Advances in visual prostheses: engineering and biological challenges

Borda, Eleonora; Ghezzi, Diego

doi:10.1088/2516-1091/ac812c

Cited by 31 publications

(22 citation statements)

References 219 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Naive Stimulation Algorithm (NSA) for the epiretinal implant described in [15] is tested in patients with the Argus II epiretinal implant. Many improvements to the original algorithms that use more advanced image processing methods such as Difference of Gaussian [16], constrained optimization [17], and deep learning [18,19] obtained better results in simulation [20]. The NSA algorithm for epiretinal implant most tested in patients is transforming a digital image into electrical stimulation using a downscaling operation of a camera image that matches the intensity of a pixel to the amplitude or frequency.…”

Section: Naive Stimulation Algorithmmentioning

confidence: 99%

Learning to see via epiretinal implant stimulation in silico with model-based deep reinforcement learning

Lavoie,

Besrour,

Lemaire

et al. 2024

Biomed. Phys. Eng. Express

View full text Add to dashboard Cite

Objective: Diseases such as age-related macular degeneration and retinitis pigmentosa cause the degradation of the photoreceptor layer. One approach to restore vision is to electrically stimulate the surviving retinal ganglion cells with a microelectrode array such as epiretinal implants. Epiretinal implants are known to generate visible anisotropic shapes elongated along the axon fascicles of neighboring retinal ganglion cells. Recent work has demonstrated that to obtain isotropic pixel-like shapes, it is possible to map axon fascicles and avoid stimulating them by inactivating electrodes or lowering stimulation current levels. Avoiding axon fascicule stimulation aims to remove brushstroke-like shapes in favor of a more reduced set of pixel-like shapes. Approach: In this study, we propose the use of isotropic and anisotropic shapes to render intelligible images on the retina of a virtual patient in a reinforcement learning environment named rlretina. The environment formalizes the task as using brushstrokes in a stroke-based rendering task. Main Results: We train a deep reinforcement learning agent that learns to assemble isotropic and anisotropic shapes to form an image. We investigate which error-based or perception-based metrics are adequate to reward the agent. The agent is trained in a model-based data generation fashion using the psychophysically validated axon map model to render images as perceived by different virtual patients. We show that the agent can generate more intelligible images compared to the naive method in different virtual patients. Significance: This work shares a new way to address epiretinal stimulation that constitutes a first step towards improving visual acuity in artificially-restored vision using anisotropic phosphenes.

show abstract

Section: Naive Stimulation Algorithmmentioning

confidence: 99%

Learning to see via epiretinal implant stimulation in silico with model-based deep reinforcement learning

Lavoie,

Besrour,

Lemaire

et al. 2024

Biomed. Phys. Eng. Express

View full text Add to dashboard Cite

show abstract

“…It will even inevitably cause great damage to the tissue during the implantation if occupying too much space, resulting in serious inflammatory response, which is unfavorable to the efficient operation of neural prosthesis. [7,9,15] Taking visual prosthesis as an example, the space of the human eye is limited. In order to improve the image resolution, it is necessary to implant electrode arrays with as many channels as possible, and the corresponding functional modules should be increased accordingly, which will lead to the increasing volume of the implants.…”

Section: Neural Interface Challengesmentioning

confidence: 99%

“…The high‐density artificial retina technology is one of the typical representatives, generating impressive social and economic values. [ 4c,7 ]…”

Section: Introductionmentioning

confidence: 99%

Challenges and Opportunities of Implantable Neural Interfaces: From Material, Electrochemical and Biological Perspectives

Zeng

Huang

2023

Adv Funct Materials

View full text Add to dashboard Cite

The desirable implantable neural interfaces can accurately record bioelectrical signals from neurons and regulate neural activities with high spatial/time resolution, facilitating the understanding of neuronal functions and dynamics. However, the electrochemical performance (impedance, charge storage/injection capacity) is limited with the miniaturization and integration of neural electrodes. The “crosstalk” caused by the uneven distribution of elctric field leads to lower electrical stimulation/recording efficiency. The mismatch between stiff electrodes and soft tissues exacerbates the inflammatory responses, thus weakening the transmission of signals. Though remarkable breakthroughs have been made through the incorporation of optimizing electrode design and functionalized nanomaterials, the chronic stability, and long‐term activity in vivo of the neural electrodes still need further development. In this review, the neural interface challenges mainly on electrochemistry and biology are discussed, followed by summarizing typical electrode optimization technologies and exploring recent advances in the application of nanomaterials, based on traditional metallic materials, emerging 2D materials, conducting polymer hydrogels, etc., for enhancing neural interfaces. The strategies for improving the durability including enhanced adhesion and minimized inflammatory response, are also summarized. The promising directions are finally presented to provide enlightenment for high‐performance neural interfaces in future, which will promote profound progress in neuroscience research.

show abstract

“…In this favorable moment, neurotechnology is receiving great financial attention, but only a few companies will succeed. [ 118,119 ] The road to market is long, unpredictable, and costly, while business cases only sometimes justify the investment. [ 120 ] Neurotechnology primarily targets neurological disorders, overall affecting a considerable portion of society.…”

Section: Figurementioning

confidence: 99%

Engineering Materials for Neurotechnology

Ghezzi

2023

Adv Eng Mater

Self Cite

View full text Add to dashboard Cite

A crucial issue for a neural interface is the chronic foreign body reaction (FBR) upon implantation. [1][2][3] Avoiding the FBR, or at least reducing its severity, is necessary to preserve the long-term functioning of neural interfaces. [4][5][6] Yet, this is still an open research question. The FBR is traditionally described as a passive barrier encapsulating the device, altering the neural interface, and impairing implant-to-cell communication. [2,7,8] Besides acting as a passive insulating barrier, soluble factors released during the inflammatory and fibrotic process can accelerate device degradation and electrode corrosion. [9] The FBR is inevitable whenever any material becomes in contact with body fluids. [10] In contrast, materials engineering and advanced fabrication processes can minimize its severity, for example, by improving the mechanical compliance of the implant, decreasing the device size, and reducing implantation and chronic trauma. [3] One of the most common limitations of neural interfaces is the mechanical mismatch between the stiff neural implant and the soft neural tissue. This mismatch intensifies the neural damage and FBR caused by insertion, compression, and motion. [11] The neural tissue is curved, soft, dynamic, and subject to deformations caused, for example, by blood pulsation, respiratory pressure, and natural body movements. [11,12] By contrast, most neural interfaces are static and struggle to conform to neural tissue in static and dynamic conditions. A large body of research in neural interfaces addresses this issue using flexible, conformable, or stretchable neural interfaces (Figure 1).Polyimide (PI), parylene-C, and SU-8 are widely used materials in flexible neural interfaces. [13][14][15][16][17][18][19][20][21][22] The advantages of these materials are their biochemical and thermal stability, and their compatibility with clean-room microfabrication processes. [23][24][25] However, they exhibit high Young's modulus (GPa range) and limited elastic deformation (<3%). [26] Ultra-thin neural interfaces made of these materials still reach low bending stiffness if sufficiently thin (<15 μm) despite the high Young's modulus. [27] They can conform to a cylindrical body such as a nerve or a smooth brain surface (Figure 1a), thus reducing the FBR severity. [23] Moreover, ultra-thin neural interfaces with a small cross-sectional area also can minimize tissue damage during penetration (Figure 1b). [22,24,28] Conformability to cylindrical or smooth surfaces is sufficient in rodents, but more is needed for large and complex surfaces with gyri, like in primates, where spherical conformability is required. Because of the limited elastic deformation, conformability to convoluted surfaces is difficult to achieve with these materials. Compliance with neural macro-and micro-movements is also reduced. [26] Elastomers exhibit a lower Young's modulus of at least three orders of magnitude (MPa range) and a much broader elastic regime under strain (>20%). Bending stiffness is influenced by both thick...

show abstract

Advances in visual prostheses: engineering and biological challenges

Cited by 31 publications

References 219 publications

Learning to see via epiretinal implant stimulation in silico with model-based deep reinforcement learning

Learning to see via epiretinal implant stimulation in silico with model-based deep reinforcement learning

Challenges and Opportunities of Implantable Neural Interfaces: From Material, Electrochemical and Biological Perspectives

Engineering Materials for Neurotechnology

Contact Info

Product

Resources

About