Visual information is channelled into three distinct parallel pathways at the level o f the retina. The parvocellular pathway (P-pathway) carries cone-specific information regarding colour and high-resolution vision. The magnocellular pathway (M-pathway), in contrast, receives inputs from both cone and rods, and is concerned with achromatic and low-resolution vision, as well as phasic responses. The konicocellular pathway (K-pathway) is thought to process
Chapter 2 The Retina
information from the S-cones (Casagrande, 1994; Hendry & Reid, 2000). Figure 2.7 illustrates the route of these pathways from retina to primary visual cortex.
eye optic nerve optic chiasm optic tract lateral geniculate nucleus optic radiation primary visual cortex
Figure 2.7 Schematic diagram of the visual pathway (From Hubei, 1988).
Upon leaving the retina, ganglion cell axons join to form the optic nerve - a bundle of between 1 and 2.2 million visual fibres (Jonas et al, 1992). The information from the right and left eyes is kept separate until the optic chiasm. Here the fibres from the nasal retina of both eyes decussate, switching into the optic tract on the opposite side of the brain. This crossing ensures that information from the left side of the visual field is carried to the right side of the brain, and vice versa.
The visual fibres continue without synapsing along the optic tracts to reach the lateral geniculate nucleus (LGN). The LGN is responsible for preparing visual information to be passed on to the visual cortex. It is composed of 6 layers, two of which contain large cells which process magnocellular information, with the remaining 4 layers being responsible for the parvocellular pathway. The tonic midget ganglion cells which predominate in the central retina project to the parvocellular layers of the LGN, whilst the phasic large parasol ganglion cells project to the magnocellular layers (Rodieck, 1985). Cells situated between these layers receive information from the K- pathway (Casagrande, 1994). The structure of the LGN may be seen in the histological section shown in Figure 2.8. The lateral geniculate nucleus not only relays the visual information on to higher cortical centres, but also receives input from
Chapter 2 The Retina
cortical and subcortical centres of the brain, and reciprocal information from the visual cortex, which allows a modification of the signal (Lachica, 1993).
magno
cellular
parvo
cellular
konio
cel
Figure 2.8. Histological section of the LGN showing the parvocellular, magnocellular, and konicocellular regions {from Gouras, 2001).
Following the first extra-retinal synapse in the LGN, information is conveyed via the optic radiations to the striate cortex. The optic radiations are a group of fibres, which spread out as they pass laterally and inferiorly through the white matter of the cerebral cortex. They pass through the temporal and parietal lobes of the brain before converging on the primary visual cortex in the occipital lobe (Warwick, 1976).
The primary visual cortex (also known as the striate cortex, or VI) is organised into a number of layers. Six layers were originally identified and labelled by Brodmann, but these have since been subdivided. The structure of VI has also been categorised according to cytochrome oxidase (CO) staining, and has been found to consist of staining areas, known as ‘blobs’, and non-staining areas called ‘interblobs’. Increased cytochrome oxidase staining is indicative of high cellular activity. Cells within the blobs appear in general to have strong colour sensitivity, but poor orientation selectivity {Regan, 1989). Blobs receive input from the
Chapter 2 The Retina
magno-, parvo- and konicocellular pathways. In the extrastriate region V2 o f the visual cortex, cytochrome oxidase staining shows thin and thick stripes, and pale interstripes.
The 3 parallel pathways for vision each project to different areas and layers o f VI {Hubei &
Wiesel, 1972). Information from the P- and M- layers of the LGN are projected to layer 4 o f
the primary visual cortex, which is divided into sublayers 4A, 4B, 4Ca, and 4Cp. Signals from the P-pathway project to layer 4Cf3, and from the M- pathway project to 4Ca. From here, parvocellular information is relayed to the interblobs o f layers 2 and 3 o f VI, via layer 4A. then to the interstripes o f extrastriate region V2. The magnocellular processing pathway proceeds from 4Ca to layer 4B, from where information is passed both to the interblobs o f layers 2 and 3 o f VI, where it converges with the P-pathway, and to extrastriate regions MT, V2 and V3 {Schmolesky, 2000; Regan; 1989). The K-pathway bypasses layer 4, and projects directly to the blobs o f layer 3, from where information is passed to the thin stripes o f V2
{Ding, 1997). Figure 2.9 illustrates a schematic diagram o f the connections between the LGN
and layers o f the striate cortex.
(Thick (Inter
stripes) stripes) SC (Thin stripes) V2 Pulvinar MT V3 V2 V2 VI 4A 4B 4 C a 4C p 6B 6A Blobs Blobs 2 + 3 interblobs M M+P P
LGN Magnocellular Parvocellular Konicocellular
Figure 2.9. The hierarchical arrangement o f cortical layers, showing the input from LGN to VI, and the feedback and feedforward pathways. Small arrows indicate minor pathways, large arrows indicate major pathways {From Regan et al, 1989).
Chapter 2 The Retina
have spiny dendrites. Pyramidal cells have been found in all layers except 1 and 4, whilst stellate cells are located throughout the layers o f the striate cortex, and it is with these neurones that axons from the LGN synapse. The smooth or sparsly spinous intemeurones mainly form inhibitory GABAergic synapses, and have few or no spikes on their dendrites. Layers 2 and 3 o f VI have numerous projections to the extrastriate cortical regions (V2, V3, V4, MT) where further processing occurs, and feedback from these regions is processed in layers 1 and 2. VI layers 5 and 6 are responsible for providing feedback to subcortical regions such as the thalamus, midbrain, and pons (Schmolesky, 2000). Feedback to the superior colliculus, for example, is important for visual orientation, foveation, and control o f saccadic eye movements {Horton, 1992). The feedback from layer 6 to the LGN is important in providing a pathway for the striate cortex to modulate its own input {Casagrande & Ichida,
1989).
In addition to the input to the primary visual cortex from the LGN, and feedback from higher visual processing regions, further information is received by VI from other areas including the brainstem, basal forebrain nuclei, thalamus, and pulvinar {Casagrande & Ichida, 2003). The neurones o f VI exhibit a retinotopic organisation reflecting the location o f their receptive fields within visual space. The area o f the cortex devoted to each part o f the retina is determined by the size o f receptive fields in that area. The central retina therefore has a relatively high cortical representation as a result of the higher density of ganglion cells in this region {Horton, 1991).
The primary visual cortex also has a columnar organisation {Hubei & Wiesel, 1972) with neurones organised according to function and structure {See Figure 2.10). All neurones within an ocular dominance column, for example, will be innervated to the same extent by the right and left eyes {Levay et al, 1980). Similarly, cells within an orientation column will respond maximally to stimuli o f the same orientation. There is an ordered progression in orientation preference as the columns are transversed {Hubei et al, 1977). The visual cortex may be construed as having an ‘ice-cube’ structure consisting o f functional modules, within each are cells which have every possible combination o f orientation, dominance and colour selectivity. The ocular dominance columns o f layer 4C are innervated completely by either the right or the left eye. These monocular outputs later converge such that other layers are binocular, and innervated to varying extents by each eye {Horton, 1992).
Chapter 2 The Retina
Figure 2.10. Schematic diagram of ‘ice-cube’ organisation of the primary visual cortex. The white ovals represent the ‘blobs’. The coloured ‘pinwheels’ indicate orientation columns, with each colour representing sensitivity to a specific orientation. The light and dark grey columns separated by black lines represent the ocular dominance columns. The upper slice consists of two functional units, each containing neurones sensitive to all combinations of orientation, ocular dominance and colour coding (From Van Gisbergen, 2004).
The primary visual cortex is just the first site of analysis of the visual information. From here the processed neural signals are distributed to more than 30 other specialised visual association areas where further interpretation occurs (Horton, 1992). Visual areas and visual association areas in one hemisphere are connected to the corresponding regions of the opposite hemisphere via the posterior portion of the corpus callosum {Regan, 1989).
The extrastriate visual cortex receives nearly all visual information from VI (except for a minor stream directly from the LGN and pulvinar, which may be responsible for blindsight)
(Cowey & Stoerig, 1991). The main extrastriate projections of VI are to V2, V3, V4, and
V5/MT. The inputs to these different regions arise from different groups of neurones in VI
{see Figure 2.9). As described above, area V2 shows patterns of cytochrome oxidase staining,
which are seen as thin stripes, thick stripes, and interstripes. The input to the thick stripes is dominated fairly directly by information from the magnocellular pathway, whilst the input to the thin stripes and interstripes appears to be dominated by parvocellular information, through a more indirect pathway. The majority of cells within the thin stripes lack orientation selectivity, and show colour opponent receptive fields, whilst those within the interstripes tend to be orientation selective, but lack colour coding. In comparison, the receptive fields of cells within the thick stripes predominantly show orientation selectivity, and sensitivity to
Chapter 2 The Retina
retinal disparity, a necessity for depth perception. Within the stripes and interstripes a columnar organisation is continued for orientation, disparity or colour opponency, depending on the nature o f cells within the stripe (Boyd & Matsurbara, 2003).
Projections from the thin stripes and interstripes then carry information to the cortical area V4, involved in form processing. Thick stripes project to the motion processing centre MT, as well as sending information to the subcortical superior colliculus. The retinal disparity sensitivity o f these cells is likely to play a role in the generation o f eye movements.
A retinotopic organisation continues at the level of V2, with a regular progression in the location o f the receptive fields o f cells as you move across each stripe. However, there is actually a complete visual field representation for each stripe type such that the visual map of V2 consists of 3 interleaved maps (Roe & T s’o, 1995).
Visual area V3 is divided into dorsal and ventral halves, which represent the lower and upper quadrants o f the visual field respectively. These are termed dorsal V3 and the ventral posterior area (VP). Only dorsal V3 receives direct projections from VI, originating in the magnocellular dominated layer 4B. However, both VP and dorsal V3 form connections with V2, MT, and the parietal lobe (Felleman et al, 1997). In VP most cells are sensitive to colour and orientation, but not direction, whilst those cells in dorsal V3 tend to be less colour selective, and more direction selective (Boyd & Matsubara, 2003).
The other extrastriate region o f the visual cortex to receive a primarily magnocellular input is MT. MT is a heavily myelinated area, which represents the complete contralateral visual field. It receives a strong magnocellular input from Layers 4B and 6 o f VI, and the thick stripes o f V2 (Movshon & Newsome, 1996). Area MT has a columnar organisation of cells according to direction and retinal disparity tuning. Cells in MT are also arranged in columns according to either wide-field or local motion contrast sensitivity (Albright et al, 1984). Local motion contrast is achieved by antagonistic receptive field surrounds, and allows detection o f motion boundaries. Wide-field motion contrast provides information regarding global motion, allowing orientation o f an individual within their environment. Information from V3 and MT is relayed to the temporal and parietal lobe areas (Boyd & Matsubara, 2003).
Area V4 is located between ventral V3 and MT. It receives projections from VI and V2, and has inputs from both magno- and parvocellular pathways. Projections from VI originate in layers 2 and 3, in blobs and interblobs. Both thin stripes and interstripes o f V2 project to V4, but input from the two is kept segregated. Cells in V4 are sensitive to parameters which might
Chapter 2 The Retina
be involved in object recognition e.g. colour, orientation, size and binocular disparity. This information is then relayed to the inferotemporal areas o f the temporal lobe, which also receive input directly from V2. The inferotemporal cortex contains areas which are sensitive to combinations o f stimulus features. For example, some neurones will respond selectively to faces. There is a columnar organisation o f cells, with cells within a column responding to similar features (Wang et al, 1998).
Chapter 3 Age-Related M acular Degeneration
CHAPTER 3 : AGE-RELATED MACULAR DEGENERATION
3.1 Introduction
Age-related macular degeneration (AMD) is the leading cause o f blindness in the Western World (Leibowitz et al, 1980; Klein et al, 1995; 1999). It is a condition which ravages the central retina and can have a devastating effect on an individual's lifestyle and independence. Age-related macular degeneration was first described by Haab in 1888, although according to Verhoeff and Grossman (7955), the clinical features may have been described as early as 1875 by Pagenstecher. Over the past century, our understanding o f the condition has improved, but is still far from complete, and treatments remain limited in their efficacy.
Age-related maculopathy (ARM) is the early form of the disease, characterised by soft indistinct or distinct drusen and RPE abnormalities, but with limited visual function deficits
(Bird et al, 1995; Mitchell et al, 1995; Sarraf et al, 1999). This may then develop into either
nonexudative (dry) or exudative (wet or neovascular) AMD. Dry AMD is characterised by the presence o f drusen and atrophy o f the retinal pigment epithelium. It is associated with a slow decline in visual acuity. Exudative AMD starts with the growth o f new, fragile vessels from the choroidal circulation, which penetrate through breaks in Bruch’s membrane to proliferate beneath the retina. In exudative AMD vision may be lost more rapidly, over a larger area, and often at an earlier age than in dry AMD, due to serous or haemorrhagic exudates from the choroidal neovascular vessels. Although far less common than the nonexudative form of the disease, exudative AMD is responsible for eighty-nine percent o f cases o f legal blindness attributed to AMD (Ferris, 1983; Ferris et al, 1984).