Cortical Activity Research Paper

The roots of optical imaging can be traced back at least 60 years, to the observations of Wilder Pennfield (1933) who noted minor changes in the appearance of exposed cortex during epileptic seizures. On account of associated vasospasms, the seizing regions looked alternately light and dark. Though such changes must be large indeed to be visible to the naked eye, they resemble smaller ones that accompany normal events. This wasn’t recognized at the time—indeed these changes were deemed pathological. But even if it had been, the technologies needed for using them to map normal activity would not become available for many years.

A few years after Pennfield’s initial observations, Kelin and Millikan found activity dependent changes in the optical properties of cytochromes. Hill and Keynes (1949) subsequently recorded the first activity dependent optical signals, in specially prepared nervous tissues. However, these and subsequent efforts soon were diverted by an emerging focus on single neurons. Hence, the possibility of using light to monitor slow changes in the activation of cortical tissue was neglected until Blasdel and Salama (1986) showed that slow, activity-driven changes in light reflectance can be used to visualize organizations that had intrigued neuroscientists for decades: the arrangement of ocular dominance and/orientation slabs in the primary visual cortex. Though their initial observations have been replicated many times, using a variety of slightly modified procedures, the fundamental strategy stays the same. All rely on the small, activity driven reflectance changes resembling those noted by Pennfield, all try to drive these alternately between two extremes, and all try to map associated reflectance changes with some differential imaging technique.

1. Primary Visual Cortex

It is not an accident that the first comprehensive optical maps were obtained from the primary visual cortex of macaques—an area ripe for this kind of analysis, after decades of brilliant discoveries showing the importance of ocular dominance and/orientation selectivity, and the likely organization of these proper-ties in slabs (see Fig. 1). In addition, the primary visual cortex of macaque monkeys is particularly well suited to optical investigation since most of it is remarkably large and flat and, due to its location on top of the occipital lobe, conveniently accessible. It also was not an accident that this was first achieved by visual neuroscientists, since many of the problems with differential imaging resemble those in vision— problems whose solutions are most familiar to those thinking about them full-time.

In order to appreciate the nature of the problem at hand, it is important to understand a few aspects of primate area V1. Light from the outside world is refracted onto the retina, where it directly forms the first of several representations including one in a layer of ganglion cells, whose axons form the optic nerve and project centrally to the lateral geniculate nucleus (LGN). Neurons in six separate but retinotopically aligned layers of the LGN receive this information and project directly to V1. The thing to keep in mind, therefore, is that V1 is not the first two-dimensional representation of visual space, nor does it reflect retinotopic arrangements in any simple way. Instead, it reflects the cumulative consequences of successive transforms, each one of which modifies the representation of space in simple but important ways.

These modifications lie beyond the scope of this review. However, they can be summarized as a tendency to reduce some form of redundancy with each and every step. The consequence of a center-surround receptive field in the retina or LGN, for example, is to minimize responses to diffuse or gradually changing light. Light that doesn’t change with distance is redundant and can be ignored. This makes it possible to concentrate on more important events, like those where light changes suddenly, associated with an edge.

The reduction in redundancy continues in the LGN and primary visual cortex (also known as the striate cortex, area 17, and V1), where redundant views from the two eyes are merged. Because the convergence is usually incomplete, there is a residual tendency for one eye to dominate in alternating 0.5 mm wide slabs. These are known as ‘ocular dominance slabs’ or ‘ocular dominance columns’ and are illustrated in Fig. 1a.

At the time information converges from both eyes to form a single binocular map, the cortex sees the emergence of a new response property—orientation selectivity—that is characterized by selective responses to particular orientations of edges. This property too can be characterized as a reduction in redundancy, since strings of discontinuous events that occur along a single axis in space can now be summarized by a single neuron. Like ocular dominance, this property is characterized by vertically aligned neurons, with those lying above one another responding only to similar orientations. What is particularly important about the vertical alignment of ocular dominance and/orientation columns, in regard to optical imaging, is that their depth invariance makes it possible to ignore the depth of light penetration, which is difficult to determine. In spite of their clear importance, ocular dominance and/orientation selectivity patterns had proven difficult to map.

2. Differential Video Imaging

A solution to this problem is provided by the small changes in reflectance—similar to those described by Pennfield (1933)—that accompany normal activity as well. Due to recent advances in the development of practical imaging technologies (e.g., cameras, image processors, computers, etc.) reflectance differences as small as 0.01 percent—approximately 500 times smaller than ones that can be seen by the unaided eye—can easily be seen (Blasdel 1992a, b). Accordingly, any area of cortex that can be exposed and activated physiologically, can be mapped with this technique.

Though a number of procedures have been de-scribed for doing this, all entail minor variations of the same technique. All exploit small, activity-dependent changes that cause active cortex to reflect less light; all seek to modulate them between two extremes, and all seek to average responses to increase ratios of signal to noise. The specific mechanisms that couple reflectance to activity are controversial and likely to reflect different contributions at different wavelengths. Since shorter wavelengths (600 nm) are strongly absorbed by hemoglobin, signals obtained with them are dominated by vascular events: changes in blood volume and possibly even the oxygenation of hemoglobin. Since longer wavelengths (720 nm) are absorbed much less strongly by hemoglobin, signals monitored with them are freer from vascular events and more likely to reflect differences in scattering due to changes in ion concentration, water movement, and the volumes of active cells; they are also less pronounced (for a review, see McLoughlin and Blasdel 1998).

The first apparatus used for differentially imaging these changes relies on a low power microscope (Blasdel 1992a) that uses co-axial illumination to irradiate the cortex with monochromatic light (720 nm ±20 nm). Reflected light is collected by an objective that relays it to an imaging device—usually a TV or CCD camera—that transduces it into a video signal that is digitized and recorded by an image processor that averages thousands of frames for each response to improve ratios of signal to noise.

The point of differential imaging is to see changes that are too small to be seen by themselves. This is especially true in the case of cortical responses, where the optical changes associated with orientation preference or ocular dominance may be a hundred times smaller than those associated with cortical activation alone. By stimulating the cortex twice, however, with stimuli that activate it to approximately the same degree, it is possible to modulate a single variable and visualize the small reflectance changes that develop in phase. In the case of ocular dominance patterns, for example, the responses to either eye alone may be huge. Yet because common mode changes to each eye are similar, a differential subtraction (i.e., images achieved during stimulation of one eye are subtracted stimulating both eyes alternately with orthogonal orientations (in this case, vertical and horizontal) and differentially imaging the changes that emerge. (e) Differential images of retinotopic space, obtained by stimulating complementary stripe like compartments on alternate trials and differentially imaging the changes. Due to the retinotopic organization of cortex, the stimulation of vertical stripes in space induces vertical bands (that run parallel to border, represented by arrow), while (f) The stimulation of horizontal stripes in spaces produces bands that intersect the V1 border at right angles from images averaged during stimulation of the other) can remove these common mode events and reveal differences specific to the two eyes.

Once the cortex has been exposed (by removing the scalp, cranium, and overlying meninges) and immobilized by a thin plastic or glass plate, it is possible to see the surface of the cortex (Fig. 2a) and differentially image activity. While the dense covering of superficial blood vessels might seem to present a problem, they become virtually transparent at 720 nm. Due to the weak absorption of 720 nm light by hemoglobin, the same blood vessels that dominate Fig. 2a become virtually transparent in Fig. 2b. This is the image seen by the camera imaging all responses that are compared.

Differential imaging works by comparing two images of the cortex (see Figs. 2c–2f ) in slightly different states that under ideal circumstances differ with respect to only one variable. If done carefully, the resulting difference image reveals a pattern specific to the variable. By subtracting images of the cortex responding to the left eye from images of it responding to the right eye, for example, one obtains an image of ocular dominance columns, as illustrated in Fig. 2c. By subtracting images of the cortex responding to horizontal edges from images of it responding to vertical edges, one obtains a pattern associated with orientation selectivity, where dark and light regions rep-resent selectivity for vertical and horizontal (Fig. 2d). Similarly, by differentially imaging responses to complementary parts of space, it is possible to visualize their cortical representations. The vertical stripes in Fig. 2e, for example, elicited bands running parallel to the V1/V2 border (which represents vertical), while horizontal stripes (Fig. 2f ) induced bands that intersect the V1/V2 border at right angles.

Fig. 3 shows 8 differential images (a–h) of orientation, obtained by comparing responses to each of two perpendicular gratings, with the orientation of each pair rotated by 22.5 in successive frames. As one can see, the small ‘x’ placed at the center of a vertical iso-orientation band lies at the center of a dark patch in Fig. 3a, but moves gradually to the side in successive frames until it lies over a white patch in Fig. 3e. From the relative displacements of light and dark regions relative to this x, it is possible to see that these patterns depend only on the orientations of stimuli used to induce responses that are compared.

3. Orientation Preference Patterns

As one might note from the images in Fig. 3, the gray values at various locations reveal the degree to which each of the two opposed orientations (indicated in the lower right hand corner of each frame) was effective at driving responses. They do not indicate which orientation was optimal, however, since this information can only be inferred by comparing responses to many orientation pairs. Because this information appears distributed across all the images in Fig. 3a–3h, it is easiest to extract from all of them at once, by converting the pixel values in each image into vectors (which use phase to keep track of the angle used), that can be added to reveal the distribution of preferences and selectivities at all locations (see Blasdel 1992a for more details). The outputs of this operation are illustrated in Fig. 3j, where each short line indicates the degree of selectivity and preferred orientation, through its length and/orientation. As one can 43у, some of these are quite short, indicating no selectivity at all; while many are quite long, indicating sequences of orientation preference that rotate gradually in swirls, from side to side. In order to estimate the shapes of orientation ‘columns,’ it is possible to calculate con-tours along which orientation preferences do not change. The resulting iso-orientation contours (Obermayer and Blasdel 1993), which resemble Hubel and Wiesel’s (1974) iso-orientation slabs, are illustrated for the same region of cortex in Fig. 3k.

As one can see, this pattern differs from that expected (see Fig. 2c). Rather than appearing in long arrays of infinitely long columns, iso-orientation slabs appear limited to 0.5–1.0 mm in length and to sequences that are no more than 0.5–1.0 mm across. This wasn’t known before, and has become apparent only on account of optically imaged patterns, with results from all laboratories showing approximately the same thing (see also T’so et al. 1990, Bartfeldt and Grinvald 1992). A particularly surprising phenomenon concerns orientation singularities that occur at regular intervals in the centers of ocular dominance columns where iso-orientation contours converge (Blasdel 1992b; Bonhoeffer and Grinvald 1993). One example appears highlighted by a white circle in Fig. 3k. These are the main events that disrupt iso-orientation contours, and prevent them from getting longer. Nevertheless, iso-orientation contours seem to congregate in equally regular linear zones that subtend 0.5–1.0 mm and predominate near ocular dominance borders. One example is outlined by a white rectangle in Fig. 3k. Within these zones, iso-orientation contours appear particularly well organized in linear arrays of slabs, more or less as Hubel and Wiesel (1974) predicted (see Fig. 1c). Note, that within these zones, iso-orientation slabs also have a striking tendency to cross ocular dominance borders at angles of approximately 90°—more or less as suggested by Hubel and Wiesel’s model, in Fig. 1c.

From this relatively simple analysis of optical imaging maps, it is possible to conclude that orientation preferences organize according to one of at least two competing schemes—a linear one, confined to 0.5–1.0 mm patches, where orientation preferences change linearly along one axis while remaining constant along the other; and a discontinuous, circularly symmetric one, that entails continuous changes in preferred orientation around a point singularity. The linear scheme predominates in two-dimensional patches, 0.5–1.0 mm across, while the pinwheel singularities that disrupt them lie somewhere in between. Also, these results make it clear that within the linear zones iso-orientation contours and ocular dominance borders cross at right angles.

The true strength of optical imaging is apparent from the comparisons that it enables. Since it only takes a few minutes to obtain each pattern with a technique that appears to be relatively benign, it is possible to collect many different types of response profile from the same cortical regions. This is what makes it possible to calculate maps of orientation preference, for example, where responses to many different orientations need to be compared. It also is what enables comparisons between ocular dominance and/orientation maps from the same regions, allowing them to be related precisely for the first time. By comparing these with maps of retinotopic representation from the same region (see Figs. 2e, 2f ) it is possible to obtain unprecedented insights into cortical organizations.

Bibliography:

1. Bartfeld E, Grinvald A 1992 Relationships between orientation-preference pinwheels, cytochrome oxidase blobs, and ocular-dominance columns in primate striate cortex. Proceedings of the National Academy of Science USA 89: 11905–9
2. Blasdel G G 1992a Differential imaging of ocular dominance and/orientation selectivity in monkey striate cortex. Journal of Neuroscience 12(8): 3115–38
3. Blasdel G G 1992b Orientation selectivity, preference, and continuity in monkey striate cortex. Journal of Neuroscience 12(8): 3139–61
4. Blasdel G G Campbell D in press Functional retinotopy of monkey visual cortex. Journal of Neuroscience
5. Blasdel G G, Salama G 1986 Voltage-sensitive dyes reveal a modular organization in monkey striate cortex. Nature 321: 579–85
6. Bonhoeffer T, Grinvald A 1993 The layout of iso-orientation domains in area 18 of cat visual cortex: Optical imaging reveals a pinwheel-like organization. Journal of Neuroscience 13(10): 4157–76
7. Grinvald A, Lieke E, Frostig R P, Gilbert C, Wiesel R M 1986 Functional architecture of cortex revealed by optical imaging of intrinsic signals. Nature 324: 351–64
8. Hill D K, Keynes R D 1949 Opacity changes in stimulated nerve. Journal of Neuroscience 7: 3378–415
9. Hubel D H, Freeman D C 1977 Short communications: projection into the visual field of ocular dominance columns in macaque monkey. Brain Research 122: 336–43
10. Hubel D H, Wiesel T N 1972 Laminar and columnar distribution of geniculo-cortical fibers in the macaque monkey. Journal of Comparative Neurology 146: 421–450
11. Hubel D H, Wiesel T N 1974 Sequence regularity and geometry of orientation columns in the monkey striate cortex. Journal of Comparative Neurology 158: 267–93
12. Hubel D H, Wiesel T N 1977 Ferrier lecture. Functional architecture of macaque monkey visual cortex. Proceedings of the Royal Society of London Series B Biological Science 198(1130): 1–59
13. Hubel D H, Wiesel T N, Stryker M P 1978 Anatomical demonstration of orientation columns in macaque monkey. Journal of Comparative Neurology 177: 361–80
14. LeVay S, Hubel D H, Wiesel T N 1975 The pattern of ocular dominance columns in macaque usual cortex revealed by a reduced silver stain. Journal of Comparative Neurology 159(4): 559–76
15. McLoughlin N P, Blasdel G G 1998 Wavelength-dependent differences between optically determined functional maps from macaque striate cortex. Neuroimage 7: 326–36
16. Obermayer K, Blasdel G G 1993 Geometry of orientation and ocular dominance columns in monkey striate cortex. Journal of Neuroscience 13: 4114–29
17. Penfield W 1933 The evidence for a cerebral vascular mechanism in epilepsy. Annals of Internal Medicine 7: 303–10
18. Schiller P H, Finlay B L, Volman S F 1976 Quantitative studies of single-cell properties in monkey striate cortex. III. Spatial frequency. Journal of Neurophysiology 39: 1334–51
19. Ts’O D Y, Frostig R D, Lieke E E 1990 Functional organization of primate visual cortex revealed by high resolution optical imaging. Science 249: 417–20
20. Wilson H R, Levi D, Maffei L, Rovamo J, DeValois R L 1989 The perception of form: retina to striate cortex. In: Spillman L, Werner J S (eds.) Visual perception: The Neurophysiological Foundations, Academic Press, New York, pp. 231–72