Cited by 44Neuroscience and Neural EngineeringTactile and Sensory InteractionsPain Management and Treatment

The DOI matches its Crossref record; 223 of the 223 references it cites were found in a registry or an index; 8 of the 185 sources its claims cite were checked for support.

Advances in visual prostheses: engineering and biological challenges

Open access
Borda, Eleonora, Ghezzi, Diego·2022·Progress in Biomedical Engineering

Abstract

Vision is an extraordinary sense through which we can appreciate the beauty of the world we live in, gain invaluable knowledge and communicate with others using visual expression and arts. On the contrary, blindness is a threatening medical condition disrupting the life of affected people and their families. Therefore, restoring sight is one of the open challenges of our society. Today, the synergistic convergence of science and technology holds the potential to provide blind patients with artificial vision using visual prostheses: a type of implantable medical device able to reactivate visual neurons using electrical stimulation. Although clinical trials showed that vision restoration is still far away, significant technological advances make visual prostheses a valuable solution for blind patients. This review is not only a description of the state-of-the-art. Instead, it provides the reader with an update on recent developments, a critical discussion of the open challenges, and an overview of promising future directions.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

This review synthesizes engineering and biological challenges of visual prostheses across the retina, optic nerve, LGN, and V1, noting that retinal implants are the most clinically advanced but still provide vision substitution rather than full restoration. It highlights quantitative limits in visual field and resolution, electrode–tissue interface issues, image encoding, wireless power, biological variability, and market/regulatory lessons from first-generation devices.

Key findings

  • Visual prostheses are classified by target along the visual pathway: retina, optic nerve, LGN, and V1. Only retinal prostheses have reached clinical approval, and only for retinitis pigmentosa (RP): Argus® II, IRIS® II, and Alpha IMS/AMS; none remain on the market.
  • Best reported visual acuities were 20/460 with the subretinal PRIMA implant in AMD patients and 20/546 with the subretinal Alpha-AMS in RP patients. Grating acuities ranged from 20/1260 with the epiretinal Argus® II to 20/364 with the subretinal Alpha-AMS.
  • A visual angle of 30°–35° was identified as the minimal requirement for mobility and daily tasks under simulated prosthetic vision. POLYRETINA is described as a high-density wide-field epiretinal prosthesis with 10,498 photovoltaic pixels over approximately 43° (about 140 mm²).
  • Temporal limits include phosphene fading; flicker fusion (30 Hz or above) was achieved only for the PRIMA device in AMD patients. Market challenges include RP affecting about 1.5–2 million people, but only 12% having visual acuity worse than counting fingers and 0.5% having no light perception; Argus® II cost 115,000 EUR in Europe to 150,000 USD in the USA, with one patient estimating total device, surgery, and rehabilitation costs at 497,000 USD.

Methods

  • Narrative review; the paper states that no new data were created or analysed.
  • Synthesizes literature on the visual system, device categories by anatomical target, clinical and preclinical trials, and quantitative, technological, biological, and market/regulatory challenges.
  • No systematic search strategy, inclusion criteria, or quantitative meta-analysis is described.

Limitations

  • The paper does not include a formal limitations section; as a narrative review, it lacks systematic search/selection criteria and quantitative synthesis (evident from the text).
  • The authors state key knowledge gaps: insufficient understanding of the neural visual code in blind individuals, unclear predictiveness of animal models, and variable retinal remodelling and plasticity across patients.
  • Evident from the text: many cited outcomes come from heterogeneous, device-specific clinical and preclinical reports, and first-generation approved retinal prostheses were discontinued, limiting long-term generalizability and patient support.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1088/2516-1091/ac812c

Cites

223 of 223 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

112 Not checked yet · 34 Own interpretation · 27 No basis given · 4 Partly supported · 1 Supported · 1 Supported so far · accuracy not yet measured

Verify all 177 references — free account
  • Probably not a substantial claim. 25
  • It may state two separate claims as one. 52
70 of 70
  • The authors state this as an assumption.

    Neuroplasticity triggered by learning and rehabilitation will improve artificial vision over time.

    “Initially, phosphenes might be irregular and difficult to interpret, but it is assumed that neuroplasticity triggered by learning and rehabilitation will improve artificial vision over time (figure 3(b)) [20].”
  • The paper relies on this without support.

    The entire visual field is mapped over the anatomical area in all structures mentioned (retina, optic nerve, LGN and V1).

    “The entire visual field is mapped over the anatomical area in all the structures mentioned (retina, optic nerve, LGN and V1).”
  • The paper relies on this without support.

    Stimulation parameters play a crucial role in the characteristics of phosphenes, such as brightness, duration, size or colour.

    “The stimulation parameters play a crucial role in the characteristics of phosphenes, such as brightness, duration, size or colour.”
  • The paper relies on this without support.

    When stimulation resolution is low, phosphenes might fuse into a blob during synchronous activation of multiple electrodes.

    “When the stimulation resolution is low, as it is for the Argus® II or epicortical visual prostheses, it has been shown that phosphenes might fuse into a blob during synchronous activation of multiple electrodes.”
  • The paper relies on this without support.

    Artificial vision can be hampered by poor temporal resolution.

    “Artificial vision can also be hampered by poor temporal resolution.”
  • The paper relies on this without support.

    Little is known about visual targets other than the retina.

    “Meanwhile, little is known about other visual targets.”
  • The paper relies on this without support.

    Decreasing electrode size imposes constraints on the electrode material to maintain low impedance and high charge injection capacity.

    “However, decreasing the electrode size imposes constraints on the electrode material to maintain low impedance and high charge injection capacity.”
  • The paper relies on this without support.

    Exceeding the charge injection limit of electrodes will cause irreversible damage to both electrodes and tissue.

    “Exceeding the charge injection limit of electrodes will cause irreversible damage to both electrodes and tissue.”
  • The paper relies on this without support.

    Matching neural tissue mechanical properties with implanted ones would help improve long-term stability and efficacy of stimulation.

    “For visual prostheses, regardless of whether they are epineural or intraneural, matching the neural tissue mechanical properties with the implanted ones would help in improving the long-term stability and efficacy of the stimulation.”
  • The paper relies on this without support.

    Visual information processing is a crucial step.

    “Visual information processing is a crucial step.”
  • The paper relies on this without support.

    Saliency detection works efficiently for image or static object recognition tasks.

    “This method works efficiently for image or static object recognition tasks.”
  • The paper relies on this without support.

    A drawback of machine learning methods is the high training cost required for good accuracy.

    “A drawback of these methods is the high training cost required for good accuracy.”
  • The paper relies on this without support.

    Phosphene maps are specific to each patient and might change over time.

    “Such a customised approach is needed because phosphenes maps are specific to each patient and might change over time.”
  • The paper relies on this without support.

    Most image processing algorithms and closed-loop approaches have been tested so far in-silico or in-vitro.

    “Most of the image processing algorithms and the closed-loop approaches have been tested so far in-silico or in-vitro.”
  • The paper relies on this without support.

    Neurostimulation has been routinely performed with electrodes wired to an electric pulse generator.

    “Neurostimulation has been routinely performed with electrodes wired to an electric pulse generator.”
  • The paper relies on this without support.

    In general, a prosthesis is validated for its efficacy in animal models.

    “In general, a prosthesis is validated for its efficacy in animal models.”
  • The paper relies on this without support.

    It remains unclear whether experimental animals are appropriate models to predict efficacy in humans.

    “However, it remains unclear whether experimental animals are appropriate models to predict efficacy in humans.”
  • The paper relies on this without support.

    Even when blind animal models are used, it is difficult to establish how well they reflect human diseases.

    “Even when blind animal models are used, it is difficult to establish how well they reflect human diseases.”
  • The paper relies on this without support.

    Remodelling might be extremely variable from patient to patient.

    “In addition, remodelling might be extremely variable from patient to patient.”
  • The paper relies on this without support.

    Variability could also be found in residual oculomotor behaviour, cellular preservation along the visual pathway, and remaining neuroplasticity.

    “Variability could also be found in the residual oculomotor behaviour, cellular preservation along the visual pathway, and remaining neuroplasticity.”
  • The paper relies on this without support.

    The downside of qualitative functional vision tests is limited relevance out of a controlled laboratory environment.

    “The downside of these tests is the limited relevance out of a controlled laboratory environment.”
  • The paper relies on this without support.

    Medical device regulation is a crucial step for any kind of neuroprosthetic device.

    “Medical device regulation is another crucial step for any kind of neuroprosthetic device.”
  • The paper relies on this without support.

    Neurotechnology is a rapidly growing field, with several new companies entering the business.

    “Neurotechnology is a rapidly growing field, with several new companies entering the business.”
23 of 23

Citations

Cites

Show all 222 references

Research works printed but not linked: 5

  1. LeRoy, 1755, Mémoire où l’on rend compte de quelques tentatives que l’on a faites pour guérir plusieurs maladies par l’électricité, Hist. Acad. R. Sci., 60, 87
  2. Foerster, 1929, Beitrage zur pathophysiologie der sehbahn und der spehsphare, J. Psychol. Neurol., 39, 435
  3. JC, 1958, Electronics brings light to the blind, Radio Electron., 29, 53
  4. Button, 1962, Visual responses to cortical stimulation in the blind, J. Iowa Med. Soc., 52, 17
  5. Michael, 2020, Video-mode percepts are smaller than sums of single-electrode phosphenes with the Orion® visual cortical prosthesis, Investigative Ophthalmol. Vis. Sci., 61, 927

Cited by

Sorted by citations

See all 44

Metadata

Authors
Borda, E., & Ghezzi, D.
Affiliations
École Polytechnique Fédérale de Lausanne (CH)
Corresponding author
Ghezzi, Diego
Venue
Progress in Biomedical Engineering
Published
2022-07-01
Volume
4
Issue
3
Pages
032003
ISSN
2516-1091 (ISSN-L)
OpenAlex
W4285495317
Cited by
44
References
222
Citations by year
2026: 1 · 2025: 14 · 2024: 17 · 2023: 11 · 2022: 1
FWCI
4.75
Type
journal article
Language
en
Open access
Locations
  • Progress in Biomedical Engineering (publishedVersion, cc-by)
  • Infoscience (Ecole Polytechnique Fédérale de Lausanne) (submittedVersion, cc-by-nc-nd)
Funders
  • Medtronic
  • École Polytechnique Fédérale de Lausanne
Topics
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neuroscience and Neural Engineering
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Tactile and Sensory Interactions
  • Health Sciences › Medicine › Anesthesiology and Pain Medicine › Pain Management and Treatment
Keywords
retinal prosthesis, blindness, vision restoration, biomedical implants, artificial vision, open problems

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-10-08, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-08 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review