Honeycomb-shaped electrical neural interface for artificial retinas

Abstract

We present a three-dimensional configuration of an electrode array for neuronal stimulation designed to enhance migration of retinal cells into the subretinal space. Walls surrounding each pixel align the electric field vertically, matching the orientation of retinal bipolar cells and thereby lowering the stimulation threshold. These walls also decouple the effect of pixel width on field penetration depth, allowing pixel size to be scaled down to the cell dimensions. Inner retinal cells migrate into the electrode cavity, enabling highly efficient stimulation. By aligning the electric field in one dimension along the cavity, the stimulation threshold current density does not increase significantly with decreasing pixel size, eliminating the quadratic increase seen in planar arrays.

Description

[Technical Field]

[0001] This invention relates to the electrical stimulation of nerve cells. [Background technology]

[0002] Conventional stimulation arrays for artificial vision devices have a planar structure of active and return electrodes. This planar structure presents serious difficulties when scaling artificial vision devices to have pixels dense enough to provide small, useful vision. As pixel size decreases, two main problems arise: increased crosstalk between adjacent pixels and a higher pixel stimulation threshold (i.e., the current density required to obtain a cellular response). The increase in stimulation threshold due to smaller pixel size is a particularly troublesome problem, as the current density required at the desired pixel size for retinal implants becomes biologically dangerous. Therefore, mitigating these limitations of planar stimulation arrays would represent progress in this field. [Overview of the Initiative] [Problems that the invention aims to solve]

[0003] This invention provides a three-dimensional configuration of an electrode array for neuronal stimulation designed to improve the migration of retinal cells into the subretinal space. Walls surrounding each pixel vertically align the electric field, aligning it with the direction of the bipolar cells in the retina, thereby lowering the stimulation threshold. These walls also decouple the effect of pixel width on electric field penetration depth, allowing pixel sizes to be reduced to cell dimensions. Because inner retinal cells migrate to the honeycomb well, these neurons reside within the electrode cavity, enabling highly efficient stimulation. By aligning the electric field one-dimensionally along the cavity, the stimulation threshold current density does not increase significantly with decreasing pixel size, avoiding the quadratic increase seen in planar arrays. This three-dimensional electrode configuration offers the potential to restore vision with pixels much smaller than those in planar electrode arrays. Similar three-dimensional arrays can also be used for electrical neural interfaces with the brain in other applications.

[0004] In a preferred embodiment, the pixel cavity has a depth between 10 μm and 100 μm and a width between 5 μm and 100 μm. The current density at the active central electrode is preferably 0.01 A / cm². 2 ~1A / cm 2 The overall width of the device is preferably between 0.5 mm and 5 mm, making it suitable for retinal implantation. The charge density injected per pulse into the active electrode is preferably 0.1 mC / cm². 2 ~10mC / cm 2 It is between these two points.

[0005] Applications of the present invention include artificial vision devices for restoring vision in retinal degeneration, and general high-resolution electrical neural interfaces. The implementation of the present invention provides the advantageous effect of reducing the size of the pixels in the electrical neural interface to the size of a cell. [Brief explanation of the drawing]

[0006] [Figure 1] Figure 1 is a schematic diagram illustrating the operation of a planar array for nerve cell stimulation. [Figure 2A] Figure 2A shows a first embodiment of the present invention. [Figure 2B] Figure 2B shows a second embodiment of the present invention. [Figure 3-1] Figure 3-1 consists of (A) and (B) and illustrates the difference between a planar stimulus array (A) and a non-planar stimulus array (B). [Figure 3-2] Figure 3-2 consists of (C) and (D), which show the tissues that make up the upper layers of the implants in Figure 3-1 (A) and (B), respectively, and illustrates the anatomical structure of the retina. [Figure 4A] Figure 4A shows an image of the fabricated cavity array. [Figure 4B] Figure 4B shows an image of the fabricated cavity array. [Figure 4C] Figure 4C shows an image of a honeycomb-shaped implant under the retina of an animal. [Figure 5] Figure 5 shows the histology of the retina integrated with the implanted honeycomb array. [Figure 6A] Figure 6A shows the simulated potentials of a planar array (top) and a honeycomb array (bottom). [Figure 6B] Figure 6B shows experimental and simulated threshold current density versus pixel size for a planar array, compared to safety limits and compared to simulated threshold current density versus pixel size for a honeycomb array. [Modes for carrying out the invention]

[0007] (A) below describes the general principles related to embodiments of the present invention. (B) below describes the details of experimental examples.

[0008] (A) General principles To better understand embodiments of the present invention, it is useful to first examine the operation of a prior art planar stimulation array 100, shown in Figure 1. This figure is a cross-sectional view of a row of pixels 102, 104, and 106. The "on" pixels (102 and 106) generate an electric current, which is injected into the electrolyte through a central active electrode (e.g., 108 of pixel 102), and the current is then collected by a separation return electrode 110 outside (or between) the pixels. The flow of the current is indicated by the black arrows. The current flowing through the electrolyte generates a potential gradient (indicated by shading). The current spreads in all directions from the active electrode, affecting cells in adjacent pixels as well. For example, the "off" pixel 104, labeled X, receives parasitic stimulation from adjacent pixels, resulting in a decrease in contrast, which is an undesirable outcome.

[0009] Figure 2A shows a cross-section of a first embodiment of the present invention. In the embodiment of Figure 2A, the stimulation array 200 includes pixels 202, 204, and 206. The "on" pixels 202 and 206 generate an electric current, which is injected into the electrolyte through a central active electrode (e.g., 214 of pixel 202). The current flows upward through the electrolyte (tissue) and is collected by a return electrode, i.e., a return electrode 216, located at the top of wall 218, to complete the circuit. The current through the electrolyte generates a potential gradient (shown by shading) primarily in front of the "on" pixels, thereby enabling localized stimulation of the retina. Note the lighter shading above pixel 204 in Figure 2A compared to pixel 104 in Figure 1. The large-capacitance material (return electrode 216) deposited on top of the conductive honeycomb wall 218 ensures that most of the current flowing through the electrolyte is collected at the top of the wall, and the current collected through the sides of the conductive wall is negligible due to the small capacitance at the interface between the metal and the electrolyte. This configuration makes it possible to fabricate a highly return electrode with a honeycomb structure without requiring passivation or insulation of the side walls.

[0010] More precisely, a first embodiment of the present invention is a device for electrical stimulation of nerve cells, comprising an array of cavities configured to allow the movement of nerve cells within the cavities. Each cavity has a floor (e.g., 212 in Figure 2A) and a conductive wall (e.g., 218 in Figure 2A). Each cavity has a first electrode (e.g., 214 in Figure 2A) positioned in its floor and a second electrode (e.g., 216 in Figure 2A) positioned on the top of the wall and perpendicularly separated from the corresponding floor electrode. During operation of the device, an ionic current flows through the contents of the cavities. The capacitance of the second electrode per cavity is greater than the capacitance of the conductive wall per cavity, thereby allowing the second electrode to preferentially collect the ionic current from the sides of the conductive wall. Preferably, the capacitance per unit area of ​​the second electrode is at least 100 times greater than the capacitance per unit area of ​​the conductive wall.

[0011] Figure 2B shows a cross-sectional view of a second embodiment of the present invention, where the stimulation array 250 includes pixels 252, 254, and 256. The "on" pixels 252 and 256 generate an electric current, which is injected into the electrolyte through a central active electrode (e.g., 214 of pixel 252). The current flows upward through a honeycomb well surrounded by an electrically insulating wall 262 and is then collected by a separation return electrode 264 outside the pixel. The current through the electrolyte primarily generates a potential gradient (shading) in front of the "on" pixels, allowing for much better localized stimulation of the retina than in the absence of vertical walls. Note that the shading above pixel 254 in Figure 2B is lighter compared to pixel 104 in Figure 1.

[0012] More precisely, the second embodiment of the present invention is a device for the electrical stimulation of nerve cells, including an array of cavities configured to allow the movement of nerve cells within the cavities. Each cavity has a floor (e.g., 212 in FIG. 2B) and electrically insulating walls (e.g., 262 in FIG. 2B). Each cavity has a first electrode (e.g., 214 in FIG. 2B) disposed on its floor. The device includes a common return electrode (e.g., 264 in FIG. 2B) disposed outside the array of cavities. During operation of the device, an ionic current flows through the contents of the cavity. The electrically insulating walls improve the stimulation efficiency and reduce crosstalk between adjacent cavities of the device by moving the ionic current vertically within the cavity.

[0013] The preferred configuration of the embodiment of FIGS. 2A - 2B is as follows. The depth of the cavity is preferably between 10 μm and 100 μm. The width of the cavity is preferably between 5 μm and 100 μm. The depth of the cavity is preferably greater than the width of the cavity. The array of cavities is preferably periodic, and in this case, the cavity is preferably hexagonal. In a preferred embodiment, the nerve cells are retinal cells. Preferably, the charge density injected into the first electrode of the cavity is between 0.1 mC / cm 2 and 10 mC / cm 2 during operation.

[0014] In the foregoing description, it was convenient to refer to each of the cavities as a cavity having walls. However, in most cases, in an array structure (e.g., as in FIG. 4A), it is clear that the walls are shared between adjacent cavities. In the case of a shared wall, portions of the wall are arbitrarily assigned to each of the two adjacent cavities. In practicing the present invention, it is not affected in any way by how this theoretical division is made. Similarly, local return electrodes (e.g., 216 in FIG. 2A) are, in most cases, similarly shared between adjacent cavities. In the case of a shared return electrode, portions of the return electrode are arbitrarily assigned to each of the two adjacent cavities. In practicing the present invention, it is not affected in any way by how this theoretical division is made.

[0015] (B) Details of Experimental Examples (B1) Introduction The electronic approach to vision restoration has advanced rapidly, and there are systems that have been approved for clinical use in patients blinded by hereditary retinal degenerations (retinitis pigmentosa, RP), as well as systems that are in clinical trials. Patients implanted with the retinal prosthesis Argus (trademark) II (Second Sight Medical Products Inc., Sylmar, California, USA) or the subretinal Alpha IMS / AMS (trademark) (Retina Implant AG, Reutlingen, Germany) due to RP showed improved performance in walking and visual exploration, and the best-reported visual acuities were 20 / 1260 and 20 / 546, respectively. Although favorable results, these beneficial effects are not sufficient to assist in age-related macular degeneration (AMD), the most common form of retinal degeneration, where patients typically retain peripheral vision with a visual acuity of 20 / 400 or better but lose high-resolution central vision.

[0016] The US statutory limit for blindness, a visual acuity of 20 / 200, corresponds geometrically to a pixel pitch of approximately 50 μm. To ensure the safety of charge injection across the electrode-electrolyte interface, there are limitations on the minimum electrode size. Furthermore, crosstalk between adjacent electrodes increases as the pixel size decreases. The latter problem can be addressed by providing a circumferential return electrode for each pixel, but this technique further reduces the penetration depth of the electric field into the tissue.

[0017] The separation of electrodes and tissue in the subretinal space can be reduced using pillar electrodes. As retinal cells in the inner granular layer (INL) migrate to fill the space of such 3D implants, target neurons move closer to the stimulating electrode, lowering the stimulation threshold. However, such pillar electrodes only reduced the stimulation threshold by half, and no significant reduction was achieved with pixel sizes smaller than 55 μm. The underlying problem limiting electrode size lies in the shape of the electric field, which spreads from a small electrode and returns to another electrode beneath the target cells.

[0018] Here, we present a novel 3D geometry for subretinal prostheses to overcome these limitations and enable the reduction of pixels to the size of cells. We call this a honeycomb configuration. In this approach, the return electrode is lifted by vertical insulating walls surrounding each pixel, aligning the electric field vertically and matching the direction of the bipolar cells in the retina, thereby lowering the stimulation threshold. We compare the lifted return electrode 306 in Figure 3-1(B) with the planar configuration in Figure 3-1(A), where 302 is the stimulating electrode and 304 is the substrate on which the stimulating electrode 302 is placed.

[0019] Figure 3-2(C) shows that planar pixels with circumferential return generate a locally confined electric field with short vertical extension. Cells within the electric field are polarized by the potential difference over their entire length. Bipolar somatic cells and axon terminals are located in the INL (Inner Granular Layer) and IPL (Inner Reticuloplasmic Layer), respectively. Therefore, the potential is expressed as the potential relative to the potential midway through the IPL. Figure 3-2(D) shows that the return electrode at the top of the insulating wall generates a vertical dipole confined to the local pixel volume, thereby maximizing the vertical potential drop across the entire target cell layer. The magnitude of the current (length of the arrow) is shown on a logarithmic scale. The potential difference with respect to the center of the IPL (57 μm) is shown in grayscale for a current of 68 nA.

[0020] These barriers decouple the depth of the electric field extension from the pixel width, allowing the pixel size to be reduced to the size of a cell. First, we investigate the anatomical integration of such 3D structures with the retina using implants of 20, 30, and 40 μm pixels, and then quantify the electrical stimulation capacity using a network-mediated retinal stimulation model validated by experimental measurements. The results of this experiment show that this technique opens the door to artificial vision with visual acuity better than 20 / 100, which is highly beneficial not only for patients who are completely blind due to RP, but also for restoring central vision in a much larger number of patients with AMD.

[0021] We previously described the migration of retinal cells to the opening of a membrane implanted in the subretinal space and its effect on the subretinal prosthesis. However, in the configuration of that membrane, there was an opening at the bottom of the implant that allowed current to propagate beneath the implant. However, in reality, artificial retinal implants are usually solid. That is, current does not pass through them, and therefore, as shown here in Figures 3-2(C) and 3-2(D), it can only flow upward. Therefore, we here describe various shapes and materials of return electrodes suitable for optimal shaping of the electric field using such a stimulation array.

[0022] (B2) Results (B2a) Anatomical integration Figures 4A to 4C show subretinal honeycomb implants. Figure 4A shows an image of a 1 mm wide device with a honeycomb structure 25 μm deep with 40 μm (*), 30 μm (**), and 20 μm (***) pixel pitches. The fourth quadrant contained a 10 μm pitch structure, but this was not fabricated because it exceeded the processing limit. This is called the "planar" region because it lacks walls. Figure 4B is a high-magnification image of the 30 μm pitch honeycomb. Figure 4C is an OCT image of the subretinal implant in an RCS rat 6 weeks post-surgery. The scale bar is 200 μm in Figure 4A, 50 μm in Figure 4B, and 100 μm in Figure 4C.

[0023] To evaluate the integration of the honeycomb with the degenerated retina, 1 mm diameter silicon arrays were implanted in the subretinal space of RCS rats (P180-300, n=6) for 6 weeks. Each array was divided into quadrants containing honeycomb and planar regions with 40 μm, 30 μm, and 20 μm pitches (Figure 4A). The arrays were fabricated from silicon using the Bosch etching process to define a honeycomb chamber to a depth of 25 μm (Figure 4B). Integration of the implant with the retina was monitored in vivo by optical coherence tomography (OCT) (Figure 4C). Six weeks after implantation, INL was barely detectable on OCT above the honeycomb region (Figure 4C, right), but was visible above the planar quadrant (Figure 4C, left) and outside the implant, indicating that INL was migrating into the cavity.

[0024] As shown in Figure 5, histological examination confirmed the migration of INL cells, and there were no visible signs of fibrosis or trauma. The retinal structure remained preserved, with the INL, internal plexiform layer (IPL), and ganglion cell layer (GCL) clearly present. Some INL cells remained on the honeycomb wall, but the cavity was completely filled with densely packed cells up to the base of the array. The black arrows point to the location of the original wall. The wall was removed after the sample was embedded and refilled with epoxy for sectioning. The scale bar in Figure 5 is 40 μm.

[0025] A comprehensive assessment of retinal integration and immune response was performed using 3D confocal imaging of the entire retina with implants. 3D reconstruction reveals that most of the observed honeycomb cavities are densely packed with INL(DAPI). A lateral view of a single honeycomb row shows complete transition to the base of the array.

[0026] In a degenerate RCS-controlled retina, the dilating microglia process within the IPL exhibits a dormant state of microglia, while microglia beneath the INL expand the process via the degenerate outer plexiform layer (OPL). In subretinal implants, microglia in the IPL are observed similarly to those in the control retina, and the dilating process exhibits a dormant state of microglia. In planar implants, microglia are located near the device surface. The presence of cortical responses with these active implants indicates that microglia on subretinal prostheses do not interfere with electrical stimulation. With honeycomb implants, the microglia process spreads primarily along the upper wall, with minimal spread into the wells.

[0027] The degree of retinal integration was assessed by analyzing cell density as a function of height from the base of each cavity size. A mean of 50%, 45%, and 54% of INL cells were found in cavities with honeycomb pitches of 40, 30, and 20 μm, respectively. Since the electric field can spread over the walls (Figure 6A), more cells can be stimulated.

[0028] (B2b) Modeling of retinal responses in vivo Figure 6A shows 0.5 A / cm². 2Figure 6B shows the calculated potentials of a planar array (top) and a honeycomb array (bottom) relative to the center of the IPL (z=57μm) for the active electrode current density. Figure 6B shows the experimental thresholds (data points) and calculated thresholds for the planar device (dashed line) and honeycomb device (dotted line) with respect to the current density on the active electrode. The planar model (both binary and linear, as described below) reproduces the trends observed in experimental measurements. The honeycomb array significantly reduces the stimulation threshold (dotted line), enabling safe operation of devices with pixels smaller than 40μm. Maximum charge injection by SIROF (3mC / cm² in a 10ms pulse) 2 The ) is indicated by a dashed line. Here, SIROF is an abbreviation for Spattairidium oxide film. Other biocompatible high-capacity materials such as IrOx deposited by various means (electroplating, chemical vapor deposition, etc.), porous Pt, PEDOT, and carbon nanotubes can also be used for high-capacity electrodes.

[0029] To evaluate the advantages of honeycomb-shaped arrays, we used a model of retinal stimulation via a network. To validate this model, we first compared the modeling results with in vivo stimulation thresholds measured in rats using planar subretinal photovoltaic implants of various pixel sizes, and then calculated the stimulation thresholds for honeycomb arrays of various sizes.

[0030] A complete modeling of this system requires converting the simulated electric field generated by the device (Figure 6A) into the response of intraretinal neurons, applying the subsequent processing through the network to retinal ganglion cell (RGC) activity, and finally converting the retinal output into visually evoked cortical potentials (VEPs), which is a very complex modeling task with several unknowns. This task can be simplified by the assumptions that (1) stimulation through the network induces RGC activity that follows a sigmoid curve, and (2) the cortical response is driven by the sum of retinal signals and follows a sigmoid curve. The inventors approximated the sigmoid dependence of the retinal response through the network to the electric field from previous experiments using two extreme modeling methods: (1) a step function that models a binary transition across the stimulus threshold, and (2) a linear function that models a gradual increase in neuronal output in response to increasing stimulation. As a result, the total retinal response is calculated by integrating the cellular response of the entire volume of INL with either (1) a binary coefficient (i.e., calculating only the portion of the INL volume that exceeds the stimulation threshold) or (2) a cellular response proportional to its polarization. Thus, the two models are called "binary" and "linear," respectively.

[0031] The retinal electric field was calculated using a finite element model of the entire array in COMSOL Multiphysics™ 5.0, assuming a steady-state current, and Maxwell's potential equations were solved using the electrostatic module. The calculated electric field was then converted to the retinal response using both binary and linear models. Each model had only one fitting parameter for relating the retinal output to the amplitude of the cortical response. In the binary model, it was the percentage of INL cells that needed to be activated to elicit the cortical response, while in the linear model, it was the slope of the linear fitting. Irradiance was converted to current density based on the pixel shape and the light-to-current conversion efficiency of the two-diode pixels. Thus, to fit the binary model, the electric field and irradiance were calculated, and the percentage of INL cells exceeding the stimulation threshold (4.8 mV for a 10 ms anode pulse, see section "Methods") was calculated for all pixel sizes. Using previously recorded experimental thresholds for 140 μm, 70 μm, 55 μm, and 40 μm pixels, it can be seen that 8.27 ± 1.42% of the INL volume must exceed the stimulus threshold to elicit a VEP response. Both the binary and linear models produce very similar scaling of the stimulus threshold and pixel size, which indicates that the crucial factor is the shape of the electric field rather than the details of the sigmoid response curve.

[0032] (B2c) Stimulus threshold using honeycomb array The stimulation thresholds of the honeycomb-type arrays were calculated using binary model parameters verified by comparison with in-vivo stimulation thresholds obtained with planar implants (see details in the "Methods" section). These thresholds are significantly lower than those for planar pixels of the same size with respect to the current density on the active electrodes for 10-ms pulses and do not increase much with decreasing pixel size (Figure 6B). Planar arrays with circumferential return have problems with a sharp decrease in potential along the vertical axis due to (a) the radial spread of the electric field from the active electrodes and (b) the return electrodes on the same plane (Figure 3-2(C)). As a result, planar pixels less than 40 μm require current densities greater than the safe charge injection limit of SIROF for 10-ms pulses (>30 mA / cm 2 or a charge density >3 mC / cm 2 ). Placing the return electrodes on the insulating honeycomb walls surrounding the pixels causes the current to flow mainly upward from the active electrodes (Figure 3-2(D)), thereby significantly increasing the depth at which the potential exceeds the stimulation threshold. Furthermore, in the honeycomb-type arrays, the depth of penetration of the electric field into the tissue is set by the height of the walls and is thus decoupled from the pixel width. Therefore, the stimulation threshold with respect to the current density is independent of the pixel width (Figure 6B), and the pixel size can be reduced to a size limited only by retinal movement, i.e., only by cell dimensions. The improvement of the stimulation threshold is summarized in Table 1.

[0033] [Table 1] Stimulation threshold current density calculated for various pixel sizes TIFF0007843610000001.tif61160 Asterisks (*) indicate current densities for 10-ms pulses that exceed the SIROF charge injection limit.

[0034] In addition to the stimulation threshold, honeycomb electrodes significantly improve the spatial selectivity of electrical stimulation (i.e., contrast between adjacent pixels), which is essential for high visual acuity. To replicate the grid patterns used in in vivo assessments of visual acuity, alternating rows of ON and OFF pixels are used to simulate the electric field distribution from the array. In both electrode configurations, increasing the current density increases the positive potential above ON pixels and the negative potential above OFF pixels. The insulating walls of the honeycomb shape prevent lateral spreading of the electric field, thereby widening the dynamic range of selective activation of ON pixels. Furthermore, the electric field spreads over the honeycomb-shaped walls, stimulating cells up to 40 μm from the base of the cavity within safe charge injection limits, enabling activation of up to 99% of the entire retina without crosstalk.

[0035] (B3) Consideration Our research offers a path to improving the spatial resolution of artificial retinas far beyond the limits of current systems. Until recently, the best visual acuity achieved in clinical trials for artificial retinas was 20 / 546, observed in two patients using Alpha IMS / AMS subretinal implants [2,20]. For 70 μm pixels, this artificial vision device performs twice as poorly against the sampling limit, leading some to conclude that visual acuity will not improve by continuously reducing the pixel size. However, the bottleneck in these devices may be electrical rather than biological. In the unipolar configuration of this implant, where the active electrode of each pixel shares a common detachment return electrode, strong crosstalk occurs between adjacent electrodes, significantly reducing spatial contrast. An alternative design with an active electrode and a return electrode for each pixel improves spatial resolution associated with the localization of the electric field. In fact, a recent clinical trial using a 100 μm pixel photovoltaic subretinal implant with a localized return electrode demonstrated a visual acuity of up to 20 / 460, which is only 15% below the sampling limit for this pixel size (20 / 400). Furthermore, in rats, similar implants with 70 μm and 55 μm pixels provided grid visual acuity consistent with the sampling limit. This suggests that the sampling density limit of the stimulus array can be reached when visual acuity adequately limits the electric field of each pixel. However, further reductions in pixel size are limited by a rapid increase in the stimulus threshold, which rises close to the safety limit with 40 μm wide planar pixels (Figure 6B).

[0036] The fundamentally different shape of the stimulation array (honeycomb configuration) described in the embodiment of the present invention addresses this fundamental limitation. The current from the active electrode at the bottom of the cavity flows upward due to the confinement of the pixels by the insulating wall (Figures 3-1(B) and 3-2(D)). This change in the shape of the electric field dramatically lowers the stimulation threshold. Furthermore, the unidirectional flow of current fundamentally alters the scaling of the stimulation threshold depending on the pixel size. In this case, the required current does not change with the size of the electrode, so the current density increases inversely with the area of ​​the electrode, i.e., proportional to the square of the radius. If the ratio of electrode size to pixel width is kept constant, smaller pixels require higher current density, but this is limited by material properties. However, the situation is different with one-dimensional current flow. That is, since the current density required for a particular potential drop does not change with the pixel width, pixels can be made much smaller while maintaining the same current density on the electrode. This means that, in the case of photovoltaic pixels, the threshold irradiance should remain approximately the same for all pixel sizes, as long as the relative dimensions of the electrodes to the pixel width do not change.

[0037] The honeycomb design is uniquely suited to subretinal placement due to the ability of intraretinal neurons to migrate into the spaces within the subretinal environment. Our research has shown that medial retinal neurons readily migrate into small wells with a width of 18 μm (pixel pitch of 20 μm). Tissue viability after 6 weeks indicates that the diffusion of oxygen and nutrients from the retinal vascular system above the implant is sufficient for cell survival within a 25 μm high wall. Since no lower limit of integration was observed in our research, the pixel width may continue to shrink, but it is certain that it does not fall below a cell size of approximately 10 μm. Further experiments are needed to determine the exact minimum within this range. Even without further reduction in pixel size, an array of 20 μm pixels allows for the achievement of spatial resolution that matches the natural visual acuity of rats, which should be 20 / 100 better than human visual acuity.

[0038] Even though the majority of the INL migrates to the cavity, the rest of the retinal structure appears unaffected, with the INL, IPL, and GCL clearly visible. Since all INL cell connections to RGCs are located above the honeycomb wall of the IPL, retinal signal processing is expected to remain unaffected by this migration. Lateral connectivity between bipolar cells via horizontal cells connected to the photoreceptor terminals of the outer plexiform layer (OPL) may be lost in degenerated retina due to the absence of photoreceptors. The number of microglia in the honeycomb wall was found to be similar to that of the planar device, and since the latter induces VEP throughout the animal's life, the immune response with both implants appears to be within acceptable limits. Interestingly, using the honeycomb shape localizes microglia to the top of the wall, allowing neurons to migrate closer to the stimulating electrode.

[0039] Existing subretinal artificial vision devices with separated return electrodes can also incorporate vertical walls around the pixels. In such a unipolar configuration, the polarization of bipolar cells moving into the well should be more efficient due to the vertical alignment of the electric field, and therefore the stimulus threshold and crosstalk between adjacent pixels should be reduced, as shown in Figure 2B. However, in the absence of local return electrodes, the potential in dark pixels is affected by the light entering adjacent pixels. Thus, the dynamic range of the patterns displayed in such an array will be lower than in the case of the bipolar electrode configuration shown in Figure 2A.

[0040] In a steady state, the current density at the electrode surface is proportional to the capacitance per unit area. Therefore, if the electrode is composed of two materials with very different capacitances, the current will mainly flow through the material with the higher capacitance. Because of this phenomenon, it is possible to fabricate vertical walls for 3D electrodes from conductive materials only if the capacitance is much lower than that of the material deposited on top of the wall for the return electrode. Although the wall is made of a conductive material, the current flowing through the electrolyte is very small and the capacitance per unit area is very small, so in the liquid, the wall behaves as if it were an insulator. For example, the wall has a capacitance of 0.01 mF / cm² in physiological saline. 2It can be electroplated with gold having a capacitance of the order of , but the return electrode at the top of the wall has a capacitance of 1-10 mF / cm 2 It is made from iridium oxide with a capacitance of 1 / 2. Since the metal wall is conductive, IrOx can be electroplated onto the top of such a wall. Alternatively, IrOx can be deposited by sputtering, in which case the conductive wall connects this coating to the electrical circuits on the device surface. The wall can also be made from other metals such as platinum, aluminum, or molybdenum.

[0041] To prevent current from flowing between the electrolyte and the sidewall, the sidewall can also be coated with a non-conductive material. For example, the surface becomes non-conductive through oxidation of aluminum or molybdenum. The wall can also be insulated by additional processes, including, but not limited to, atomic layer deposition or photolithography of non-conductive materials. Alternatively, the wall can be fabricated from an insulator. In this case, the return electrode deposited on such a wall would be connected to the electrical circuit via a conductive track deposited on the wall for this purpose.

[0042] The pixels within the implant may be photovoltaic, meaning they can use a photodiode connected between an active electrode and a return electrode to convert light striking the pixel into an electric current. Alternatively, the current can be supplied to the electrodes via a wired connection.

[0043] To enable the movement of cells into the honeycomb-shaped cavities, their width needs to exceed the cell size, i.e., have a width greater than 5 μm. If the width of the cavity significantly exceeds the depth, the beneficial effects of the cavity become negligible. As shown in Figure 6B, for retinal stimulation, pixels larger than 100 μm become negligible. For use in subretinal implantation, i.e., to enable convenient tiling of the rigid array to follow the curved surface of the human eye, the size of the array must be in the range of 0.5 - 5 mm, more commonly 1 - 3 mm. At these dimensions, the array does not necessarily need to use a flexible material as described above, but for larger arrays, a flexible substrate can be incorporated to conform to the eye. The depth of the cavity must enable the movement of cells from the inner granular layer. That is, it is in the range of the thickness of the inner granular layer and the subretinal fragmentation layer, i.e., 20 - 70 μm. From the tissue survival rate after 6 weeks of implantation in vivo, it can be seen that perforations at the bottom of the wells are not necessary for the flow of additional nutrients and tissue survival.

[0044] The side walls can be designed to enable explantation of the implant while at the same time improving the mechanical stability within the subretinal cavity. As shown in this study, a completely smooth vertical wall does not exert large mechanical forces on the tissue when the device is removed from the tissue. However, the tissue moving into the cavity serves as a means to fix the device laterally with respect to the retina. An additional overhang at the top of the wall can be introduced by electroplating over a trench that guides the electroplating procedure. In this case, since the cavity opening do < dc is smaller than its width, the stability of the implant along the z-axis can be further improved. However, since this may make the explantation of the device more traumatic, the correct configuration can be determined based on the patient's age and the likelihood of explantation.

[0045] As shown in Table 1, the stimulation threshold is in the range of 0.01 - 1 A / cm 2 and the pulse width is 10 ms. Therefore, the electrode material is 0.1 - 10 mC / cm 2It is necessary to select a range that allows for the injection of charge densities within that range.

[0046] When the implant is tilted relative to the incident light, the honeycomb-shaped reflective sidewalls help direct the radiation to the photosensitive area of ​​the implant at the bottom of the well. For this purpose, creating the walls with metal using electroplating is advantageous due to the high reflectivity of metal to visible and infrared light. Other highly reflective materials and coatings can also be used for this purpose, especially for light incident on the walls at roughly a graze angle.

[0047] (B4) Method (B4a) Passive honeycomb implant Passive honeycomb implants were fabricated from crystalline silicon wafers using two mask layers to generate patterns for deep silicon etching. Wafers primed with hexamethyldisilazane (HMDS) were spin-coated with 2 μm of negative photoresist (AZ5214-IR) and processed to define the honeycomb walls. This resist was further treated with UV for 15 minutes to enhance selectivity during subsequent etching. A 25 μm deep cavity was formed in the exposed silicon region using a Bosch etching process. After removing the honeycomb defining resist, a 14 μm thick photoresist (7.5% SPR220-7, 68% MEK, and 24.5% PGMEA) was spray-coated onto the wafer and processed to define release trenches around an array approximately 1 mm wide. A second Bosch process was applied to create these release trenches, after which the photoresist was removed. A 60 μm thick protective photoresist was spray-coated onto the wafer, followed by back grinding (Grinding and Dicing Services, Inc., San Jose, California, USA) from the honeycomb base to a thickness of 500–50 μm. Then, excess silicon remaining in the XeF2 gas was etched to complete the implant release. The resulting structure is shown in Figures 4A and 4B. The cavities are arranged in a hexagonal honeycomb pattern with pitches of 40 μm, 30 μm, and 20 μm, with walls 4 μm, 3 μm, and 2 μm thick and 25 μm high, respectively, on a 10 μm thick base. The fourth quadrant was designed for a 10 μm pitch honeycomb, but these were not fabricated as they exceeded the processing limits of the lithography system available to the applicant. This region is referred to herein as the “planar” quadrant. To prevent in vivo dissolution, a 50 nm thick oxide layer was grown on the surface of the silicon implant.

[0048] (B4b) Animals and surgical procedures All experimental procedures were carried out in accordance with the facility guidelines and the ARVO statement on the use of animals in ophthalmic and visual research. Animal care and subsequent transplantation were performed using rats with retinal degeneration obtained from a Royal College of Surgeons (RCS) colony maintained at the Stanford Animal Facility. Honeycomb arrays were implanted in N=6 animals, with implantation performed between P180 and P300 to ensure complete degeneration of photoreceptors. Animals were anesthetized by intramuscular injection of a mixture of ketamine (75 mg / kg) and xylazine (5 mg / kg). A 1.5 mm incision was made through the sclera and choroid 1.5 mm posterior to the limbus. The retina was lifted by injection of saline solution, and the implant was inserted into the subretinal space. The conjunctiva was sutured with nylon 10-0, and a topical antibiotic (bacitracin / polymyxin B) was applied to the eye postoperatively. Surgical success and retinal reattachment were verified using optical coherence tomography (OCT) (HRA2-Spectralis; Heidelberg Engineering, Heidelberg, Germany). Animals were euthanized 6 weeks after transplantation. Additional animals in the control group had planar active implants and were sacrificed after a long in vivo experiment of approximately 6 months.

[0049] (B4c) Whole-mount retinal imaging The animals were euthanized by intracardiac injection of Beuthanesia, the eyes were enucleated, and rinsed with phosphate-buffered saline (PBS, Gibco; Thermo Fisher Scientific, Sunnyvale, California, USA). The anterior segment and lens were removed, the implant position was confirmed under a stereomicroscope, and the eye cup was cut into a 3mm x 3mm square at the center of the implant and fixed in 4% paraformaldehyde (PFA; EMS, Pennsylvania, USA) at 4°C for 12 hours. The implant was retained in place to prevent tissue damage or reorganization upon removal. The samples were permeabilized in 1% Triton X-100 (Sigma-Aldrich, California, USA) in PBS at room temperature for 3 hours. The samples were placed in 10% bovine serum albumin (BSA) blocking buffer and subsequently incubated with two primary antibodies at room temperature for 12 hours. Specifically, this procedure was performed with rabbit anti-IBA1 antibody (1:200; Wako Chemicals, Virginia, USA) and mouse anti-glutamine synthetase (GS, 1:100; Novus Biologicals, Colorado, USA) in PBS containing 0.5% Triton X-100 and 5% BSA. Samples were washed with 0.1% Triton X-100 in PBS (PBS-T) at room temperature for 6 hours and incubated with the two secondary antibodies at room temperature for 12 hours. Specifically, this procedure was performed with donkey anti-rabbit AlexaFluor 488 (1:200; Thermo Fisher Scientific, Sunnyvale, California, USA) and donkey anti-mouse CY3 (1:200; Jackson ImmunoResearch Inc., Pennsylvania, USA), and counterstained with 4',6-diamidine-2-phenylindole (DAPI) in PBS. After washing with PBST for 6 hours, the samples were mounted using the mounting medium Vectashield (H-1000; Vector Laboratories, Burlingham, California, USA).

[0050] 3D imaging was performed using a Zeiss LSM880 confocal inverted microscope with Zeiss ZENBlack software. The image plane was acquired through the entire thickness of the retina using a Z-stack, with the upper and lower limits defined at the internal limiting membrane (ILM) and 10 μm below the base of the honeycomb cavity, respectively. The stack was acquired at the center of each honeycomb quadrant using a 225 μm × 225 μm acquisition area, a 360 nm Z-axis step, and a 40x oil immersion objective lens with a 0.55 μm pinhole.

[0051] (B4d) Image Analysis The confocal dataset was analyzed using the FiJi distribution in ImageJ. To analyze cell density in the wells and on top of the implant, the contrast of individual XY planes was first maximized to ensure 0.3% channel saturation to compensate for brightness variations at various Z-direction positions within the stack. Next, the XY planes were decolorized and the background was subtracted. A Gaussian blur filter (σ=3 pixels, 0.42 μm) was applied to smooth the brightness variations within individual cells. Then, the XY planes were passed through an edge detection filter, and the final image was created from a combination of the processed image and the edge-detected image with the background subtracted. For cell density analysis, the channel threshold was adjusted (default method) to obtain a binary representation of the cells. Next, considering the area occupied by the honeycomb wall, the cell density in the XY plane was calculated as a percentage of the area occupied by cells. To account for local variations in retinal histology, each honeycomb unit was analyzed independently, and the cell density was normalized to the maximum value within the unit stack. The percentage of INL contained within the cavity was calculated as follows: In formula TIFF0007843610000002.tif3395, D(z) is the relative density of cells per XY plane as a function of height (z) from the base (z=0), z' is the wall height (25 μm), and z'' is the end of INL, defined as a point where D(z) < 0.02.

[0052] Six embedded devices, each containing honeycomb structures of 40 μm, 30 μm, and 20 μm sizes, were imaged and analyzed. Three devices had damaged 20 μm honeycomb structures and were therefore excluded from the analysis.

[0053] (B4e) Histological preparation After confocal imaging, the samples were rinsed with buffer and fixed in 1.25% glutaraldehyde solution at room temperature for 24 hours. Next, they were fixed with osmium tetroxide at room temperature for 2 hours and dehydrated stepwise with alcohol and propylene oxide. After overnight immersion in epoxy (without DMP-30) at room temperature (using Electron Microscopy Sciences-Araldite-EMbed, RT13940, Mollenhauer's kit), the samples were left in a 70°C oven for 36 hours. The epoxy block was then trimmed until the silicon implants were exposed. To prevent damage to the cutting knife and formation of silicon fragments from the honeycomb structure, the silicon implants were removed using XeF2 etching (Xactix e-1, 23°C, 3 Torre). The block was then refilled with epoxy and placed in a vacuum desiccator for 2 hours, followed by overnight baking at 70°C. Refilling of the voids remaining after etching of these implants provided structural support during cutting. Sections with a thickness of 700 nm (cut with Reichart UltracutE) were stained with toluidine blue for use in an optical microscope.

[0054] (B4f) Modeling of electric field and retinal stimulation The electric field of the retina was calculated using a 3D finite element model of the complete array in COMSOL Multiphysics 5.0, and Maxwell's potential equations were solved using an electrostatic module, assuming a steady-state current. The modeled array was 1 mm in diameter and 30 μm thick, consisting of hexagonal pixels of various sizes as shown in Table 2, with a return electrode connected to a single mesh.

[0055] [Table 2] Number of pixels and shape of the modeled photovoltaic array TIFF0007843610000003.tif67161

[0056] The electric field is calculated in a volume (cube, side length = 10 mm) with a defined ground (0 potential) around it. The modeled artificial vision device functions as a closed system where all current injected from the active electrode is collected by the return electrode. The boundary conditions of the electrode surface are defined as having a uniform current density corresponding to the steady state.

[0057] The current density at the steady-state electrode-electrolyte interface is proportional to the electrode capacitance per unit area. Therefore, even if the sidewalls are not insulated, the current flows mainly through the SIROF-coated upper surface. For example, when a SIROF coating is used as the return electrode material, its capacitance becomes approximately 1000 times that of gold (10 mF / cm²). 2 vs 0.01 mF / cm 2 The ratio of the area of ​​the return electrode located at the top of the wall to the area of ​​the sidewall per pixel is equal to the ratio of half the width of the wall to its height. For a wall with an aspect ratio (height to width) of 5:1, the ratio of the area of ​​the top to the side of the wall is approximately 10:1. Thus, the capacitance of the sidewall is 1 / 100th of the capacitance of the SIROF return electrode deposited at the top of the honeycomb per pixel. Therefore, even if the wall is made of metal, it functions as if the wall were insulating, as the electrolyte can only accept very low currents. This greatly simplifies the manufacturing of the honeycomb, as simple electroplating is possible without the need to insulate the sidewalls.

[0058] Two electrode configurations were investigated: (1) a planar configuration with a local return electrode (Figure 3-1(A)), and (2) a honeycomb configuration with a local return electrode (Figure 3-1(B)). In both configurations, a common return electrode collects the currents generated by all active pixels such that the sum of the collected currents equals the sum of the currents injected into the individual active electrodes. The sidewalls of the honeycomb configuration are nonconductive. In previous studies using active devices, in vivo spatial resolution was evaluated by projecting grid patterns of various spatial frequencies with 100% contrast. Maximum resolution corresponds to activating alternating rows of pixels (creating ON rows and OFF rows). To replicate this configuration in the simulation, the electric field distribution was calculated using the same activation scheme, and the electric field was analyzed at the center of the array where crosstalk between adjacent pixels was highest. To correlate electrical simulations with light intensity using our photovoltaic prosthesis, the total current per pixel was calculated based on the diode area of ​​a two-diode configuration and the measured light-to-current conversion efficiency (0.40, 0.31, 0.26, and 0.24 A / W for 140 μm, 70 μm, 55 μm, and 40 μm pixels, respectively).

[0059] The retinal stimulation threshold was assessed using a network-mediated activation model. This approach assumed a network-mediated stimulation threshold defined by the voltage drop between bipolar cells. In an external electric field, the intracellular medium becomes equipotential within microseconds, resulting in hyperpolarization and depolarization at the proximal and distal cell membranes of the anode, respectively. Using the stimulation threshold current density through the retinal network of large electrodes with 10 ms pulses obtained from the literature, the mean resistivity of the retina (1000 Ω·cm), and the mean length of bipolar cells estimated as the distance from the center of the INL to the center of the IPL (37 μm), a potential difference threshold of 4.8 mV from somatic cells to axon terminals for anode stimulation was calculated. The cathode threshold of -21 mV was calculated based on network-mediated activation curves (scaled to match the calculated anode threshold) measured in rat retina with anode and cathode stimulation.

[0060] Importantly, while our model is based on stimulation current density derived from the literature, it should be noted that the calculated transcellular voltage is linearly proportional to retinal resistivity. There is no consensus on this in the prior literature. Transcellular voltage is used only as a means to evaluate the boundary of the activation zone within the tissue relative to the stimulation threshold. Assumed variability in retinal resistivity linearly affects the potential of the same current, including the threshold potential. Therefore, these variability do not affect the boundary of the activation zone because they are calculated in relation to the stimulation threshold.

[0061] The inventors approximated the sigmoid dependence of the network-mediated retinal response to an increasing electric field using two extreme modeling approaches based on previous experiments: (1) a step function that models a binary transition across the stimulus threshold, and (2) a linear function that models the gradual increase in neural output in response to increasing stimulation. As a result, the total retinal response is calculated by integrating the cellular response across the entire volume of the INL with either (1) a binary coefficient (i.e., calculating only a portion of the INL volume above the stimulus threshold), or (2) a cellular response proportional to its polarization.

Claims

1. A device for electrical stimulation of nerve cells, It comprises an array of cavities configured to allow the migration of nerve cells within the cavity, Each of the cavities is a space defined by a floor and a conductive wall perpendicular to the floor, the horizontal cross-sectional shape of the space is a regular polygon that can be densely arranged in a plane, and the array of cavities is created in a form in which a plurality of hollow regular polygonal prisms forming the walls of each of the cavities are arranged periodically and densely. Each of the aforementioned walls is made entirely of a conductive material. Each of the cavities has a first electrode positioned in the floor portion, Each of the cavities has a second electrode positioned on the upper surface of the wall corresponding to each cavity and spaced vertically apart from the corresponding first electrode in the floor. During the operation of the device, an ionic current flows through the contents of the cavity. The apparatus is characterized in that the capacitance of the second electrode is greater than the capacitance of the conductive wall within each cavity, and as a result, the second electrode preferentially collects ion current from the side surface of the conductive wall.

2. The apparatus according to claim 1, The apparatus is characterized in that the depth of the cavity is between 10 μm and 100 μm.

3. The apparatus according to claim 1, The apparatus is characterized in that the width of the cavity is between 5 μm and 100 μm.

4. The apparatus according to claim 1, The apparatus is characterized in that the depth of the cavity is greater than the width of the cavity.

5. The apparatus according to claim 1, The apparatus is characterized by being configured to convert light striking the apparatus into an electric current using a photodiode connected between the first electrode and the second electrode.

6. The apparatus according to claim 1, The apparatus is characterized in that the horizontal cross-section of the cavity is hexagonal.

7. The apparatus according to claim 1, The apparatus is characterized in that the nerve cells are retinal cells.

8. The apparatus according to claim 1, During operation of the apparatus, the charge density injected into the first electrode of the cavity is 0.1 mC / cm². ２ ~10 mC / cm ２ A device characterized by being such.

9. The apparatus according to claim 1, The apparatus is characterized in that the capacitance per unit area of ​​the second electrode is at least 100 times greater than the capacitance per unit area of ​​the conductive wall.

Images