Motion Perception in Glaucoma Patients: A Review

Magnocellular and parvocellular subcortical visual pathways 

The primate retina contains 20 or more different types of ganglion cells.55 The in vitro preparation of primate retina shows three types of ganglion cells that project to the lateral geniculate nucleus (LGN): the parasol (Pα), midget (Pβ), and small bistratified cells.16, 77 The morphology, physiology, and central projections of these cells have been well studied in primate retina.20, 111, 112 The parasol ganglion cells have a larger dendritic field and cell body size than the midget ganglion cells at any given eccentricity. The small bistratified ganglion cells tend to have larger dendritic field diameters than the midget cells and smaller than the parasol cells.112 For all cells, the retinal ganglion cell size correlates with the axon diameter and eccentricity.77, 78

Perry et al78 used retrograde horseradish peroxidase (HRP) labeling to illustrate that the large majority of Pα cells in the retina project to the magnocellular layers of the LGN, whereas the large majority of Pβ cells in the retina project to the parvocellular layers of the LGN. Thus, it has become a common practice to refer to the retinal ganglion cells by their putative connections to the LGN, namely, parasol cells as magnocellular (M) cells and midget cells as parvocellular (P) cells.2, 39, 74 For the sake of consistency we will adopt this practice and refer to the parasol and midget ganglion cells as M and P cells, respectively, although we note that it is possible that not all parasol cells and midget cells make these connections.

There is some evidence that a third class of cells, the small bistratified cells,18, 19 project to the koniocellular (K) layers of the LGN, in parallel with the parasol and midget cells.18, 66 The bottom part of Fig. 1 summarizes the current view of ganglion cell projections to the LGN. This picture should not be considered complete, because the details of most other ganglion cell type are still not known.

View Large Image Download to PowerPoint

Fig. 1. A simplified view of visual pathways presented in a schematic form from retinal ganglion cells to the higher levels of cerebral cortex. The projection of the M cells in the retina to the M layers of the LGN forms the magnocellular pathway and similarly the projection of P cells in the retina to the P layers of the LGN forms the parvocellular pathway. The axons of the magnocellular and parvocellular layers of LGN project to 4Cα and 4Cβ of the primary visual cortex. Recently recognised small bistratified cells project to koniocellular layers of LGN. The dorsal (parietal) visual pathway proceeds from primary visual cortex and end in the parietal cortex and the ventral (temporal) visual pathway also proceeds from the primary visual cortex and ends in the temporal cortex. Note that inspite of an early segregation of the M and P projections, the parietal and temporal pathways present complex pattern of connections. M Cells (magnocellular ganglion cells); P Cells (parvocellular ganglion cells); LGN (lateral geniculate nucleus); Magno (magnocellular layer); Parvo (parvocellular layers); Konio (koniocellular layers); V1 (primary or striate visual cortex); V2, V4 and MT (middle temporal area) are cortical (extra striate) areas. (Modified from Merigan and Maunsel.72)

There are considerable differences in the physiological properties of the two pathways. The M cells have higher contrast sensitivity than P cells.52, 79 The conduction velocity of the M pathway is faster than the P pathway.30, 65, 96 The time course of the response in the M pathway tends to be transient, whereas it is more sustained in the P pathway.23, 45, 96 The M cells are sensitive to high temporal and low spatial frequency stimuli, and P cells are sensitive to low temporal and high spatial frequencies.24, 45 The M pathway responds to broad-band light of different wavelength, whereas the P pathway mediates visual acuity71 and color.24, 69, 94 Recent research indicates that blue-yellow color is mediated by the small-bistratified ganglion cells and red-green color by the P ganglion cells.17

This functional segregation between the M and P pathways has been illustrated by behavioral studies of monkeys with ablated sections of LGN. Schiller et al showed that lesions of the P layers resulted in marked decreases in spatial contrast sensitivity; fine stereopsis; and pattern, color, texture, and shape discrimination.94, 95 These functions were unaffected by lesions of the M layers. In comparison, lesions of the M layers caused specific decreases in flicker (especially at high temporal frequencies) and motion sensitivity. Interestingly, low spatial frequency stereopsis, brightness discrimination and some aspects of shape discrimination were unaffected by either P or M lesions, but they were lost if both types of layers were lesioned, suggesting that these functions were carried by both M and P layers. Another study68 showed that M lesions resulted in a reduced sensitivity to high temporal frequency contrast flicker, but that the critical fusion frequency and contrast sensitivity for low temporal frequency and moderate spatial frequency stimuli were unaffected by such lesions.

Dorsal and ventral cortical visual pathways 

The axons from the M layers of the LGN project to the lamina 4C-α of the primary (striate) visual cortex (V1), while the axons from P layers project to layer 4C-β.78 Thus, the M and P visual pathways from the retina to the LGN remain clearly separated up to the level of V1. Subsequently, at the level of cortical processing, the segregation is only partial.72 However, two pathways dominate the complex network of connections in the visual cortex: the dorsal and ventral visual pathways (Fig. 1).105, 106 The dorsal visual pathway proceeds from V1 via the thick stripe regions of V2, mainly to the middle temporal (MT) area (specialized in motion processing) and to areas in the posterior parietal cortex. It receives a major contribution from the M pathway.72 The ventral visual pathway proceeds from V1, via the thin and interstripe regions of V2, mainly to V4 (specialized in color processing) and to areas of the inferotemporal (IT) cortex. It receives a major contribution from the P pathway.

The dorsal visual pathway plays a prominent role in motion perception, spatial localization and sensory motor coordination,40 whereas the ventral visual pathway is involved in the visual identification of colors, patterns, or objects.105

In spite of this gross functional specialization, the dorsal and ventral visual pathways should not be considered as a simple prolongation of the M and P subcortical streams. For instance, there is evidence of a moderate P input to cortical area MT.68 In addition, some aspects of visual motion perception are unaffected by M lesions of the LGN.70

Fig. 1 summarizes the main current knowledge about the neural transmission in the visual system. The dorsal and ventral pathways are grossly parallel, but there are over 300 known connections between them.72 A simplified view of the dorsal and ventral cortical visual pathways of the rhesus monkey is shown in Fig. 2.

View Large Image Download to PowerPoint

Fig. 2. Schematic representation of the dorsal and ventral cortical visual pathways showing simplified view of the left hemisphere of a rhesus monkey. The dorsal pathway starts from the occipital lobe (OL) and ends in the inferior parietal lobe (IPL) and ventral pathway starts from OL and ends in the infertemporal cortex (IT). The positions of primary visual cortex (VI), middle temporal lobe (MT), IPL located within posterior parietal cortex are roughly indicated.

In summary, the two major subcortical visual pathways (M and P), which proceed from the respective ganglion cells, show a clear segregation up to the LGN. Subsequently, the mapping of the subcortical to the cortical (dorsal and ventral) visual pathways is not one-to-one as there is mixing of the M and P streams within the visual cortex.25

M cell loss 

Immunolabelling for calcium binding and neurofilament (NF) proteins conducted on retinal tissue of monkey showed no significant differences between the amacrine, horizontal, and bipolar cells in glaucomatous and normal eyes. However, retinal ganglion cells defined by their content of NF proteins, many of which were likely to correspond to M ganglion cells, were greatly reduced in number in glaucomatous eyes as compared to normal eyes.108

Glovinsky et al compared 32 temporal midperipheral retinal locations in 4 monkeys with monocular glaucoma with the normal fellow eye.39 They demonstrated that the distribution of ganglion cell sizes of upper and lower retinal areas were similar. However, more large ganglion cells were lost in the lower than in the upper retina. The optic nerve zones corresponding to these retinal areas also showed a loss of large diameter axons. Quigley et al induced chronic glaucoma in one eye of 10 monkeys (Macaca fascicularis), whose retinal and optic nerve anatomy is close to that of humans.85 They compared the glaucomatous eye to the normal fellow eye, dividing each optic nerve into 16 equal sections. Regardless of the location in the optic nerve, there was a preferential loss of axons of large diameter, which suggests that large ganglion cells are preferentially lost, because ganglion cell size correlates with axon size.77, 78 Similarly, in 7 monkey eyes with chronic intraocular pressure (IOP) elevation, Dandona et al demonstrated that the M layers of the LGN showed less transport of radioactive substance than the P layers.21 In the eyes with normal IOP, the labeling of the M layers was equal or slightly heavier than in the P layers.

In human eyes, a comparison of the size and number of retinal ganglion cells in 6 post-mortem eyes with glaucoma and 5 age-matched post-enucleated normal eyes showed fewer remaining large ganglion cells in the retinal area with atrophy in glaucomatous eyes.83 Also, in a post-mortem comparative study of glaucomatous and normal eyes, the optic nerve fibers larger than the mean diameter were lost earlier than the smaller fibers.84 The loss was spread throughout the nerve, but it was 2 to 3 times more in the vertical poles. LGN autopsy sections of individuals with and without glaucoma showed that the density of M cells was significantly smaller in the glaucoma group than in the control group. No such effect was measured in the P layers.13

Chronic experimental glaucoma in monkeys is similar to chronic human glaucoma, both clinically and histopathologically.89 Therefore, there is a general agreement between monkey and human studies for a preferential damage to large ganglion cells, and the cause of ganglion cell death is commonly attributed to the elevated IOP. The damage to optic nerve fibers in chronic glaucoma tends to respect the following rules. First, at a given location within the optic nerve head, large fibers are lost at a relatively higher proportion than small fibers, because they are more sensitive to the elevated IOP.39 Second, the degree of fiber loss varies as a function of the position within the optic nerve head. Nerve fibers in the superior and inferior poles of the optic nerve typically exhibit the greatest susceptibility to glaucomatous damage.84 It is not clear why large fibers are sensitive to elevated pressures. It could be due to the fact that they are found mainly in the weaker part of the optic nerve head, which corresponds to the superior and inferior poles. Indeed the connective tissue and glial cell structures are thinner in this area as compared to the nasal and temporal areas.81, 87, 88 However, in a recent study, Radius has suggested as an alternate theory, that in eyes with elevated pressure the absorption of nutrients from the axon membrane is decreased due to decreased tissue perfusion.86 This would affect the normal functioning of large fibers more so because of their larger surface-to-volume ratio.

P cell loss 

Although there seems to be an agreement on the predominant loss of large fibers in chronic glaucoma, it should be emphasized that retinal ganglion cells with large axons are not necessarily M cells.47, 84, 99 M cells constitute only about 10% of the ganglion cell population.78 Therefore, some large P cells could also be included among the large cells that are lost.39 Indeed, a reduction in the proportion of P and M cells in the monkey LGN has been demonstrated in cases of chronic glaucoma.99, 107, 114 As far as acute glaucoma is concerned, one monkey eye with acute IOP elevation showed less radioactive labeling in the P layers, as compared to the M layers of the LGN.21 Therefore, both chronic and acute glaucoma preparations have also provided evidence for P losses in the LGN.

In cats, large (Y) ganglion cells are more tolerant to acute IOP elevation than small (X) cells.43, 97 This could be due to the fact that Y cells tend to have more intracellular reserves of ATP, oxygen, and potassium than X cells during ischemia produced by brief elevation of IOP.42, 58 Particular care should be taken, however, when comparing cat and human ganglion cells. Cat X and Y cells share similarities with primate Pβ and Pα cells, respectively (e.g., in terms of relative soma size, relative axon size, relative conduction velocities, and relative transience of response). However, at the level of LGN, very few primate M cells (6–25%) show the marked non-linearity of response that cat Y-ganglion cells do,7 and some primate P cells (0.6–0.7%) have Y-like response characteristics.24, 52

Ganglion cell distribution 

The fact that the glaucomatous damage starts in the mid-peripheral retina is also compatible with a predominant loss of large fibers.39 Indeed, retinal ganglion cells located near the fovea have small cell bodies and comparatively small axons because ganglion cell size increases with eccentricity for both M cell and P cells.111

Interpreting histopathological studies 

The studies reviewed here do not conclusively demonstrate the specificity of M cell losses in glaucoma. Resolving the discrepancies of the results is difficult for several reasons. First, most human histological data come from eyes that are obtained after post-mortem or enucleation.83 Such tissue may not be identical to that obtained from otherwise healthy glaucoma patients. Second, although similar techniques are used to elevate IOP in all studies, the duration and level of IOP elevation may play a role in the type of ganglion cells that are affected.39, 43, 99 Third, the methodologies used so far are by their nature cross-sectional and cannot evaluate the relative progresses in the loss of different cell types.

Despite these difficulties, we can conclude that most of the histopathological studies suggest that the neural loss in glaucoma is partially selective for the M pathway, although the P pathway is not spared. The losses also seem to affect mainly the large cells, which is consistent with greater losses at the peripheral retina where ganglion cells have larger cell bodies and axon diameter than in the fovea.

In spite of this evidence, we should be cautious at predicting that there will be a selective loss in functions subserved by the M pathway, such as motion perception. Indeed, the hypothesis of reduced redundancy47 states that the observed functional deficits will also depend on the degree of redundancy present in the different cell populations subserving different functions.

Current clinical glaucoma psychophysics 

A major principle of current clinical glaucoma assessment is the testing of different points of the visual field, in a search for patterns of field loss that are considered typical of glaucoma. Modern differential light-sensitive (DLS) automated perimetry for visual field testing has historical origins in manual perimetry, and it has not been specifically constructed to test the M or P pathways.

In DLS automated perimetry, the patient looks into a white hemispherical bowl, at a small fixation target with the head is stabilized. The stimulus is usually a broadband visible light of constant size but of variable luminance. It is briefly presented (0.1–0.2 sec) on a uniform background of white light at different eccentricities. The minimum luminance required to detect the stimulus at each location is determined. Estimation of reliability factors such as fixation losses, false-positive, and false-negative responses are considered before interpretation of the visual fields. Different commercial varieties of DLS perimeters (e.g., Octopus, Humphrey Field Analyser (HFA), Dicon, etc.) are available. In the following paragraphs, we have kept the term HFA wherever the original authors have used it.

In glaucoma, visual field testing is usually conducted with the Goldman size III target (0.43°). This stimulus seems to detect advanced rather than mild visual field changes in humans44 because the autopsy of human glaucomatous eyes shows 20–40% of ganglion cell losses before field defects are detected with DLS perimetry.82, 83 In addition, the achromatic stimuli (white light on a white background) that are typically used in DLS automated perimetry have been reported to be nonspecific in their stimulation of various types of ganglion cells.15, 95 Therefore, DLS perimetry may not detect the earliest functional losses, and the goal of new psychophysical tests is to specifically evidence early pathological changes of the visual system in glaucoma.

Studies targeting the m or p pathway 

Based on the anatomical and functional data above, many studies have explored the psychophysical correlates of glaucoma-induced deficits in visual perception. In most of them, the visual functions are evaluated by targeting either the M or P pathway, but some have targeted both pathways in the same patients. Their purpose is to reveal visual functions that are impaired in glaucoma, reveal selective versus non-selective losses in glaucoma, and search for diagnostic tools for glaucoma.

Studies Targeting the M Pathway 

Frequency-doubling perimetry (FDP),12, 50, 63, 80, 100 pattern-discrimination perimetry,31, 32 and scotopic sensitivity38, 101 are some of the tests introduced to target the M pathway. FDP is briefly described here, as some investigators believe that it can be used as a diagnostic12, 22, 63, 100 and a screening50, 80, 113 tool for glaucoma. The stimulus is a large (10 × 10°) low-spatial frequency sinusoidal grating (0.25 c/°) consisting of black and white bars undergoing a rapid counter phase flicker (25 Hz). In the normal eye at a certain level of contrast, the spatial frequency appears to double (the frequency doubling illusion).54 Glaucoma patients fail to detect the FDP stimulus in regions that match grossly the part of the visual field that is found defective through DLS perimetry.80, 100 It has been argued that the failure to detect the FDP stimulus could be due to a loss of My cells (a subset of M cells) in glaucoma.53, 62 However, this is still strongly debated, as some authors report the responses of glaucoma patients might not be due to a doubling illusion effect, but rather to a change in contrast or an apparent flicker of the stimulus.103, 104 Also, the isolation of My cells by the FDP test remains to be clearly determined.24, 57

Studies Targeting the P Pathway 

Short-wavelength automated perimetry (SWAP)37, 48, 92 and high-pass resolution perimetry (HRP)6, 14, 36, 67 have been used to target the P pathway. It has been reported that SWAP detects earlier48 and more extensive92 visual field defects than HFA. Previously it was considered that SWAP involves the P pathway, but now it is thought that SWAP isolates the blue cone pathway via the small bi-stratified sub-population of ganglion cells. Whether the axons of these ganglion cells project to the K cell interlaminar layers of LGN is still under study, but there is some evidence for this.66

Studies Targeting M and P Pathways 

Bassi et al5 studied contrast sensitivity, stereoacuity, and color vision in glaucoma patients and concluded that the results reflect a selective damage to the M pathway by referring to the functional properties of this pathway.59, 60 However, this conclusion is questioned as stereopsis may also be mediated by the P pathway95 and deficits in color perception have been reported by others in glaucoma.28, 93 Peripheral resolution acuity measurements using gratings that were either stationary or dynamic (phase reversed at 30 Hz) revealed that dynamic resolution was significantly more affected than static resolution, suggesting a dominant M function loss in glaucoma.2 In another study, a battery of psychophysical and electrophysiological tests reflect glaucomatous damage to a varying degree, and do not point to a specific M or P loss.41

In glaucoma patients and suspects, a comparison of motion-automated perimetry (MAP) (see below for a thorough description) and SWAP was performed in one study,91 and in a second study90 MAP, SWAP, FDP, and HFA were compared. In both studies, each test successfully identified a subset of eyes as abnormal. In the latter study, many more OHT subjects (17 of 37) gave abnormal results with FDP as compared with the results of HFA (2 of 37). These OHT subjects were selected without reference to the results of the HFA. The percentage of OHT subjects, which show abnormal results on FDP, was much higher than those which might be expected to convert to glaucoma. Therefore, it suggests that FDP is not a selective screening tool but may be useful in combination with other perimetry tests. Because SWAP, FDP, and MAP are very likely to stimulate different types of ganglion cells, these studies support non-specific ganglion cell loss. As suggested by Sample et al,91 it could also be that individual differences exist in the type of ganglion cells that are first affected in glaucoma.

In conclusion, psychophysical studies reveal an early selective loss in glaucoma either by targeting the M or the P pathways. However, most of the studies which compare the integrity of these pathways in the same group of patients reveal that visual functions subserved by both the M and P pathways are affected5, 41, 90 and that the tests targeting the M pathway seem to be more sensitive for detecting glaucoma.2, 91

Methods 

The tests of motion perception include global (full field) and local (at eccentricities) motion tests. The most commonly used stimuli are dynamic sparse random dots generated on a computer screen. Such stimuli are of methodological interest in ophthalmology, because the motion sensitivity seems relatively immune to blur variations for these stimuli. Indeed an optical blur of 3.25 diopters (D) does not modify coherent motion thresholds for a range of stimulus displacement and speed,4 and up to 8 D of optical blur weakly modified motion thresholds for a range of contrast values.102 Such stimuli are used to measure 1) the minimum and the maximum perceived displacements, referred as Dmin and Dmax, respectively, (Fig. 3); and 2) the ability to report or detect the motion of a subset of coherently moving dots, within a population of randomly moving dots. The corresponding percentage of coherently moving dots (PCD) ranges between 0% and 100%, and it produces graded intensities of motion signal (Fig. 4). PCD stimuli can be presented either globally or in small regions of the visual field in perimetry tests (Fig. 5).

View Large Image Download to PowerPoint

Fig. 3. The minimum perceived displacement within a random dot kinematogram is called Dmin and the maximum perceived displacement is called Dmax.

View Large Image Download to PowerPoint

Fig. 4. Random dot kinematograms are used to study global motion perception. A: All dots move in random directions, hence no coherent motion is perceived. B: 50% of the dots move in the same direction, in a coherent motion.

View Large Image Download to PowerPoint

Fig. 5. Diagram of motion perimetry. The enlarged target shows 50% of the dots moving at random and 50% moving in one direction (bold dots). X = the fixation target. (Redrawn from Wall et al.110)

The different psychophysical tasks are direction discrimination (indicating the motion direction), motion detection, and motion localization (pointing to the area of movement).

Displacement Thresholds 

Dmin and Dmax thresholds in central vision have been measured over a 2.9° or 19° field of vision in a glaucoma and OHT patients.10 The results reveal a larger impairment in motion sensitivity in glaucoma patients for Dmin, as compared to Dmax. However, the Dmin values were poorly correlated with the visual fields on HFA. Bullimore et al interpret their results as the consequence of a preferential damage to the larger M ganglion cells.10 This is in agreement with the fact that magnocellular cells are more sensitive to small displacements than P cells.56

In another study, peripheral displacement threshold was measured at 15° in the temporal field just above and below the blind spot.35 The target was a 2° by 2° vertical line oscillating at 2.5 Hz. Thresholds were measured as the minimum amplitude of oscillation that could be detected. A general increase of displacement thresholds occurred in glaucoma and OHT patients. However, as noted elsewhere,10 the histograms suggest that the peripheral displacement thresholds were elevated in only 9 of 16 (56%) glaucoma patients and in 10 of 45 (22%) OHT subjects.

Coherent Motion Studies: Global Tests 

In several studies, coherent motion thresholds in central vision were measured in normals, and glaucoma and OHT patients. Motion thresholds for a large stimulus (60° width), with low (0.03 dots/degree2) dot density, fell significantly outside the normal range for POAG and OHT patients.98 However, there was a considerable overlap in performance between patients and normal subjects. Because the results did not correlate with the HFA test, the authors suggest that the deficits in motion perception might occur early in the pathology.

Global testing with PCD also showed a poor discriminatory power between POAG and normal subjects in two other studies. First, with a 19° stimulus, and dot density of 7 dots/degree2 no significant difference in motion thresholds between patients and normal subjects was found.10 Second, the same result was obtained with a 25.1° stimulus and a dot density of 0.83 dot/degree2.9

Contrasting with these results, when a 24.5° stimulus with a density of 0.42 dots/degree2 was examined, a significant elevation of the coherent motion thresholds was observed in 70.6% of POAG patients.102 The deficits were more frequently detected with a speed of 12.5°/second, rather than a speed of 4.2°/second. In addition, there was a significant correlation between HFA visual defects and motion thresholds for high speeds, but not for low speeds. Motion deficits were detected in OHT patients, but they were not significant.

In this latter study, a possible reason for the difference in the results of the two speeds is that fast motion tends to be coded by the M pathway, while slow motion may involve the P pathway.70 However, the results obtained by Trick et al102 contradict the lack of difference between glaucoma patients and controls found by Bosworth et al,9 as described above. The major difference between the two studies is that Trick et al102 used a contrast (15.6% for the highest velocity) that was half that used by Bosworth et al.9 Considering the fact that the M pathway is sensitive to low contrast, a possible explanation is that the stimuli used in the study by Trick et al targeted more specifically the M pathway and led to a better glaucoma discriminatory power for their test.

In any case, the best sensitivity observed across different stimuli in the study of Trick et al is 70.6% for glaucoma patients. Therefore, it can be concluded that global testing through PCD is not an adequate tool for glaucoma screening and diagnosis. This could be due to the fact that global stimuli address both pathological and normal regions, which limits their discriminatory power for glaucoma.46 Also, a centrally placed stimulus would fail to cover the damaged visual areas in many early glaucoma patients.51

Coherent Motion Studies: Perimetry 

Because of the poor sensitivity of global motion testing, several authors designed perimetric motion tests, which have the double benefits of testing motion perception and isolating areas that are affected by glaucoma.

A first study by Bosworth et al showed that the regions of visual loss, as defined by the HFA visual fields, had a significant lower threshold for coherent motion than the regions that were relatively spared by the disease.8 This study gave some methodological support (in terms of the sensitivity of the test) to the use of restricted size stimuli (here 7.3° wide) to map motion sensitivity in different regions of the glaucomatous retina.

A motion perimetry study was performed by Wall and colleagues, who used targets of 20 different sizes (0.25–21°).110 The size of the target varied from trial to trial, and a 2-1 staircase procedure was used. These targets were randomly presented at 44 visual field locations within the 54 locations of the 24-2 HFA test (some of the most eccentric points were skipped). The target was a field of static dots, within which a circular area of moving dots was shown. In this area, 50% of the dots moved coherently, at 30°/second. The target size was decreased, until the smallest motion target could be localized. The POAG patients showed a significant increase of the motion size thresholds, as compared to controls, in many loci of the visual field. The loci expanded over most of the defects revealed by HFA, but they were usually larger than the defects. In addition, motion perimetry identified defects following nerve fiber bundle-like patterns, which were missed by the HFA.

The same procedure showed that OHT patients had slightly higher motion-size thresholds than controls.109 Although on average the difference was not significant, 6 OHT patients presented field defects following a nerve fiber bundle-like pattern. Therefore, motion perimetry used by Wall et al seems to be highly effective in glaucoma screening and in diagnosing OHT subjects.109, 110

Using a fixed target size of 3 degrees, Joffe and colleagues tested each quarter of the visual field along the horizontal, vertical, and oblique meridians (Fig. 6A).46 The target was presented at 22 locations, including the fovea and eccentricities of 9°, 15°, and 21°. The dot speed was 2.63°/second. The subject's task was to indicate the direction of coherent motion. The PCD was decreased from 100%, until the motion direction was wrongly reported. Under foveal testing, no significant difference was found between glaucoma patients and controls. However, the mean peripheral thresholds (at 21° eccentricity) were significantly different between the glaucoma and control groups. This difference was the largest in the superior visual field (SVF) (Fig. 6B), but the four quadrants of the visual field of glaucoma patients presented a decrease in motion sensitivity. As for glaucoma suspects, the deficits occurred only in the SVF. Overall, significant SVF deficits occurred in 10 of 12 patients, and in 8 of 15 suspects, and the deficits tended to increase with eccentricity.

View Large Image Download to PowerPoint

Fig. 6. A: Stimulus location at various eccentricities. Along the horizontal axis, negative numbers indicate normal hemifield locations and positive numbers indicate temporal hemifield locations. B: Group mean motion sensitivity as a function of eccentricity along the sup vertical meridian for each group. Both patient group exhibit lower sensitivity than controls at all eccentricities with greatest deficit at most eccentric location. (Fig. 6A was redrawn and Fig. 6B was reprinted from Joffe et al46 with permission of Vision Research.)

A similar procedure was conducted by Bosworth et al in which a 7.3° stimulus was presented in 15 locations, including the fovea, and led to similar results (see Table 1, lower 2 rows) for glaucoma patients.9 However, the results of Bosworth et al9 differed from Joffe et al46 in that 1) Bosworth et al do not report the high difference between the upper and the lower visual fields, and 2) more importantly, the percentage of glaucoma suspects found abnormal in the study of Bosworth et al9 is 26%, which is considerably lower than the 53% obtained in the study of Joffe et al.46 There are a number of possible reasons for this inter-study difference in discriminatory power. For example Joffe et al46 used a smaller stimulus of 3°, which might be better at detecting small motion scotomas than the 7.3° stimulus size used in the study by Bosworth et al.9 In addition, the two studies use different criteria for normal controls, with Bosworth et al9 defining normals as having IOP of ≤ 21 mm Hg, cup/disk ratio of < 0.6 and normal HFA 24-2 fields and Joffe et al46 using the definition of IOP < 21, cup/disk ratio < 0.4 and normal HFA 30-2 fields. Hence, it may be possible that more borderline suspects may be detected in the study by Joffe et al.46

Table 1. Comparison of Several Studies on Motion Perception in Glaucoma

	Psychophysical test		References	Stimulus Size (deg)		Eccentricity (deg)	Speed (deg/s)	Diagnosis	No. of Patients	% of Patients with Elevated Thresholds
	Task	Stimulus
	Motion detection	Dmin	Bullimore et al10 (1993)		2.9	0	—	OHT POAG	23 15	9% 67%
	Motion detection	Dmax	Bullimore et al10 (1993)		19	0	—	POAG	15	13%
	Motion detection	Line	Fitzke et al35 (1987)		2 × 2	15 (above & below the blind spot)	—	OHT POAG	45 16	22% 56%
	Direction discrimination	Coherent motion (global)	Bullimore et al10 (1993)		19	0	10	POAG	15	20%
	Direction discrimination	Coherent motion (global)	Silverman et al98 (1990)		60	0	—	OHT POAG	14 37	21% 43%
	Direction discrimination	Coherent motiona (global)	Trick et al102 (1995)		24.5 24.5	0 0	4.2 12.5	OHT POAG OHT POAG	16 17 16 17	37.5% 58.8% 12.5% 70.6%
	Motion localisation	Coherent motion (perimetry)	Wall et al109 (1997) Wall et al110 (1995)		0.5–21	24	30	OHT POAG	27 25	22% 96%
	Direction discrimination	Coherent motion (perimetry)	Joffe et al46 (1997)		3	21	2.83	Glaucoma suspects POAG	15 12	53% 83%
	Direction discrimination	Coherent motion (perimetry)	Bosworth et al8 (1997)		7.3	24	2.83	Glaucoma suspects POAG	31 19	26% 84%

a Results taken at 15.6% contrast for cardinal direction analysis.

A separate issue to consider is that, in the study by Joffe et al,46 the 53% of glaucoma suspects giving abnormal results on motion perimetry is a higher percentage than expected and may be overcalling defects, because from clinical practice we know that only a very small percentage of glaucoma suspects convert to glaucoma.73 This percentage is even more unusual given the stringent criteria used in the study by Joffe et al46 for defining glaucoma suspects (IOP of > 21mm Hg, cup/disk of < 0.4 and normal HFA 30-2 fields) compared to the study of Bosworth et al9 (IOP > 24 mm Hg, cup/disk ratio of ≥ 0.6 and normal HFA 24-2 fields). Thus, motion perimetry used in the study by Joffe et al46 appears not to be very selective for potential glaucoma patients and further study is required to explain the large number of suspects found abnormal.

In conclusion, motion perimetry reveals a decrease in motion sensitivity in specific areas of the visual field. All three studies9, 46, 110 found motion perimetry defects in similar areas as the HFA, and detected abnormalities in a proportion of OHT or glaucoma suspects (26%, 53%, and 22%, respectively). These results clearly support the need for larger population testing and also for working out the differences in their chosen experimental parameters.

Discussion of the studies on motion perception 

As summarized in Table 1, there is a large variation in sensitivity across the motion tests. The motion thresholds are elevated in 13–96% of the POAG patients, and 21–53% in OHT or glaucoma suspects.

In addition to the differences in the stimulus parameters such as size, dot density, and presentation time used in the various studies, each study compares patients to different normative databases (visual field examinations were done using different DLS perimeters). Also, considering the database of normal subjects (presenting totally normal visual fields), the question arises whether a given percentage present early glaucoma deficits creating losses that were not detected by DLS perimetry.

The reports on the minimum-displacement thresholds yield similar results,10, 35 with less than 67% of glaucoma patients being considered abnormal. On one hand, the score of 67% abnormal patients found by Bullimore et al10 in central vision suggest a deficit in motion perception occuring prior to HFA detection. On the other hand the 56% score obtained by Fitzke et al35 seems to be very low, considering that the tested regions above and below the blind spot are field areas often affected in early glaucoma.26, 27, 29 This contradiction could either be due to the longer presentation duration (2 seconds) used by Fitzke et al or to the fact that these authors used a line stimulus with a determined shape, which could stimulate the P pathway, whereas Bullimore et al used random-dot motion which could address more specifically the M pathway. Indeed it has been found that line movements and random dot movements have different psychophysical properties in terms of the sensitivity to their temporal and spatial frequencies,75 which is compatible with the idea that the motion of line patterns might also be processed by the P pathway.

Overall, the best sensitivity to glaucoma is obtained in the three studies of motion perimetry.9, 46, 110 The elevated motion thresholds were 83–96% for POAG patients. These rates are higher than those obtained with global motion tests. Also, these tests are designed to assess small localized areas of the visual field. Interestingly, these perimetry studies delineate defects in parts of the visual field that are considered normal with the HFA. In this regard, Joffe et al46 find that the highest deficits occur in the superior hemifield, as is usually found by the HFA. This confirms the fact that the greatest damage to the optic nerve head occurs in the inferior pole,11 and it supports the hypothesis that the arcuate fibers of the superior field are particularly susceptible in early glaucoma.1 However, it contrasts with the finding, that motion deficits occur also in the inferior field for glaucoma patients and suspects.9, 110 In any case, all three studies detected deficits in areas where no defects were detected with HFA, and in most cases the extent of field losses was larger if depicted by motion perimetry. Together with the high rate (53%) of OHT patients found abnormal,46 it suggests that early glaucomatous damage affecting motion perception occurs in these areas.

Finally, the high sensitivity (84%) and specificity (100%) obtained with motion perimetry9 for glaucoma patients are close to the reported sensitivity (87–97%) and specificity (90–99.8%) of FDP.49, 76 However, motion perimetry is only a promising screening tool, as we have seen above that FDP has been validated in studies involving large12, 61, 64 numbers of patients whereas motion perimetry has been tested only on three restricted populations.