Introduction

Many blind individuals will not benefit from the development of a retinal prosthesis. The reason for this is simple: retinal prostheses rely on the circuitry of the brain to transmit electrical signals from the eye to the center of visual processing in the brain, the visual cortex. If this circuitry is not functional, a prosthesis must bypass these systems and intervene directly at the cortical level. The group of people for which this applies (see "Who Benefits") includes anyone who has sustained trauma to the optic nerve, as well as individuals with severe cases of retinal diseases that leave few functional ganglion cells. Of course, if a cortical visual prosthesis (CVP) works well, a patient that would also qualify for a retinal prosthesis could opt for a CVP instead.

The history of the cortical prosthesis begins in 1929 when Foerster investigated the effects of electrical stimulation of the occipital lobe of the human cortex [3]. He found that this stimulation caused a subject to "see" a small point of light, later called a "phosphene". This result was reproduced many times after the original experiment with both sighted subjects and blind subjects. The idea that concurrent stimulation of many sites in the brain could produce a single coherent image was postulated as early as 1953 by Krieg [4]. Because there is rough retinotopy in the visual cortex, Krieg thought it would be possible to use this technique to restore sight to the blind.

In order for a CVP to be even a remotely viable option, a permanent device for chronic stimulation of neural tissue needed to be developed. This was accomplished in 1968 by Drs. Brindley and Lewin [1]. Their device was implanted in a 52-year-old woman who had gone totally blind six months before the operation. The device had 80 electrodes, each with its own controlling unit (receiver). The set of 80 receivers sat directly beneath the pericranium, while the electrodes lay on top of the occipital cortex. As described in the paper, "To activate a given receiver, and so stimulate the cortex through its electrode, the transmitting coil of an oscillator tuned to the appropriate frequency is pressed against the scalp immediately over it." Using this system, Brindley and Lewin were able to demonstrate the plausibility of a permanent CVP.

Brindley and Lewin's device was an immediate success. An electrode voltage of about 25 V resulted in the subject's perception of a small point of light. More importantly, many electrodes could be stimulated at the same time, resulting in many small points of light. The position of the electrodes corresponded roughly to the position of the phosphenes in the visual field, and the subject was able to identify patterns of phosphenes. Of course, this device was far from ideal. One of the major problems they reported was that a single electrode could cause many phosphenes to appear. Sometimes this was a function of current, with a greater current leading to the production of more phosphenes, but other times it was independent of stimulus parameters. Additionally, there was a serious limitation on spatial resolution: electrodes spaced closer than 2-3 mm created a large strip of light when activated simultaneously. Finally, the patient reported seeing flicker in every phosphene created.

The next major advance pertaining to CVPs was made by Drs. Dobelle and Mladejovsky in 1974 [2]. They tested various parameters of electrical simulation of human visual cortex on 38 volunteers who were admitted into a hospital for non-elective cranial surgery. Even though they did not implant an actual prosthesis, they were able to provide important data because of the number of patients analyzed (compared to Brindley and Lewin's one subject). Some of the results they obtained complemented those presented by Brindley and Lewin, but others directly conflicted. For example, they also found that a single electrode could elicit a multi-phosphene response, and their subjects also reported constant flicker of the phosphenes. However, in contrast to the 1968 paper, Dobelle and Mladejovsky concluded that a constant stimulus did not produce a sustained phosphene, but rather one that grew dimmer over time and eventually faded after 10-15 sec. This is important in designing a cortical prosthesis, because it means that an image will have to be refreshed at a certain rate to prevent adaptation. Another important design consideration results from the fact that phosphenes move with eye movements. That is, if a phosphene appears in the center of vision when a subject is looking straight ahead, it will appear in the right hemisphere when the subject looks to the right. An eye position-detector is therefore needed if an image is to be stabilized in the visual field. Finally, although safety was not considered an issue while they were conducting their experiments, the currents required to produce phosphenes (3-5 mA) were potentially dangerous. A phenomenon known as "kindling" can occur when the cortex is repeatedly stimulated with high currents -- the added electrical activity can cause local seizures. With these problems in mind, researchers continued to work towards an implantable device.

Throughout most of the 1990's there were two groups working towards a permanent CVP. The first was based at the National Institutes of Health (NIH) in Washington, D.C., and was headed by Dr. E. M. Schmidt. The second was based at the John Moran Laboratories in Applied Vision and Neural Sciences at the University of Utah, headed by Dr. R. A. Normann. Both of these groups approached the problem of a cortical prosthesis differently than their colleagues in the 1970's: instead of using surface electrodes on the visual cortex, they chose to employ penetrating microelectrodes.

The first paper from the NIH group was published in 1990 [5], and it described the various parameters required to produce phosphenes from intracortical microstimulation (ICMS). The motivation for this research came from reports of low stimulus thresholds for visual cortex ICMS in primates. Indeed, instead of the 3-5 mA threshold described by Dobelle and Mladejovsky, they reported a more modest range of 20-200 uA (depending on depth of insertion, from 3-5 mm) in their human subjects. Moreover, the phosphenes they produced were described as identical to the ones generated by surface stimulation, except that these did not flicker at all. Also important is the difference in spatial resolution between surface stimulation and ICMS -- Schmidt was able evoke the percept of two distinct phosphenes with microelectrodes separated by only 0.7-1.0 mm.

Because of the encouraging data in the 1990 paper, the NIH group proceeded to manufacture a 38-microelectrode penetrating array for implantation in a blind volunteer. The subject they chose was a 42-year-old woman who had been blind for 22 years due to severe glaucoma. She was tested with the device frequently over the course of 4 months, during which time she had a percutaneous connector emanating from the back of her head. Since there was no power source incorporated into this design, she was unable to use the prosthesis outside the lab.

Initially, the results were encouraging. It was unclear whether someone who had been blind for 22 years would respond to ICMS, but they were able to evoke a response from 36 of the 38 microelectrodes. The group published an extensive account of the sensations reported by their subject, and she experienced many of the same visual sensations as Brindley and Lewin's subject [6]. The size of the phosphenes ranged from a "pinpoint" to a "nickel held at arm's length". They were colored either white, yellow, red, or blue, but not green. Sometimes the phosphenes appeared to be different distances away from the subject, and they always moved when she moved her eyes. However, the researchers noticed that most of these effects were variable by changing the current or presenting multiple phosphenes simultaneously. For example, as the number of phosphenes generated increased, they became increasingly coplanar and uniformly colored.

After a few weeks' time, the project became fraught with difficulties. Although the researchers expected to find an optimal set of stimulus parameters, they found that their subject adapted differently to different aspects of the electric stimulus, and the optimal settings changed over time. For example, the current threshold level required to produce a phosphene increased by an average of 52% over a sequence of 50 stimuli. During that same interval, the brightness of a given perceived phosphene decreased by 75% . These and other issues of accommodation could pose a significant problem in the development of a permanent CVP, because it is difficult to design a device when the specifications are continually varying. Separate from the problem of finding stable stimulus parameters, the NIH group had problems with the microelectrode array itself. By the second month of testing, more than half of the electrodes had broken. Some of this breakage was due to accidental movement during sleep, some was due to the initial insertion of the array, and still some was due to bad luck. Obviously, a permanent implant must be viable for much longer than this experimental device.

To determine whether the subject could recognize meaningful spatial patterns of stimulation, the researchers needed to map the location of the phosphenes in the subject's visual field. The reason for this is that the retinotopy in the cortex is true for gross measurements, but not for precise locations. Furthermore, the cortical representation of the visual field is distorted, not linear; stimulation of electrodes in a square grid will not produce the perception of a square. Mapping the location of phosphenes is simple with a sighted subject, but it is significantly more difficult with a blind individual. Without any point of reference, it is difficult for a blind person to reliably indicate the location of phosphenes. This was demonstrated by research conducted last summer [7], and it remains a problem in the application of a cortical prosthesis. Regardless, the NIH subject reported seeing a vertical line when a set of seven electrodes were activated simultaneously, and she suggested that "the size of the resultant image would be adequate to represent a letter 'I' or one leg of the letter 'M'." This suggests that an ICMS-based CVP might be adequate for a reading aid.

While 38 electrodes are useful in proof-of-concept experiments, a commercial CVP would require a significantly larger number of electrodes. As an incremental advance towards achieving that goal, Dr. Richard Normann has been working on the development of the Utah Intracortical Electrode Array (UIEA), a 100 microelectrode array designed for recording and stimulating single cells in the cortex [8]. Although the long-term safety and stability of the UIEA have not yet been demonstrated, they have proven that the design works and does not immediately damage the neural tissue into which it is inserted [9]. The next stage in this research is the implantation of the UIEA in a human subject.

Although the cortical visual prosthesis was conceptualized almost 50 years ago, the technology required to build such a device has become available only recently. The first generations of CVPs will likely have between 100 and 300 microelectrodes and be useful as reading aids. As the technology advances, devices with more and closer-spaced electrodes will become available, increasing the resolution so that the device can be used to navigate while walking. Eventually, the scenario described in the introduction will become standard, with a video camera processing scenes in real time and transmitting them to a high-resolution microelectrode array. Before any of this can happen, however, studies must be done on the long-term safety of ICMS and the long-term durability of a cortical implant. Furthermore, a good CVP will require precise eye-tracking to stabilize images (see History, above), an accurate method of mapping phosphenes (see 90's, above), and a way to transmit signals to the device without percutaneous connectors to minimize the chance of infection. Finally, more mundane issues will have to be considered, such as minimizing power consumption, making the device cost-effective. The timeline for these developments is difficult to predict, but the outlook is decidedly positive: cortical visual prostheses are on their way!

[1] Brindley, G.S. and Lewin, W.S. (1968) The sensations produced by electrical stimulation of the visual cortex. J. Physiol., 196, 479-493.

[2] Dobelle, W.H. and Mladejovsky, W.G. (1974) Phosphenes produced by electrical stimulation of human occipital cortex, and their application to the development of a prosthesis for the blind. J. Physiol., 243, 553-576.

[3] Foerster, O. (1929) Beitrage zur Pathophysiologie der Sehbahn und der Sehsphare. J. Psychol. Neurol., Lpz, 39, 463-485.

[4] Krieg, W. In: Functional Neuroanatomy, Second Edition. New York: Blakiston, 1953: 207-208.

[5] Bak, M., Girvin, J.P., Hambrecht, F.T., Kufta, C.V., Loeb, G.E., and Schmidt, E.M. (1990) Visual sensations produced by intracortical microstimulation of the human occipital cortex. Med. & Biol. Eng. & Comput., 28, 257-259.

[6] Schmidt, E.M., Bak, M, Hambrecht, F.T., Kufta, C.V., O'Rourke, D.K., and Vallabhanath, P. (1996) Feasibility of a visual prosthesis for the blind based on intracortical microstimulation of the visual cortex. Brain, 119, 507-522.

[7] Vogelstein, J.V. and Dagnelie, G. (1998) Phosphene Mapping Strategies for a Cortical Visual Prosthesis. Poster presentation at: 29^th Neural Prosthesis Workshop, NINDS at NIH, Bethesda, Maryland.

[8] Maynard, E.M., Nordhausen, C.T., and Normann, R.A. (1997) The Utah Intracortical Electrode Array: a recording structure for potential brain-computer interfaces. Electroencephalogr. Clin. Neurophysiol., 102, 228-239.

[9] Rousche, P.J, and Normann, R.A. (1999) Chronic intracortical microstimulation (ICMS) of cat sensory cortex using the Utah Intracortical Electrode Array. IEEE Trans. Rehab. Eng., 7, 56-68.