BRAIN-MACHINE INTERFACE:
Cortical Vision Prosthesis

John Raiti and Benjamin Kimia

I. Goal:

To provide functional visual perception to a person who has been rendered blind either due to disease or trauma.

Achieve this by bypassing the damaged elements of the visual pathway and interface directly at the cortical level (by means of an electrode array) to generate visual perception

(Maynard, 2001)

A schematic of the visual system highlighting the neural structures of the visual pathway most relevant to the development of a vision prosthesis

Statistics (National Alliance for Eye and Vision Research):

over 1 million legally blind people in the U.S. (legally blind refers to clinically measured visual acuity of 20/200 in the better eye with best correction, or visual field of 20 degrees or less)

Disease	Population Blinded by disease in the U.S.	Treatment
Retinitis Pigmentosa	100,000	Subretinal Implant
Age-related Macula Degeneration (AMD)	700,000	Epiretinal Implant
Glaucoma	120,000	Optic Nerve Implant
All causes of blindness	over 1 million	Cortical Implant

Retinal diseases:

Retinitis Pigmentosa
Age-related Macula Degeneration

Optic Nerve:

Glaucoma: optic nerve fibers are damaged and eventually die due to pressure in the eye or weakness of the support tissue around the optic disk

Difficulties due to lack of vision:

Reading and writing
Navigating through their environment
selecting clothes, making a cup of tea
Face recognition
Driving

II. Components of a vision prosthesis:

Hardware:

Input Device (ex. camera - to convert light into electrical signals)

an interface to the nervous system

Software:

convert image to appropriate stimulation pattern

a means of transforming image to retinotopic map

III. Survey of the current state of the field of visual implants:

Types of Visual Prostheses:

Subretinal:

Diseases	Symptoms of the Disease	# people in U.S. afflicted with Disease	Purpose of the Prosthesis	Type of Implant	Site of Implantation	Time of Implant	Array Specifications	Results from Implantation in Humans	Research Groups
Retinitis Pigmentosa (RP) Age-related Macular Degeneration (AMD)	RP: Gradual loss of retinal pigment epithelial (RPE) cells, and degeneration of photoreceptors (mostly rods) AMD: degeneration of the macula causes loss of central vision & destroys photoreceptors in macula region	RP = 100,000 AMD = 700,000	To be a functional substitute for the degenerated photoreceptors by transforming light into electrical signals	Microphotodiode Array (MPD)	Between the Pigment Epithelium and Retina, thus stimulates bipolar cells	Chronic	MPD array contains 5000 microphotodiodes/chip Each MPD measures 20 um X 20 um and adjacent units are separated by 10 um Diameter of chip = 2mm	After implantation, 6 people blinded by RP could pick out shapes and recognize faces (Chow)(Review of Opthamology 9(6) Aug. 15, 2002)	University of Tubingen (Prof.Zrenner) University of Stuttgart (Prof.Schubert) Optobionics Corp. (Dr. Alan Chow)

(Zrenner, 2001)
(Zrenner, 2001)(Zrenner, 2001)

Pros:

stimulating closest to photoreceptors takes advantage of retinal, thalamic, and cortical signal processing
if bipolar cells can be directly stimulated, retinotopic organization should be preserved
surgical complications not necessarily as significant as cortical approach

Cons:

requires functional retina and optic nerve pathway to convey signals to cortex
blockage of nutrients from choroid to remnant retina by the implant
very complex surgical access
subretinal implant must be externally powered ==> The amount of visible light that can reach the subretinal device is limited & insufficient to generate enough current to activate the MPD; the key to boosting the stimulation power of the MPD is with a continuous near infrared light (NIR)
it has been shown that silicon photodiodes are many orders of magnitude less sensitive than rods and cones

Epiretinal:

Diseases	Symptoms of the Disease	# people in U.S. afflicted with Disease	Purpose of the Prosthesis	Types of Implants	Site of Implantation	Time of Implant	Array Specifications	Results from Implantation in Humans	Research Groups
Retinitis Pigmentosa (RP) Age-related Macular Degeneration (AMD)	RP: Gradual loss of retinal pigment epithelial (RPE) cells, and degeneration of photoreceptors (mostly rods) AMD: degeneration of the macula (atrophy of retinal cells) causes loss of central vision & destroys photoreceptors in macula region	RP = 100,000 AMD = 700,000	To stimulate intact ganglion cells to convey meaningful visual stimuli to the impaired patient	1. Johns Hopkins University (JHU)-North Carolina State University (NCSU) device 2. MIT-Harvard Device	Ganglion cell layer	JHU-NCSU: Chronic MIT/Harvard: Acute (up to 8 hrs)	JHU-NCSU: 60 electrode array (5.5 x 5.25 mm); 10 x 6 electrode configuration MIT/Harvard: photodiode array (2 x 2 mm); 12 photodiodes in the array (4 x 3)	JHU-NCSU: patient could see spots of light in response to electrical current; spot locations were correlated with the electrodes' positions; patient was able to determine the orientation of the letter L MIT-Harvard: only animal trials	JHU-NCSU (Prof. Humayun-Prof. W. Liu) Humayun and Eugene de Juan are currently at the University of Southern California MIT-Harvard: (Prof. Rizzo-Prof. Wyatt) German Northern Consortium (Prof. Eckmiller-Prof. Heimann)

(Humayun, 1994)

How the JHU-NCSU prosthesis works:

2 units:

extraocular: (1) provides for image acquisition and processing

(2) consists of external camera, image processor chip, telemetry encoder, RF amplifier, and primary coil

intraocular: (1) receives power and data from the extraocular unit

(2) consists of a secondary coil, power rectifier and regulator, telemetry decoder, retinal stimulator chip, and an electrode array

1. Camera and electronic image-processing chip capture and convert a visual scene into pixels

2. Image data is sent via a telemetry link (laser or radio frequency modulated signal) to a retinal microchip implanted intraocularly

3. Retinal chip converts the transmitted image data and produces an appropriate pattern of small electrical currents that are applied through a grid of electrodes positioned

over the retina

Implanted the first permanent epiretinal implant system in a human (patient previously had no light perception vision due to RP) in February of 2002 (Review of Opthamology 9(6) Aug. 15, 2002)
Limitations: (1) low resolution caused by relatively large electrode diameter (0.5 mm) due to relatively high current required to elicit phosphenes

(2) electrodes are being short-circuited by the saline environment in the eye, and so require increasing amounts of current to work

(Wyatt, 1996, MIT/Harvard)

How the MIT/Harvard process works:

A tiny camera using charge-coupled devices is mounted on a pair of glasses and its output modulates the signal from a small fixed-direction laser also on the glasses
Laser beam (820 nm wavelength) simultaneously powers the implant and conveys the visual information
An array of electrodes sits on one end of a thin, flexible polyimide strip whose other end is sandwiched between two chips: a photodiode array facing the pupil of the eye (and the laser beam) and a stimulator chip which directs current to individual electrodes arrayed on the far end of the strip facing the retina and stimulates ganglion cells

Pros:

stimulating close to photoreceptors takes advantage of native processing power in the thalamus and cortex
surgical complications not necessarily as significant as cortical approach

Cons:

requires functional optic nerve pathway
may stimulate optic nerve fibers rather than cell bodies: this will greatly complicate visuotopic organization
difficult surgical access
difficult to adhere electrode array to retina

Neurotransmitter Implant

Diseases	Symptoms of the Disease	# people in U.S. afflicted with Disease	Purpose of the Prosthesis	Type of Implant	Site of Implantation	Time of Implant	Array Specifications	Results from In Vitro Tests	Research Groups
Retinitis Pigmentosa (RP) Age-related Macular Degeneration (AMD)	RP: Gradual loss of retinal pigment epithelial (RPE) cells, and degeneration of photoreceptors (mostly rods) AMD: degeneration of the macula (atrophy of retinal cells) causes loss of central vision & destroys photoreceptors in macula region	RP = 100,000 AMD = 700,000	To combine fiber optics, microfluidics, and photo-triggerable neurotransmitters in order to generate a brief pulse of light	Based on an inactive prodrug of glutamate called a "photo-trggerable glutamate" that is carried through a microfluidic conduit to many locations on the retina	Retina	Only in vitro tests thus far	No "array"	5-10 um spatial resolution corresponding to 20/40 or 20/20 acuity	Wayne State (Dr. Raymond Iezzi)

How the implant works:

(1) an inactive prodrug of glutamate called a photo-triggerable glutamate is carried through a microfluidic conduit (hair-fine channels made in planar substrates using silcon processing technologies replace beakers and tubing for automated liquid transport and handling on a sub-microliter scale) to many locations on the retina

(2) at each hole in the conduit there is a tiny fiber optic that allows the creation of a brief pulse of light, activating the glutamate pro-drug and causing a localized "chemical pixel"

(3) glutamate molecules are supplied to the eye via an implantable fluid pump (similar to the insulin pump used by some diabetics)

(4) on the other end of the fiber optic line is a computer/camera system

Limitation: neurotransmitters such as glutamate can overstimulate neurons causing cell death (apoptosis)

Hybrid Retinal Implant

Diseases	Symptoms of the Disease	# people in U.S. afflicted with Disease	Purpose of the Prosthesis	Type of Implant	Site of Implantation	Time of Implant	Array Specifications	Results from Implantation in Humans	Research Group
Glaucoma	optic nerve fibers are damaged and eventually die due to pressure in the eye or weakness of the support tissue	3 million (120, 000 blinded by disease)	To be a functional substitute for retinal ganglion cell loss and optic nerve degeneration	Hybrid	Subretinal layer and peripheral nerve graft (from peripheral sciatic nerve) transplant into optic tract	Chronic	Array substrate (20 x 6 mm) 9 pairs of stimulating sites (3 pairs x 3 pairs) with center to center spacing of 40 um each stimulating site is 20 um x 20 um	No human trials (only limited in vivo experiments with cats)	Nagoya University (Prof. Yagi)

(Yagi, 1998)(Yagi, 1999)

(Yagi, 1999) (Yagi, 1999)

Components:

(1) Micro-electrical mechanical system (MEMS)

(2) Transplanted neurons

How the prosthesis works:

(1) cultured neurons are attached onto an electrode array of the MEMS

(2) neurons and MEMS are implanted together in the subretinal space

(3) axons of the neurons will be guided with an axon guiding material (ex. peripheral nerve graft or tube which is filled with Schwann cells and Extracellular Matrix) to the lateral geniculate body (LGN) thereby connecting the prosthesis with the LGN.

Pro:

able to replace an eye with total or inner retinal degeneration

Cons:

difficulties in precisely directing axons to the LGN
developing the interface between the electronics and neurons

Optic Nerve

Diseases	Symptoms of the Disease	# people in U.S. afflicted with Disease	Purpose of the Prosthesis	Type of Implant	Site of Implantation	Time of Implant	Array Specifications	Results from Implantation into a Human	Research Groups
Retinitis Pigmentosa (RP) Age-related Macular Degeneration (AMD)	RP: Gradual loss of retinal pigment epithelial (RPE) cells, and degeneration of photoreceptors (mostly rods) AMD: degeneration of the macula (atrophy of retinal cells) causes loss of central vision & destroys photoreceptors in macula region	RP = 100,000 AMD = 700,000	To restore vision to patients who do not have functioning photoreceptors	Self-sizing Spiral Cuff Electrode	Around the Optic Nerve	Chronic	Cuff that contains 4 electrodes spaced at 90 degree intervals	Patient could see distinct phosphenes over a large portion of the visual field when electrical stimuli were applied to the optic nerve	University of Catholique Science Appliquees Microelectronics Lab (Belgium) (Prof. Veraart)

Self-sizing Cuff Electrode (shown in green):

uses an extracellular "cuff" electrode that surrounds the optic nerve
contains four electrical contacts
cable passes through the skull and courses below the skin and along the outer surface of the skull; it then passes down the neck to exit the skin below the clavicle
stimulation was either:

monopolar - using a surface indifferent anode
biploar - between two contacts within the cuff

Phosphene characteristics:

most phosphenes were reported as a set of dots (2-5) arranged in a cluster, or arranged in rows, arrays, or lumps of 6 - 30
solid lines, bars, or triangles with no dot structure were occasionally reported, usually near perception threshold
often reported as colored: yellow, blue, white;

Pros:

surgical complications not necessarily as significant as cortical approach

Cons:

requires functional optic nerve pathway
visuotopic organization requires placing electrodes at many closely spaced regions of the optic nerve
will require complex electrode array to provide any useful patterned vision
very difficult surgical access

Cortical Prosthesis

Diseases

Symptoms of the Disease

# people in U.S. afflicted with Disease

Purpose of the Prosthesis

Type of Implant

Site of Implantation

Time of Implant

Array Specifications

Results from Implantation into a Human

Research Groups

Severe cases of Retinitis Pigmentosa (RP) and Age-related Macula Degeneration

Glaucoma

Diabetic Retinopathy

RP: Gradual loss of retinal pigment epithelial (RPE) cells, and degeneration of photoreceptors (mostly rods)

AMD: degeneration of the macula (atrophy of retinal cells) causes loss of central vision & destroys photoreceptors in macula region

Glaucoma: optic nerve fibers are damaged and eventually die due to pressure in the eye or weakness of the support tissue around the optic disk

RP = 100,000

AMD = 700,000

Glaucoma =
3 million
(120,000 currently blinded by the disease)

To restore vision to patients who have sustained trauma and/or disease to the optic nerve, and people with severe cases of retinal diseases that leave few functional ganglion cells

Surface Electrode Array: Dobelle: 64 electrode

Intracortical Microstimulation: Normann: 100 microelectrode Utah Electrode Array

Dobelle: mesial surface of the right occipital lobe

Normann: Striate (visual) cortex of cat

Chronic

Dobelle: 64 electrodes

Normann: 100 microelectrode

Dobelle: Subject obtained visual acuity of approximately 20/120

Normann: only in cat

University of Utah (Prof. Richard Normann)

NIH - (Prof. Schmidt)Neural Prosthesis Program

Dobelle Group

Illinois Institute of Technology (Prof. Troyk)

Wayne State University (Prof. J.P. "Pat" McAllister)

Surface Cortical Electrodes:

Later efforts by Dobelle et al. (1976) included the implantation of a 64 channel platinum disk electrode array on the surface of the occipital cortex of blind patients

these electrodes were interfaced with a camera consisting of a 100 x 100 charge-coupled phototransistor array
random letters were used to stimulate the camera, which in turn used to stimulate only 6 out of 64 electrodes
prosthesis allowed the blind patients to recognize 6-inch characters at 5 feet (approx. 20/1200 visual acuity)
phosphene brightness was a logarithmic function of stimulating current amplitude

64 electrode cortical implant

(Dobelle, 2000)

Basic design scheme:

Camera -->Edge Detector --> Array --> Phosphene generation

Intracortical microstimulation (ICMS)

example: Utah Electrode Array

The large number of penetrating electrodes in the UEA presents a very large surface area to the cortex and the implanted array tends to self anchor to the cortical tissues
Such an array has the strong advantage that it floats in the cortical tissues as the cortex moves due to respiration and blood pumping, or due to skeletal displacements, the array moves with it, thereby producing little or no relative motion between the electrode tips and the neurons near its active tips
Pros:

the only therapeutic approach for individual with non-functional retinas and/or optic nerves
implant site is robust and protected by skull
easy surgical access
relatively high density electrode implantation has been demonstrated
phosphene thresholds are low (1 - 10 uA range)
electrode array architecture is likely to be well suited for application in other sensory or motor regions of the cerebral cortex

Cons:

stimulation site far from photoreceptors (no retinal or thalamic processing), thus some visual processing is missing
possibly poor visuotopic organization
problems of multiple feature representations in V1 (color, lines, motion, ocular dominance)

IV. Current problems:
1. lack of basic experiments on phosphene psychophysics (Normann, et al 1999)
2. lack of an array that generates optimal resolution, field of view, and grid spacing for phosphene

non-conformal retinotopic mapping may require electronic remapping of visual inputs onto cortical electrode array

Maynard (2001)

(Maynard, 2001)
Left: A receptive field map of cat visual cortex obtained with simultaneous recordings from a UEA. The boxes show the likely position of phosphenes generated by stimulating electrodes in a row and column of the array
Right: A representation of the conformality of the receptive field map. the measured receptive fields are shown as square boxes. Lines go from the receptive field to the corresponding position on the electrode array

multiple visual representations in cortex. Visual cortex encodes color, intensity, orientation, ocular dominance, and other spatial features. It is unclear what kind of percepts will be evoked by stimulation of large numbers of cortical units
Convoluted cortical anatomy can make implantation of large arrays or many small arrays difficult. So, may need to design different structures for the array such as the Utah Slant Array (shown below):

V. Psychophysical experiments with simulators

Cha et. al. used an image and applied a mask to it
We could use:

an Edge Detection (Sobel) plus mask
our algorithm for edge detection (instead of Sobel) plus mask

We are developing an algorithm ==> gray scale images and produce perceptually salient contours

could be better form of input because it would focus on important parts of the object thereby decreasing the amount of noise and decrease the required resolution, field of view, and grid spacing

VI. Future Directions:

To develop the best form of input to the implanted electrode array

Achieve this by:

psychophysical experiments with a simulator (Cha, et al. 1992) to understand how multiple phosphenes interact
refining the parameters of grid spacing, resolution, and field of view

Further development in a cortical vision prosthesis requires:

using the 625 electrode array which has been built with the same techniques used to build the UIEA
better understanding of the biocompatibility of electrode arrays in primates
behavioral experiments in primates to determine the stability of the stimulation thresholds and evoked visual perceptions
short-term experiments in human volunteers to replicate the results of Schmidt et al. (1996) evaluating stimulation parameters for optimal phosphene generation

Similar to the work done by John Donoghue in the motor cortex, we would like to insert a probe into V1, measure the firing rates and LFP, and then feed back the information to the array for visual cortex (V1)
We may be able to use somatosensory input for perception of visual stimulation (Disbrow et al. 2000)

VII. Questions:

Implant in the LGN?
Is a camera the best source of input?
Are phosphenes the best for of image generation?

VIII. Groups

Epiretinal:

Harvard/MIT: http://rleweb.mit.edu/retina/

Mass. Eye and Ear Infirmary, Harvard Medical School (Prof. Joseph Rizzo)
Circuits and Systems Group of the Research Laboratory of Electronics, MIT (Prof. John Wyatt)

Johns Hopkins and NCSU

The Johns Hopkins University (Prof. Mark Humayun) http://www.irp.jhu.edu

North Carolina State University (Prof. Wentai Liu) http://www.ece.ncsu.edu/erl/erl_eye.html
German Northern Consortium (Germany)

University of Bonn (Prof. Ralf Eckmiller) www.nero.uni-bonn.de/ri/retina-en.html
University of Cologne (Prof. Heimann and Dr. Peter Walter) www.medizin.uni-koeln.de/kliniken/augenklinik/epi-ret3e.htm
University of Duisburg www.oe.uni-duisburg.de/Reports/jb95/ni2.html

Sub-Retinal:

Germany:

University of Tuebingen (Prof. Erhart Zrenner) www.uak.medizin.uni-tuebingen.de/depii/groups/subret/
University of Stuttgart (Prof. Schubert) www.ims-chips.de/products/retina/retina/html

America

Optobionics Corp. (Dr. Alan Chow)

Optic Nerve:

University of Catholique Science Appliquees Microelectronics Lab (Belgium) (Prof. Veraat)www.dice.ucl.ac.be/Mivip/

Cortical Implant:

University of Utah (Prof. Richard Normann) www.bioen.utah.edu/faculty/RAN/index.html
NIH - (Prof. Schmidt)Neural Prosthesis Program www.ninds.nih.gov/npp
Dobelle Group www.dobelle.com/index.html
Wayne State University (Prof. J.P. "Pat" McAllister) www.med.wayne.edu/Wayne%20Medicine/wm2000/lettherebesight.htm
Illinois Institute of Technology (Prof. Troyk) www.iit.edu/~npr/intro.html

Models:

Cha et al. (1996)----"Pixelized Vision"

Concept:

the device operates on the principle that if an electric current is passed through a single electrode into the visual cortex, then the patient "sees" a single point of light, called a PHOSPHENE (perceptual spot of light)

Methods:

A phosphene simulator was designed to mimic the sensation created by a field of phosphenes generated by an electrical array.
Since the simulator was used on healthy subjects with vision of 20/30 or better, it merely imitated the effect of an electrical array without any actual implantation
a video camera encodes the image and projects it to a small monochrome monitor
a perforated mask was placed between the eye and the monitor to create the illusion of phosphenes

two set of masks were used:

first set had a fixed-field so that as the number of pixels increased, the spacing between pixels decreased
the second set had fixed-spacing so that as the number of pixels increased the field increased
each set contained four masks which had: 1024 (32x32), 625 (25x25), 256 (16x16), or 100 (10x10) pixels
the 32x32 mask was the same for both sets and a clear mask was also provided as a control

Conducted a series of psychophysical experiments with a portable "phosphene" simulator to obtain estimates of suitable values for electrode number and spacing
Simulator consisted of a small video camera and monitor worn by a normally sighted human subject
To simulate a discrete phosphene field, the monitor was masked by an opaque perforated film.
Results: a functional visual sense may be restored with arrays containing as few as 625 electrodes

A total of 625 pixels of appropriate visual information (placed in a square grid, with a center-to-center spacing of 4 min of arc), can provide sufficient visual information to allow sighted volunteers to read at 67% of the rate that they could read with normal vision.
625 pixels allow volunteers to navigate complex visual environments with normal speed and good levels of confidence

Conclusion:

625 electrodes implanted in a 1 cm by 1 cm area near the foveal representation of the visual cortex should produce a phosphene image with a visual acuity of approximately 20/30

Schmidt et al. (1996)

used intracortical microstimulation (ICMS)
38 microelectrodes were implanted in the right visual cortex for a period of 4 months

Left: Schematic diagram of dual "Hat-Pin" microelectrode used for ICMS
Right: Photograph of exposed surface of the right visual cortex of the blind subject. The overlaid dots, at ~2.4mm spacing, were reference points for surface stimulation. The numbers in the figure have been placed on the approximate positions of the intracortical microelectrodes. The anatomical pole of the occipital cortex is estimated to be near microelectrodes 14 and 15. The terminal portion of the calcarine fissure is marked by an arrow and superior is to the left of the arrow. A centimeter scale is shown at the lower left

Phosphene: small spot of light
Phosphenes were produced with 34 of the 38 implanted microelectrodes
Properties phosphenes:

Size: range from pinpoint to nickel (~20mm diameter) held at arm's length
Color: yellow, blue, or red (not green)
Onset response latency: 395ms
Duration: 250ms
Number: depending on stimulation parameters, an individual microelectrode could either produce a single phosphene or a pair
Threshold currents for phosphene production usually increased after repeated stimulation
Multiple phosphenes:

when 3 single phosphenes were produced with currents near threshold (microelectrodes 12, 18, 32 in parallel with 33) the subject described them as closely spaced in a vertical orientation
when 6 simultaneously presented phosphenes were in a vertical orientation and all just touching, the subject reported that the size of the resultant image was adequate to form a letter of the alphabet, such as "I" or the branch of a letter such as "M"

Movement: the position of the perceived phosphene moved in the direction of eye movements

Phosphene Map:

Threshold currents for phosphene generation with trains of biphasic pulses were as low as 1.9uA and most of the microelectrodes had thresholds below 25 uA
Key problem: after second month of testing, more than half of the electrodes had broken

Nordhausen(1996)

100 microelectrode array - Utah Intracortical Electrode Array (UIEA)
implanted it into cat striate cortex (responsible for the first level processing of visual signals) and seems to work; they plan on implanting it into a human subject
58.6% of the electrodes in the array were capable of recording visually evoked responses
average signal-to-noise (SNR) ratio from 56 of the electrodes in the array was calculated to be 5.5 : 1
average of 3.4 separable spikes were identified for each of these electrodes
orientation preference and ocular dominance maps were generated for each of the evoked responses

Left

Middle

Right

Brindley and Lewin (1968) implanted devices consisting of 80 electrodes on the visual cortex of blind patients

Left

Right

Wires through a burr hole connected each electrode to a radio receiver screwed to the outer bony surface of the skull
an oscillator coil was placed above a given receiver in order to activate the receiver via radio frequency and stimulate the cortex
blind patients were able to to see phosphenes at different positions of the visual field
showed that a chronically activated electrical stimulation multichannel device was possible
Difficulties:

interactions between phosphenes
multiple phosphenes induced by the same electrode
inconsistency of phosphenes
usage of high currents
large electrodes
occasionally pain caused by meningeal stimulation and focal epileptic activity was induced following electrical stimulation

Dobelle (2000)

Problem: it appears that they placed the electrodes on the visual association cortex (areas 18 and 19) instead of primary visual cortex (area 17); resulted in a differently shaped phosphene map than the ones reported earlier by Brindley (1968).

Movies:

Current Edge Detection Process: Sobel filters:

References

Cha, K. Simulation of a Phosphene-Based Visual Field: Visual Acuity in a Pixelized Vision System. Annals of Biomedical Engineering. 20. 439-449. 1992
Dobelle, W. Artificial Vision for the Blind by Connecting a Television Camera to the Visual Cortex. ASAIO. 46: 3-9. 2000
Gekeler F., Schwahn H., Stett A., Kohler K., Zrenner E. Subretinal microphotodiodes to replace photoreceptor-function. A review of the current state. Les Seminaires Ophtalmologiques
d'IPSEN, tome 12: 77-95. 2001
Humayun, M. Pattern electrical stimulation of the human retina. Vision Research. 39: 2569-2576. 1999
Margalit, E. Retinal Prosthesis for the Blind. Survey of Ophthalmology. 47(4):335-356. 2002
Maynard, E. Visual Prostheses. Annu.Rev.Biomed.Eng. 3:145-68. 2001
Nordhausen, C. Single unit recording capabilities of a 100 microelectrode array. Brain Research. 726: 129-140. 1996
Normann, R. A neural interface for a cortical vision prosthesis. Vision Research. 39: 2577-2587. 1999
Schmidt, E. Feasibility of a visual prosthesis for the blind based on intracortical microstimulation of the visual cortex. Brain. 119:507-522. 1996
Veraart, C. Visual sensations produced by optic nerve stimulation using an implanted self-sizing spiral cuff electrode. Brain Research. 813:181-186. 1998
Wyatt, J. Ocular implants for the blind. IEEE Spectrum. 49-53. 1996
Yagi, H. Efficient Stimulation inducing neural activity in retinal implant. IEEE 409-413. 1999