15.1  Background

Visual sensations produced by stimulation of the visual cortex in human patients were well known to German neurosurgeons, Kraus [34] and Foerster [27], as early as 1924. A number of reports has been published over the years describing the effects of electrical stimulation of the visual cortex in lightly anesthetized surgical patients [38, 39]. When their visual cortex was stimulated, patients usually report small spots of light called phosphenes.

Shaw [45] obtained a patent for a “Method and Means for Aiding the Blind”. In his system, a photoelectric tube controlled the intensity and/or frequency of an electrical stimulus that was applied directly by internal electrodes, or indirectly by external electrodes to the visual areas of the brain. Although this appears to be one of the first concepts of a visual prosthesis, actual implementation of the system has not been found.

Button and Putnam [11] demonstrated, in blind subjects, visual responses to intracortical stimulation controlled by a photoelectric cell. This allowed the subjects to identify a light source by orientation of the cell. Of the three subjects, one was able to follow a flashlight carried by an attendant 15 ft away.

15.2  Cortical Surface Stimulation

The first chronic experiment to determine the effects of stimulating the visual cortex was carried out by Brindley and Lewin [9]. They implanted an array of 80 electrodes on the medial surface of the occipital pole in a 52 year-old woman who had been totally blind for 6 months. The electrodes were platinum squares 0.8 mm on a side. They were connected to 80 radio receivers mounted to the skull, beneath the pericranium. Alternate receivers were tuned to 6.0 or 9.5 MHz. Pressing a transmitter coil on the scalp above a receiver and applying the proper frequency provided stimulation currents to the associated electrode. With the technology available at the time, 80 receivers covered half of the cranium.

When electrodes that produced phosphenes within 10° of the fovea were stimulated, the patient reported a very small spot of light, or phosphene, and described it as “the size of a grain of sago at arm’s length” or “like a star in the sky”. Phosphenes further from the fovea were sometimes elongated, “like a grain of rice at arm’s length”. The most peripheral phosphenes were round like a cloud. There were three electrodes that produced a pair of phosphenes about a degree apart and two electrodes

that produced a row of three phosphenes each about a degree apart from the next. When multiple phosphenes occurred, stimulus amplitude could not be adjusted to produce single phosphenes. For 13 electrodes, weak stimulation produced a single phosphene but higher-level stimulation produced a second phosphene in a different part of the visual field.

Other significant findings from this patient were:

1. Phosphenes always flickered regardless of stimulation parameters.

2. Phosphenes moved with eye movement.

3. Phosphenes could usually be resolved that were produced by electrodes spaced 2.4 mm apart.

4. Phosphenes usually ceased immediately at the end of stimulation, but after strong stimulation they could persist for up to 2 min.

5. Stimulation of multiple electrodes could produce simple patterns.

By improving the experimental prototype, Brindley and Lewin [9] believed that at least 200 electrodes per hemisphere could be implanted and would permit blind patients to read and navigate.

Dobelle and Mladejovsky [22] were able to conduct a series of acute experiments involving volunteers undergoing neurosurgical procedures for removal of tumors or other lesions to verify the results of Brindley and investigate the possibility of producing a visual prosthesis. Dobelle’s data are based on 16 experiments in 15 volunteers. They were able to confirm most of Brindley’s results from a single volunteer. A summary of the results obtained from Dobelle’s experiments were:

1. Phosphene chromatic effects or flicker may or may not occur. 2. Phosphenes moved with eye movement.

3. Two-point discrimination was about 3 mm.

4. Phosphenes appear immediately when stimulation is begun and end immediately upon cessation of stimulation.

5. Phosphenes fade after 10–15 s of continuous stimulation. 6. Multiple phosphenes are co-planar.

7. Thresholds ranged between 1 and 5 mA, with 3 mA being typical.

8. Electrodes of 1, 3, and 9 mm² size had similar thresholds and percepts. 9. Brightness modulation can be achieved by changing pulse amplitude.

From these studies, it was apparent that to provide a blind person with a stable image, either the subject had to learn to use head movements instead of eye movements, or the camera used by the visual prosthesis had to move with eye movement. Also, long stimulation trains had to be interrupted to compensate for phosphene fading.

Dobelle’s group chronically implanted four volunteers in the 1970s with a subdural 64-electrode array placed on the medial surface of the visual cortex of the right occipital lobe. The wires were terminated in a 72-pin micro-miniature connector encapsulated in a transcutaneous pyrolytic carbon pedestal, attached to the cranium by platinum bone screws.

Of these four volunteers, two had useful results for the future of artificial vision. One of them, blind for 10 years and implanted in 1975 at age 33, could perceive 46

useful phosphenes out of 60. Using six phosphenes with a layout similar to that of a Braille cell, he could read cortical Braille at approximately five words per min but he could only read tactile Braille at one word per min [23]. He could identify the orientation of white strips of tape on a blackboard by manipulating a video camera mounted on a joystick. His phosphene map stayed constant and his thresholds only had small changes over 10 years.

Another volunteer, blind for 7 years and implanted in 1978 at age 41 with an identical 64-electrode array could perceive 21 useful phosphenes. Over the last 25 years, his phosphene map and thresholds have stayed constant. In the late 1990s this volunteer benefited from the miniaturization of electronic components and advances in computer technology. He was the first blind volunteer to wear a miniature video camera mounted on his eyeglasses and a sub-notebook computer, a stimulator, and batteries in a waist pack [24]. Using an edge detection algorithm, the images from the video camera were processed by the computer, which selected the electrodes that produced phosphenes on or near the high-contrast areas of the images. The stimulator in turn generated the proper stimuli for the selected electrodes.

Compactness and portability of the system allowed the subject to detect and negotiate objects, follow a child walking slowly and close to him in a hallway, follow a strip of black tape on the floor, enter a room, grab a ski cap hung on the opposite wall, turn around, walk towards a mannequin and put the cap on its head. Accompanied by staff in the NY City subway system, an environment he was familiar with, he could get inside a subway car. He found it easier to differentiate the space between two cars and an open car door with his visual prosthesis than with his cane.

The results of the research done on these two volunteers, particularly the last one, were quite promising. If they could achieve all this using a single array with a limited number of phosphenes, the logical conclusion was that with two arrays, blind patients would have more phosphenes, creating images with higher resolution, therefore giving them more independence and mobility.

15.3  Intracortical Microstimulation

In cat motorsensory cortex, Stoney et al. [46] showed that thresholds for facilitation of spinal motorneuron pools by intracortical microstimulation (ICMS) could be as low as 2 mA, which is 1/100 of the threshold for producing similar effects with surface stimulation (Asanuma et al. [2]). These results led Dobelle & Mladejovsky to try ICMS in patients where the cortex was going to be surgically removed. This was not successful, possibly due to pathological involvement of the cortex in question [22].

In 1980 Bartlett and Doty [4] investigated the ability of primates to detect ICMS of the visual cortex. They advanced microelectrodes through the visual cortex and recorded the primate’s threshold for detection of the stimulus. They found thresholds significantly lower than surface stimulation, with some thresholds as low as

2 mA (0.2 ms at 50 Hz). It was not apparent if the primates were responding to phosphenes similar to those produced by surface stimulation in humans. If the primates were seeing phosphenes then it appeared that it might be possible to produce an intracortical visual prosthesis requiring much less power than using surface stimulation. This question could only be answered in human subjects.

Dr. Hambrecht, who was Director of the Neural Prosthesis Program at the National Institutes of Health (NIH), assembled a team of scientists to determine if ICMS was suitable for use in a human visual prosthesis. Protocols were approved at the NIH and at the University of Western Ontario to test patients who were undergoing surgery for excision of epileptic foci in the visual cortex. Three patients were studied in Canada for 1 h each [3] by first briefly stimulating the exposed cortex with a surface electrode and then inserting pairs of electrodes into the region where the patient reported phosphenes. As the electrodes were advanced through the cortex, the threshold for phosphene production dropped from as high as 5 mA at the surface to about 20 mA at 2–3 mm from the surface. Near threshold, the phosphenes were usually blue, yellow or red. The phosphenes did not flicker. With interleaved stimulation of two microelectrodes that were 0.7–1 mm apart, the patient reported “two blobs fusing.” When the tip separation was 0.3 mm, the percept was a singular round shape.

The next step in developing a visual prosthesis was to chronically implant a blind human volunteer with an array of intracortical electrodes. Hambrecht [29] provided an excellent review of the next study and Schmidt et al. [44] provided the details of the human experiment. This study was limited to a 4-month investigation as set out in the approved protocol.

Thirty-eight microelectrodes were implanted in the visual cortex. They consisted of 12 single microelectrodes and 18 pairs. The spacing between pairs of microelectrodes was 250, 500 or 750 mm. Two of the microelectrode leads were broken at the time of implantation and only two of the remaining 36 microelectrodes failed to produce phosphenes. Due to the untimely breakage of a number of microelectrode wires, planned pattern recognition studies could not be conducted.

The phosphenes produced by ICMS were similar to those reported in the Canadian study [3]. A summary of the results obtained with ICMS were:

1.Phosphenes never flickered.

2.Phosphenes moved with eye movement and a group of phosphenes maintained their relative positions with eye movement.

3.Stimulation of microelectrodes, with tips separated by 0.5 mm, produced separate phosphenes.

4.Phosphenes appeared immediately after the beginning of stimulation and except for rare occasions, disappeared at the termination of stimulation.

5.When stimulation continued beyond a second, phosphenes usually disappeared.

6.By interrupting a long stimulation pulse train with brief pauses, the duration of phosphene perceptions could be increased.

7.Multiple phosphenes were co-planar.