device [4–6]. The images were captured by a head mounted video camera covered by a perforated mask. They concluded that, with 625 dots (25 × 25 array) in a visual field of 1.7°, a visual acuity of 20/30 and a reading speed of 100 words of paragraph text per minute could be achieved. They found that 625 or more dots, with a field view of about 30°, allowed normal walking speed through a maze with obstacles.

For these tests, they used identically-sized simulated phosphenes on regularlyspaced grids. Cortical stimulation experiments, however, have shown that, for a cortical prosthesis, a regular grid structure is not a true representation of the phosphene map. The maps will vary depending upon the area in the visual cortex used for implanting the electrodes and the type of electrodes used. For conducting proper simulation studies, simulated phosphene maps have to correspond to the map generated as per the targeted electrode location and the type of electrode used.

Some groups might target the medial surface for implantation, and a few groups might target the lateral surface or a combination of electrodes on the lateral and the medial surface. Every research group that is targeting the cortex for electrode implantation will have to generate an estimated percept map depending on their choices of electrodes and array location, and use this map to guide expectations for this device’s performance. To generate this map, the representation of visual space on the cortex, cortical structure, and the corresponding biological responses have to be understood.

18.2  Representation of Visual Space on the Visual Cortex

One of the first published visual maps by Holmes shows representation of different visual fields on calcarine cortex, with a linear relationship of visual space to the cortex [12]. These maps were later modified by Horton and Hoyt [13]. The modified map shows the horizontal meridian running at the base of the calcarine fissure with iso-eccentricity contours from 2.5° to 40°. This map shows that the space follows a logarithmic representation till 40° of eccentricity, with the foveal area of visual field represented on a larger area of cortex. Recent experiments have supported these modified maps [17, 18]. This logarithmic representation of visual space on the visual cortex is known as cortical magnification.

In an fMRI study by DeYoe and colleagues, a consistent retinotopic organization was observed on responsive visual cortex both medially and ventrally [7]. The foveal representation was located posteriorly, near the pole, and greater eccentricities were represented anteriorly on the surface. Responses observed to visually expanding checkered rings extended from the collateral sulcus on the ventral surface, crossed the calcarine fissure and passed dorsally out onto the exposed lateral surface. The responses did demonstrate cortical magnification. The data also showed that as the eccentricity increased, there was an anterior progression of activation along with alteration of visual field meridian at transfer from one visual area to other. In another fMRI study by Levy and colleagues, it was observed that the visual cortex has a hierarchical organization that begins with the precise visual field

maps in V1, V2 and V3, which degrades on the lateral and ventral regions [16]. The lateral and ventral regions contain coarse eccentricity maps with crude representations of the polar angle. It has been observed that visual space is represented in many visual clusters. Wandell and colleagues reported observing nine human visual field map representations [22]. These observed visual field maps preserved spatial structure.

Any electrode implantation will have to consider the logarithmic nature of these representations. Electrodes implanted near the occipital pole will cover a very small visual area corresponding to a few degrees of eccentricity, and phosphenes corresponding to electrodes on the lateral surface may be perceived more eccentrically, but will lose angular specificity. If the electrodes lie in an area with multiple cluster maps, then the position of the phosphenes in visual space will be almost random. Any psychophysical study designed to estimate the performance of a cortical prosthesis device will have to consider these factors.

18.3  Cortical Stimulation Studies

Brindley and Lewin placed eighty 0.64 mm2-platinum electrodes between the medial surface of the occipital pole of the right cerebral hemisphere and the falx cerebri of a 52-year-old blind female patient [1–3]. They compared the observed phosphene maps with the Holmes map and found correlations between the two. This experiment was performed in 1968, when the revised map by Horton and Hoyt was not available. If the map published by Brindley and Lewin is compared with the map published by Horton and Hoyt, it is observed that the phosphene map does show cortical magnification. The phosphenes at the periphery of the Brindley and Lewin phosphene map were larger in size, which might be result of cortical magnification. A few irregularities were observed, such as an electrode placed close to a certain group of electrodes produced a phosphene away from the phosphenes corresponding to the group.

Dobelle and Mladejovsky published phosphene maps generated from surface electrodes placed on the right medial surface of a patient [8, 9]. These maps show discrepancies from the expected responses when compared to a logarithmic visual space map, both in terms of eccentricity of phosphenes observed, and polar angles expected from published visual maps.

Dobelle and colleagues published another set of results with 64 platinum disk surface electrodes implanted 3 mm apart on the medial surface of right occipital lobe [10]. The phosphenes followed the expected cortical magnification on visual space. A set of electrodes placed in a line close to calcarine fissure was expected to define the horizontal meridian; instead, it was almost perpendicular to it, close to the vertical meridian. It was hypothesized that the electrodes crossed into the V2 area hence showed the mirror image of the expected response from V1. This shows that the phosphenes might be produced by stimulating higher areas of extrastriate cortex. On the lateral surface, the boundaries of V1, V2 and V3 are not clearly defined.

Even with a very small area of V1 exposed on the lateral surface, though, studies have still shown the generation of phosphenes [14, 15, 19].

Schmidt and colleagues stimulated the lateral surface of visual cortex using intracortical electrodes [19]. The electrodes were placed approximately up to 22 mm away from the occipital pole on the lateral surface. Most of the phosphenes mapped were within 30° eccentricity, with few phosphenes up to 40° eccentricity. Few of the phosphenes were observed at the eccentricities and polar angles expected when compared to the visual maps discussed earlier, and a few phosphenes did not exhibit either the expected eccentricity or the polar angle.

Lee and colleagues stimulated the occipital cortex and the adjacent cortices using surface electrodes in 23 epilepsy patients [15]. The experiment shows that as electrodes are implanted away from the occipital pole, more anteriorly towards the frontal cortex, the response to stimulation varies from simple form phosphenes, to intermediate responses like triangles and diamond shapes, to complex responses like observations of color and evoking movement percepts. The initial 1–2 cm (approximate) from the occipital pole show simple responses, and 2–3 cm (approximate) show intermediate responses. If electrodes are placed on the lateral surface, they should be limited to a distance of 3 cm from the occipital pole.

Kaido and colleagues investigated retinotopic maps on the lateral surface of the occipital cortex in humans [14]. The researchers observed phosphenes of up to 80° eccentricity, stimulating up to 40 mm anterior to the occipital pole, on X-ray scale, which shows that the whole lateral surface of occipital cortex might generate phosphenes. Polar angles were preserved in a coarse manner. If electrodes are implanted too far anterior from the occipital pole, however, complex forms might be observed, as in stimulation studies by Lee and colleagues. [15].

18.4  Variability in Occipital Cortex

Stensaas and colleagues studied primary visual cortex of 52 hemispheres and found the average total area to be 2,134 mm2 [21]. The average striate cortex exposed on all four surfaces was 689 mm2, about 33% of striate cortex, and the other 67% of striate cortex (average 1,445 mm2) was buried in fissures [21]. For inter-electrode distances of 3 mm, we might be able to place only 60–80 electrodes on V1 [1–3, 8–10, 21]. It was found that, on average, only 3% (55 mm2) of primary human visual cortex extends to the occipital surface of the brain. For the average exposed V1 area of 55 mm2, if intracortical electrodes are used with inter-electrode distances of 0.5 mm, it might be possible to place 200 electrodes [19]. The variability in the visual cortex within individuals will affect cortical implantation. If surgical methods are developed to implant electrodes over the complete exposed striate cortex, then the implanted electrode numbers might vary by 30%.

Dougherty and colleagues used data from fMRI and prepared 2-D flattened representations of the cortical manifold [11]. Using these flattened maps they

Fig. 18.2  The visual map expected with placement of electrodes on the medial wall of the V1, V2 and V3 areas. Note the additional phosphenes, mostly in the lateral visual field, as compared with Fig. 18.1; V2 and V3 phosphenes are largely intermixed

reflecting the distortion due to cortical magnification [1–3, 8–10]. The gap shown on the lateral visual field corresponds to the two-thirds of area V1 that lies within the calcarine fissure. This area is inaccessible with existing surgical techniques for both surface and the intracortical electrodes. It might be possible to generate phosphenes on the lateral visual field by stimulating V2 and V3, which have representations similar to that of V1. If electrodes are implanted on V2 and V3, along with V1, the mirror image correspondence of V1 to V2 and V2 to V3 might help to generate phosphenes on the lateral surface as shown in Fig. 18.2.

Schmidt and Kaido have shown the generation of phosphenes over a wide region of visual space, while stimulating the lateral surface, and concluded that this area can be used to create phosphenes for a cortical visual prosthesis [14, 19]. From the study of Lee and colleagues we observe that the limitation of area for electrode placement area on the lateral surface area is about 3 mm from the occipital pole to get a simple percept [15]. The fMRI study by DeYoe shows that for a 3 cm radius from the occipital pole, phosphenes might be observed throughout the central 25° of the visual field [7]. The expected phosphene map is represented in Fig. 18.3. This figure is derived from an fMRI study, but as observed with intracortical electrode experiments, few phosphenes might be observed in 40°–45° [19]. If it is assumed that 50% of phosphenes from this hypothesized map are between 25° and 45° eccentricity, this will give us a map in which 50% dots of phosphenes are dropped from the initial 25° of the eccentricity map in Fig. 18.3, and redistributed across 25°–45° eccentricity, giving us a map as shown in Fig. 18.4.