Photovoltaic retinal prosthesis restores high-resolution responses to single-pixel stimulation in blind retinas

Visual prostheses, including retinal, optic nerve, and cortical implants, offer the potential to restore vision. While retinal prostheses have shown promise in clinical trials with retinitis pigmentosa patients, limitations in visual angle (often around 20°, insufficient for independent mobility) and resolution (due to electrode density and direct nerve fiber activation) have hindered widespread success and user satisfaction. Many patients cease using their implants within a few years, citing limited functionality and cognitive exhaustion. Studies suggest a visual angle of at least 30° is needed for daily tasks, and increased resolution is crucial for object recognition. Wide-field arrays, primarily epiretinal, have been proposed to address the visual angle issue, but achieving both wide field and high resolution remains challenging. This study introduces POLYRETINA, a high-density photovoltaic epiretinal prosthesis designed to overcome these limitations by employing network-mediated stimulation to achieve high resolution while maintaining a large visual field.

Existing retinal prostheses, such as the Argus II, while a significant technological achievement, suffer from limitations in visual angle and resolution, restricting their efficacy in daily life. The Argus II, with its limited visual angle of 20° and coarse resolution (6 x 10 electrodes with a 525-µm pitch), restricts user capabilities. Subretinal implants have demonstrated better visual acuity but face challenges in large-scale implementation. The need for a wider visual field (at least 30°) and higher resolution is well-established in the literature, with previous studies demonstrating the importance of these factors for successful navigation and object identification. The development of wide-field epiretinal prostheses like POLYRETINA aims to address these gaps and improve patient outcomes.

The POLYRETINA prosthesis consists of 10,498 photovoltaic pixels (80-µm diameter, 120-µm pitch) distributed over a 13 mm diameter active area, covering a visual angle of ~43°. Finite element analysis simulations were conducted to assess the mechanical stress on the pixels during shaping. The pixels, coated with titanium nitride (TiN) to enhance stimulation efficiency, were tested for electrical independence (crosstalk) using glass microelectrodes. Optoelectronic characterization involved measuring photocurrent (PC) and photovoltage (PV) at increasing irradiance levels using Kelvin Probe Force Microscopy (KPFM). Ex vivo experiments used explanted retinas from rd10 mice (a retinitis pigmentosa model) to assess the stimulation efficiency of TiN-coated pixels compared to uncoated Ti pixels. Single-electrode extracellular recordings measured retinal ganglion cell (RGC) responses to various light stimuli (large-field and single-pixel illumination, various irradiance levels, and pulse durations). Spatial resolution was evaluated using two-point discrimination tests (pixel switch) and high-contrast grating pattern reversals with variable bar widths. A thermal model assessed the iris temperature increase under various illumination conditions to evaluate safety limits. Statistical analysis methods included normality tests (D’Agostino & Pearson), t-tests, and ANOVAs.

POLYRETINA's TiN-coated pixels showed significantly higher PC and PV than uncoated Ti pixels. Single-pixel illumination reproducibly evoked network-mediated RGC responses at irradiance levels well below the maximum permissible exposure (MPE) for retinal safety. Spatial resolution was demonstrated to be at least as good as the pixel pitch (120 µm) using both two-point discrimination and grating pattern reversal paradigms. The majority of RGCs (24 out of 31) exhibited small receptive fields (RFs), ranging from 34.5 to 142.5 µm in diameter, while a smaller number showed larger RFs (184.3-282.7 µm). The response resolution achieved (at least 120 µm) is significantly better than that of current clinical implants. Thermal modeling indicated the need for irradiance adjustments to prevent excessive iris heating during continuous illumination. However, eye movements are likely to mitigate this concern, suggesting that high pulse rates are achievable using eye-tracking technology.

POLYRETINA's high pixel density and wide field of view address crucial limitations of current retinal prostheses. The achievement of high spatial resolution (at least 120 µm) through network-mediated stimulation is a significant advance. The results suggest that a significant portion of the mid-peripheral retina could be stimulated with a resolution exceeding the physiological receptive field sizes in humans. The use of long light pulses and non-rectangular waveforms minimizes the probability of undesirable direct axonal stimulation. While the study was conducted ex vivo using a mouse model, the results strongly support the potential for high-resolution, wide-field artificial vision in retinitis pigmentosa patients. The findings highlight the advantages of using a photovoltaic approach with conjugated polymers to create a mechanically compliant and high-density implant.

POLYRETINA demonstrates the feasibility of a high-resolution, wide-field retinal prosthesis. The high pixel density (10,498 pixels), large visual angle (~43°), and ability to achieve single-pixel stimulation with high spatial resolution are major advancements. Future studies should focus on in vivo preclinical trials to validate these findings, explore the impact of direct axonal stimulation, and investigate the use of near-infrared light-sensitive polymers to improve safety and reduce potential for photoreceptor activation. Optimizing stimulation parameters and considering the integration of eye-tracking for beam compensation are also important.

The study was conducted ex vivo using a mouse model. While the rd10 mouse is an established model for retinitis pigmentosa, it does not perfectly replicate the human retina. The use of explanted retinas could affect the results compared to in vivo conditions. The thermal model represents a worst-case scenario, and the impact of eye movements on iris heating is an area requiring further investigation. While the resolution achieved is a major advancement, the generalizability of these findings to the human fovea needs further consideration due to the differing RGC receptive field sizes.