cortex. They have developed an image processing system that mimics the human retina. The signals from the retina module are converted into neuromorphic pulsecoded signals through a circuit that emulates the function of the retinal ganglion cells [1]. They are currently performing in vitro experiments (for biocompatibility, in vivo animal experiments (acute and chronic)) and working towards human implants. Initially they will use the Utah Electrode Array [37] while developing a 3D probe array.

15.6.5  Lateral Geniculate Stimulation

Pezaris and Reid at Harvard Medical School [40] have demonstrated in primates that microstimulation in the lateral geniculate nucleus (LGN), which is the relay between the retina and the visual cortex, produced localized visual percepts. To assess the effects of microstimulation of LGN in a primate, an eye movement task was used with visual targets presented on a computer screen or through microstimulation. Saccades made to electrical targets were comparable to saccades made to optical targets. They estimate that 200–300 stimulation sites are available in the LGN. This would be adequate for reading with a visual prosthesis. However, developing the required electrode arrays and implanting them in the LGN is a formidable task.

15.7  Microelectrode Arrays and Stimulation Hardware

The University of Michigan has a long history in the development of multi-site silicon stimulating probes [54, 55]. Their resent development is a 64-site wireless microstimulator (Interstim-2B) [36]. Up to 32 chips can be connected in parallel to drive 2,048 stimulation sites. This should be more than adequate for any currently planned visual prosthesis.

PolySTIM Neurotechnologies Laboratory in Montreal, Canada, Has developed a power efficient stimulator for an intracortical visual prosthesis [19].

Delbeke et al. [20] have developed a microsystem based stimulator for an optic nerve prosthesis.

15.7.1  Miniature Cameras

A group at Shanghai Jiao Tong University, Shanghai, China have developed a micro-camera that can be implanted in eye and powered by a solar array positioned in front of the iris [12]. Since phosphenes move with eye movement, an eyemounted camera should help to stabilize the perceived image. The camera provides a 32 × 32 element image, which with their simulation studies allowed a subject to recognize simple scenes. Through simulations, they also found that a 12 × 12 array

of pixels was sufficient to recognize Chinese characters [13]. With a 10 × 10 array, the recognition level dropped to slightly under 50%.

The retinal visual prosthesis group at the University of Southern California, USA, is also developing a camera implantable in the eye.

PolySTIM Neurotechnologies Laboratory has developed a CMOS multimode digital image pixel sensor (MIPS) for a visual prosthesis [43]. Three selectable operation modes are combined in the proposed MIPS: a high dynamic range logarithmic mode, a linear integration mode, and a novel differential mode between two consecutive images. This last mode allows 3D information for a cortical stimulator.

15.7.2  Animal Models

The major groups that are investigating the entire realm of aspects leading to human implants of a visual prosthesis are employing animal models at some stage of their work. Other groups are just looking at animal models and how they might apply to a visual prosthesis.

The group at IIT/UC [6] have chronically implanted arrays of microelectrodes in non-human primates to evaluate intracortical stimulation. One of the major findings was that the stimulation package originally developed under an NIH contract, as described on the IIT web site [32], could not be connected to the intended number of microelectrodes at surgery. Small electrode-stimulator modules had to be developed that used telemetry to transmit power and stimulation in formation. At MIT, Tehovnik and colleges [47, 48] have used moveable microelectrodes to map the generation of saccadic eye movements and study how these data might be applicable to a visual prosthesis. DeYoe [21] and Bartlett [5] at the University of Rochester used moveable microelectrodes to study stimulation parameters and laminar distribution of phosphene production in non-human primates. These studies will aid in the development of a human visual prosthesis.

15.7.3  Image Processing and Phosphene Mapping

Part of the CORTIVIS project is the development of a bio-inspired visual processing front-end that would be placed between the photosensor array and the stimulator for an intracortical visual prosthesis [18]. The images are processed by a set of separate spatial and temporal filters that mimic the functions of the photoreceptors, amacrine and bipolar cells in order to enhance specific features of the captured visual image.

The C-Sight Visual Prosthesis Group in China has been studying tactile phosphene mapping in sighted subjects using a head mounted display for the simulated phosphenes and a 19 in. touch screen to record the subject’s tactile position [14, 15].

PolySTIM Neurotechnologies Laboratory has surveyed image processing strategies that can be used with a visual prosthesis [10].