was kept at a constant 50 &#181;m.
The model described here showed that increasing the suprachoroidal electrode size at 0.25 ms pulse width increased the activation threshold of the model neuron (Figure 9A). This result reflected the in vivo findings from Liang et al.12, who showed that the cortical activation threshold increases with the increasing electrode size at this pulse width.
The magnitudes of the model&#39;s activation thresholds differ from the empirical findings because of several factors. Firstly, this model only involves a single RGC of a specific type, which may not be present in the group of cells being activated in the in vivo study. Next, this model did not include a retinal network, which may facilitate the activation of RGCs through excitatory inputs from the bipolar cells. Another possible reason for the discrepancy is the electrode-retina distance. It is possible that the electrode-retina distance in the in vivo study was lower than in this model due to anatomical variability or the surgery. Consequently, we overestimated the electrode-retina distance and thus the activation threshold. It is also important to note that, although this was not demonstrated in our results, modeling a single cell threshold would often underestimate the in vivo cortical threshold. This is due to the technical limitations in cortical measurements (primarily relating to the signal-to-noise ratio) that the cortical activity is typically only detected after multiple retinal ganglion cells have been activated. As a result, a discrepancy in the magnitude of the retinal and cortical activation thresholds is to be expected. Despite these differences, this model successfully showed the increasing trend of the activation threshold due to the increase in the electrode size. This resulted from the absence of an area of high electric field compared to its surroundings when the electrode size is increased, which did not favor neural activation22.
Next, we observed the action potential characteristics to validate the model described here. The latency, or the time between the stimulus onset and the peak of the action potential spike, ranged from 1-2.2 ms (Figure 9B). This corresponded to the short-latency spiking due to non-network mediated retinal activation23. The spike width of this model was 1 ms, and this is in the same range as the spike widths of rabbit RGCs measured in vitro24.
In the second stimulation protocol, only the location of the neuron in the x-axis (along the axon&#39;s length) relative to the electrode was varied. At a distance 0, the centroid of the soma section was immediately above the center of the disk electrode. Negative distance means that the disk electrode was positioned closer to the axonal side, while positive distance means that the disk electrode was positioned closer to the dendritic side. The model showed that the lowest threshold was achieved when the narrow segment of the axon was immediately above the disk electrode, and it increased as the x-distance became larger (Figure 9C). Moving the electrode further toward the distal axon produced a lower threshold as compared to moving the electrode toward the dendrites due to the presence of the axon initial segment and the narrow segment where the sodium channels are more prevalent. This result agreed with the in vitro finding from Jensen et al.13, where rabbit RGCs were stimulated with an ultrafine microelectrode, and the activation threshold was the highest when the electrode was shifted closer to the dendrites.
<img alt="Figure 9" class="xfigimg" src="/files/ftp_upload/63792/63792fig09.jpg" /> 
Figure 9: The results of the modeling method. (A) The activation thresholds for a retinal ganglion cell located above the disk electrode. The electrode radius was varied (50, 150, 350, and 500 &#181;m) and the threshold increased with increasing electrode size. (B) The neuron model&#39;s action potential shape at 0.25 ms pulse width. The action potentials at threshold for different electrode sizes have the same spike width of 1 ms, but the latency increased with increasing electrode size. The stimulus onset time was 1 ms and the cathodic phase caused a depolarization at the membrane but not enough to cause an action potential. (C) The neuron was shifted along the x-axis and the thresholds of activation showed that the lowest threshold was achieved by the neuron whose soma was located right above the center of the electrode. The radius of the electrode was 50 &#181;m. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/63792fig09large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
Supplementary Figure 1: Initialization of the finite element model. The types of Study and Physics determine the list of equations solved in the model. These were set during the initial creation of the FEM model file, but can also be modified/added after the model has been created. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary_Figure_1.psd" target="_blank">Please click here to download this File.</a>
Supplementary Figure 2: Changing the unit of length. The length unit and angular unit determine the units used in the geometry definition process. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary_Figure_2.psd" target="_blank">Please click here to download this File.</a>
Supplementary Figure 3: Inserting a material property. The material properties were defined for each domain in a 3D model. The available material properties were listed in the Material Properties in the Material Setting window. For the electric potential calculation, only the Electrical Conductivity property was defined. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary_Figure_3.psd" target="_blank">Please click here to download this File.</a>
Supplementary Figure 4: Creating a parametric study to loop over a list of parameter values. A parametric study allowed the FEM software to automatically repeat the computations and change the electrode radius value for each repeat. The calculation results were stored for each repeat. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary_Figure_4.psd" target="_blank">Please click here to download this File.</a>
Supplementary Figure 5: Importing the neuron morphology from the SWC file. The neuron computational suite was capable of reading SWC file acquired from neuronal tracing. The imported file contains information on the morphology and topology of each neuron segment. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary_Figure_5.psd" target="_blank">Please click here to download this File.</a>
Supplementary Figure 6: Automating FEM operations by defining a method. A method was defined by writing a script to automate processes in the FEM software that cannot be done by defining a parametric study. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary_Figure_6.psd" target="_blank">Please click here to download this File.</a>
Supplementary Figure 7: Integrating the models and automating the simulations using a general purpose programming language. The general purpose programming language was used to loop the neuron simulations, while changing the extracellular voltage file used as an input and the neural response voltage file as an output for each step in the loop. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary_Figure_7.psd" target="_blank">Please click here to download this File.</a>
Supplementary Material: Command lines for (1) Defining a voltage-dependent Cat channel. (2) Voltage- and concentration-dependent ion channels. (3) Complete .mod file. (4) Creating a biphasic pulse in the neuron simulation. (5) Calculating the coordinate of each node. (6) Application of the biphasic pulse. (7) Executing the neuron simulation. (8) Looping over a range of current amplitudes. (9) Defining a method to automate FEM simulations. (10) Running the simulations in a general purpose programming language. <a href="https://www-jove-com.remotexs.ntu.edu.sg/files/ftp_upload/63792/Supplementary Material.docx" target="_blank">Please click here to download this File.</a>

Watch this Scientific Journal Video about Computational Modeling of Retinal Neurons for Visual Prosthesis Research - Fundamental Approaches at JoVE.com

Computational Modeling of Retinal Neurons for Visual Prosthesis Research - Fundamental Approaches

We summarize a workflow to computationally model a retinal neuron's behaviors in response to electrical stimulation. The computational model is versatile and includes automation steps that are useful in simulating a range of physiological scenarios and anticipating the outcomes of future in vivo/in vitro studies.

Computational modeling has become an increasingly important method in neural engineering due to its capacity to predict behaviors of in vivo and in vitro ...

computational-modeling-retinal-neurons-for-visual-prosthesis-research

Nanyang Technological University

Research

JoVE Journal

Neuroscience

1.6K Views.  The University of Sydney. We summarize a workflow to computationally model a retinal neuron's behaviors in response to electrical stimulation. The computational model is versatile and includes automation steps that are useful in simulating a range of physiological scenarios and anticipating the outcomes of future in vivo/in vitro studies.

Video: Computational Modeling of Retinal Neurons for Visual Prosthesis Research - Fundamental Approaches

An Isolated Retinal Preparation to Record Light Response from Genetically Labeled Retinal Ganglion Cells

The first steps in vertebrate vision take place when light stimulates the rod and cone photoreceptors of the retina 1. This information is then segregated into what are known as the ON and OFF pathways. The photoreceptors signal light information to the bipolar cells (BCs), which depolarize in response to increases (On BCs) or decreases (Off BCs) in light intensity. This segregation of light information is maintained at the level of the retinal ganglion cells (RGCs), which have dendrites stratifying in either the Off sublamina of the inner plexiform layer (IPL), where they receive direct excitatory input from Off BCs, or stratifying in the On sublamina of the IPL, where they receive direct excitatory input from On BCs. This segregation of information regarding increases or decreases in illumination (the On and Off pathways) is conserved and signaled to the brain in parallel.
The RGCs are the output cells of the retina, and are thus an important cell to study in order to understand how light information is signaled to visual nuclei in the brain. Advances in mouse genetics over recent decades have resulted in a variety of fluorescent reporter mouse lines where specific RGC populations are labeled with a fluorescent protein to allow for identification of RGC subtypes 2 3 4 and specific targeting for electrophysiological recording. Here, we present a method for recording light responses from fluorescently labeled ganglion cells in an intact, isolated retinal preparation. This isolated retinal preparation allows for recordings from RGCs where the dendritic arbor is intact and the inputs across the entire RGC dendritic arbor are preserved. This method is applicable across a variety of ganglion cell subtypes and is amenable to a wide variety of single-cell physiological techniques.

This article provides a description of how to dissect and record from the isolated retinal preparation in mouse. In particular, we describe how to record light responses from a fluorescently labeled ganglion cell population and subsequently identify and analyze its morphology.

The first steps in vertebrate vision take place when light stimulates the rod and cone photoreceptors of the retina 1.  This information is then ...

Techniques for Processing Eyes Implanted With a Retinal Prosthesis for Localized Histopathological Analysis

With the recent development of retinal prostheses, it is important to develop reliable techniques for assessing the safety of these devices in preclinical studies. However, the standard fixation, preparation, and automated histology procedures are not ideal. Here we describe new procedures for evaluating the health of the retina directly adjacent to an implant. Retinal prostheses feature electrode arrays in contact with eye tissue. Previous methods have not been able to spatially localize the ocular tissue adjacent to individual electrodes within the array. In addition, standard histological processing often results in gross artifactual detachment of the retinal layers when assessing implanted eyes. Consequently, it has been difficult to assess localized damage, if present, caused by implantation and stimulation of an implanted electrode array. Therefore, we developed a method for identifying and localizing the ocular tissue adjacent to implanted electrodes using a (color-coded) dye marking scheme, and we modified an eye fixation technique to minimize artifactual retinal detachment. This method also rendered the sclera translucent, enabling localization of individual electrodes and specific parts of an implant. Finally, we used a matched control to increase the power of the histopathological assessments. In summary, this method enables reliable and efficient discrimination and assessment of the retinal cytoarchitecture in an implanted eye.

Techniques for visualizing retinal cytoarchitecture directly adjacent to individual electrodes within a retinal stimulator.

With  the recent development of retinal prostheses, it is important to develop  reliable techniques for assessing the safety of these devices in ...

Medicine

Using Looming Visual Stimuli to Evaluate Mouse Vision

The visual system in the central nervous system processes diverse visual signals. Although the overall structure has been characterized from the retina through the lateral geniculate nucleus to the visual cortex, the system is complex. Cellular and molecular studies have been conducted to elucidate the mechanisms underpinning visual processing and, by extension, disease mechanisms. These studies may contribute to the development of artificial visual systems. To validate the results of these studies, behavioral vision testing is necessary. Here, we show that the looming stimulation experiment is a reliable mouse vision test that requires a relatively simple setup. The looming experiment was conducted in a large enclosure with a shelter in one corner and a computer monitor located on the ceiling. A CCD camera positioned next to the computer monitor served to observe mouse behavior. A mouse was placed in the enclosure for 10 minutes and allowed to acclimate to and explore the surroundings. Then, the monitor projected a program-derived looming stimulus 10 times. The mouse responded to the stimuli either by freezing or by fleeing to the hiding place. The mouse&#39;s behavior before and after the looming stimuli was recorded, and the video was analyzed using motion tracking software. The velocity of the mouse movement significantly changed after the looming stimuli. In contrast, no reaction was observed in blind mice. Our results demonstrate that the simple looming experiment is a reliable test of mouse vision.

To examine mouse vision, we conducted a looming test. Mice were placed in a large square arena with a monitor on its ceiling. The looming visual stimulus consistently evoked freezing or flight reactions in the mice.

The visual system in the central nervous system processes diverse visual signals. Although the overall structure has been characterized from the retina ...

Behavior

Immunohistochemical and Calcium Imaging Methods in Wholemount Rat Retina

In this paper we describe the tools, reagents, and the practical steps that are needed for: 1) successful preparation of wholemount retinas for immunohistochemistry and, 2) calcium imaging for the study of voltage gated calcium channel (VGCC) mediated calcium signaling in retinal ganglion cells. The calcium imaging method we describe circumvents issues concerning non-specific loading of displaced amacrine cells in the ganglion cell layer.

Immunohistochemistry protocols are used to study the localization of a specific protein in the retina. Calcium imaging techniques are employed to study calcium dynamics in retinal ganglion cells and their axons.

In this paper we describe the tools, reagents, and the practical steps that are needed for: 1) successful preparation of wholemount retinas for ...

Methodology for Biomimetic Chemical Neuromodulation of Rat Retinas with the Neurotransmitter Glutamate In Vitro

Photoreceptor degenerative diseases cause irreparable blindness through the progressive loss of photoreceptor cells in the retina. Retinal prostheses are an emerging treatment for photoreceptor degenerative diseases that seek to restore vision by artificially stimulating the surviving retinal neurons in the hope of eliciting comprehensible visual perception in patients. Current retinal prostheses have demonstrated success in restoring limited vision to patients using an array of electrodes to electrically stimulate the retina but face substantial physical barriers in restoring high acuity, natural vision to patients. Chemical neurostimulation using native neurotransmitters is a biomimetic alternative to electrical stimulation and could bypass the fundamental limitations associated with retinal prostheses using electrical neurostimulation. Specifically, chemical neurostimulation has the potential to restore more natural vision with comparable or better visual acuities to patients by injecting very small quantities of neurotransmitters, the same natural agents of communication used by retinal chemical synapses, at much finer resolution than current electrical prostheses. However, as a relatively unexplored stimulation paradigm, there is no established protocol for achieving chemical stimulation of the retina in vitro. The purpose of this work is to provide a detailed framework for accomplishing chemical stimulation of the retina for investigators who wish to study the potential of chemical neuromodulation of the retina or similar neural tissues in vitro. In this work, we describe the experimental setup and methodology for eliciting retinal ganglion cell (RGC) spike responses similar to visual light responses in wild-type and photoreceptor-degenerated wholemount rat retinas by injecting controlled volumes of the neurotransmitter glutamate into the subretinal space using glass micropipettes and a custom multiport microfluidic device. This methodology and protocol are general enough to be adapted for neuromodulation using other neurotransmitters or even other neural tissues.

Methodology for Biomimetic Chemical Neuromodulation of Rat Retinas with the Neurotransmitter Glutamate In Vitro

This protocol describes a novel method for investigating a form of chemical neurostimulation of wholemount rat retinas in vitro with the neurotransmitter glutamate. Chemical neurostimulation is a promising alternative to the conventional electrical neurostimulation of retinal neurons for treating irreversible blindness caused by photoreceptor degenerative diseases.

Photoreceptor degenerative diseases cause irreparable blindness through the progressive loss of photoreceptor cells in the retina. Retinal prostheses are ...

Bioengineering

Generating Retinal Degeneration to Investigate Cellular Repair Strategies in Xenopus

Generating Retinal Injury Models in Xenopus Tadpoles

Retinal neurodegenerative diseases are the leading causes of blindness. Among numerous therapeutic strategies being explored, stimulating self-repair recently emerged as particularly appealing. A cellular source of interest for retinal repair is the M&#252;ller glial cell, which harbors stem cell potential and an extraordinary regenerative capacity in anamniotes. This potential is, however, very limited in mammals. Studying the molecular mechanisms underlying retinal regeneration in animal models with regenerative capabilities should provide insights into how to unlock the latent ability of mammalian M&#252;ller cells to regenerate the retina. This is a key step for the development of therapeutic strategies in regenerative medicine. To this aim, we developed several retinal injury paradigms in Xenopus: a mechanical retinal injury, a transgenic line allowing for nitroreductase-mediated photoreceptor conditional ablation, a retinitis pigmentosa model based on CRISPR/Cas9-mediated rhodopsin knockout, and a cytotoxic model driven by intraocular CoCl2 injections. Highlighting their advantages and disadvantages, we describe here this series of protocols that generate various degenerative conditions and allow the study of retinal regeneration in Xenopus.

Author Spotlight: Exploring Retinal Regeneration Mechanisms in the Xenopus Frog 

We have developed several protocols to induce retinal damage or retinal degeneration in Xenopus laevis tadpoles. These models offer the possibility of studying retinal regeneration mechanisms.

Retinal neurodegenerative diseases are the leading causes of blindness. Among numerous therapeutic strategies being explored, stimulating self-repair ...

Developmental Biology

Enhanced 3D Neural Retina Differentiation from Human Embryonic Stem Cells

Generating Neural Retina from Human Pluripotent Stem Cells

Retinopathy is one of the main causes of blindness worldwide. Investigating its pathogenesis is essential for the early diagnosis and timely treatment of retinopathy. Unfortunately, ethical barriers hinder the collection of evidence from humans. Recently, numerous studies have shown that human pluripotent stem cells (PSCs) can be differentiated into retinal organoids (ROs) using different induction protocols, which have enormous potential in retinopathy for disease modeling, drug screening, and stem cell-based therapies. This study describes an optimized induction protocol to generate neural retina (NR) that significantly reduces the probability of vesiculation and fusion, increasing the success rate of production until day 60. Based on the ability of PSCs to self-reorganize after dissociation, combined with certain complementary factors, this new method can specifically drive NR differentiation. Furthermore, the approach is uncomplicated, cost-effective, exhibits notable repeatability and efficiency, presents encouraging prospects for personalized models of retinal diseases, and supplies a plentiful cell reservoir for applications such as cell therapy, drug screening, and gene therapy testing.

Author Spotlight: Simple and Efficient Neural Retina Organoid Production for Disease Modeling

The present protocol describes an optimized 3D neural retina induction system that reduces the adhesion and fusion of retinal organoids with high repeatability and efficiency.

Retinopathy is one of the main causes of blindness worldwide. Investigating its pathogenesis is essential for the early diagnosis and timely treatment of ...

Subretinal Transplantation of Human Embryonic Stem Cell-Derived Retinal Tissue in a Feline Large Animal Model

Retinal degenerative (RD) conditions associated with photoreceptor loss such as age-related macular degeneration (AMD), retinitis pigmentosa (RP) and Leber Congenital Amaurosis (LCA) cause progressive and debilitating vision loss. There is an unmet need for therapies that can restore vision once photoreceptors have been lost. Transplantation of human pluripotent stem cell (hPSC)-derived retinal tissue (organoids) into the subretinal space of an eye with advanced RD brings retinal tissue sheets with thousands of healthy mutation-free photoreceptors and has a potential to treat most/all blinding diseases associated with photoreceptor degeneration with one approved protocol. Transplantation of fetal retinal tissue into the subretinal space of animal models and people with advanced RD has been developed successfully but cannot be used as a routine therapy due to ethical concerns and limited tissue supply. Large eye inherited retinal degeneration (IRD) animal models are valuable for developing vision restoration therapies utilizing advanced surgical approaches to transplant retinal cells/tissue into the subretinal space. The similarities in globe size, and photoreceptor distribution (e.g., presence of macula-like region area centralis) and availability of IRD models closely recapitulating human IRD would facilitate rapid translation of a promising therapy to the clinic. Presented here is a surgical technique of transplanting hPSC-derived retinal tissue into the subretinal space of a large animal model allowing assessment of this promising approach in animal models.

Presented here is a surgical technique for transplanting human pluripotent stem cell (hPSC)-derived retinal tissue into the subretinal space of a large animal model.

Retinal degenerative (RD) conditions associated with photoreceptor loss such as age-related macular degeneration (AMD), retinitis pigmentosa (RP) and ...

Time-Lapse Imaging of Neuronal Arborization using Sparse Adeno-Associated Virus Labeling of Genetically Targeted Retinal Cell Populations

Discovering mechanisms that pattern dendritic arbors requires methods to visualize, image, and analyze dendrites during development. The mouse retina is a powerful model system for the investigation of cell type-specific mechanisms of neuronal morphogenesis and connectivity. The organization and composition of retinal subtypes are well-defined, and genetic tools are available to access specific types during development. Many retinal cell types also constrain their dendrites and/or axons to narrow layers, which facilitates time-lapse imaging. Mouse retina explant cultures are well suited for live-cell imaging using confocal or multiphoton microscopy, but methods optimized for imaging dendrite dynamics with temporal and structural resolution are lacking. Presented here is a method to sparsely label and image the development of specific retinal populations marked by the Cre-Lox system. Commercially available adeno-associated viruses (AAVs) used here expressed membrane-targeted fluorescent proteins in a Cre-dependent manner. Intraocular delivery of AAVs in neonatal mice produces fluorescent labeling of targeted cell types by 4-5 days post-injection (dpi). The membrane fluorescent signals are detectable by confocal imaging and resolve fine branch structures and dynamics. High-quality videos spanning 2-4 h are acquired from imaging retinal flat-mounts perfused with oxygenated artificial cerebrospinal fluid (aCSF). Also provided is an image postprocessing pipeline for deconvolution and three-dimensional (3D) drift correction. This protocol can be used to capture several cellular behaviors in the intact retina and to identify novel factors controlling neurite morphogenesis. Many developmental strategies learned in the retina will be relevant for understanding the formation of neural circuits elsewhere in the central nervous system.

Here, we present a method for investigating neurite morphogenesis in postnatal mouse retinal explants by time-lapse confocal microscopy. We describe an approach for sparse labeling and acquisition of retinal cell types and their fine processes using recombinant adeno-associated virus vectors that express membrane-targeted fluorescent proteins in a Cre-dependent manner.

Discovering mechanisms that pattern dendritic arbors requires methods to visualize, image, and analyze dendrites during development. The mouse retina is a ...

Assessing Retinal Thickness and Ganglion Cell Layer Counts in Rodent Eyeballs

Preparation and Analysis of Histological Slides of Rat and Mouse Eyeballs to Evaluate the Retina

A rodent eyeball is a powerful tool for researching the pathomechanisms of many ophthalmic diseases, such as glaucoma, hypertensive retinopathy, and many more. Preclinical experiments enable researchers to examine the efficacy of novel drugs, develop new methods of treatment, or seek new pathomechanisms involved in the disease's onset or progression. A histological examination provides a lot of information necessary to assess the effects of the conducted experiments and can reveal degeneration, tissue remodeling, infiltration, and many other pathologies. In clinical research, there is rarely any chance of obtaining eye tissue suitable for a histological examination, which is why researchers should take advantage of the opportunity offered by the examination of eyeballs from rodents. This manuscript presents a protocol for the histological preparation of rodent eyeballs' sections. The procedure is presented for the eyeballs of mice and rats and has the following steps: (i) harvesting the eyeball, (ii) preserving the eyeball for further analysis, (iii) processing the tissue in paraffin, (iv) preparing slides, (v) staining with hematoxylin and eosin, (vi) assessing the tissue under a light microscope. With the proposed method, the retina can be easily visualized and assessed in detail.

This manuscript presents a protocol for the histological preparation of slides from rodent eyeballs. With the proposed method, the inner retina can be easily visualized and assessed under a light microscope. The procedure is presented for the eyeballs of mice and rats.

A rodent eyeball is a powerful tool for researching the pathomechanisms of many ophthalmic diseases, such as glaucoma, hypertensive retinopathy, and many ...

Computational Modeling of Retinal Neurons for Visual Prosthesis Research - Fundamental Approaches

Podsumowanie

Przeglądaj więcej filmów

Rozdziały w tym wideo

An Isolated Retinal Preparation to Record Light Response from Genetically Labeled Retinal Ganglion Cells

Techniques for Processing Eyes Implanted With a Retinal Prosthesis for Localized Histopathological Analysis

Using Looming Visual Stimuli to Evaluate Mouse Vision

Immunohistochemical and Calcium Imaging Methods in Wholemount Rat Retina

Methodology for Biomimetic Chemical Neuromodulation of Rat Retinas with the Neurotransmitter Glutamate In Vitro

Generating Retinal Injury Models in Xenopus Tadpoles

Generating Neural Retina from Human Pluripotent Stem Cells

Subretinal Transplantation of Human Embryonic Stem Cell-Derived Retinal Tissue in a Feline Large Animal Model

Time-Lapse Imaging of Neuronal Arborization using Sparse Adeno-Associated Virus Labeling of Genetically Targeted Retinal Cell Populations

Preparation and Analysis of Histological Slides of Rat and Mouse Eyeballs to Evaluate the Retina