Cerebral Cortex, Vol. 10, No. 1, 2-22,
January 2000
© 2000 Oxford University Press
Visual Function and Brain Organization in Non-decussating Retinal–Fugal Fibre Syndrome
1 Department of Neurology and Neuroscience, The New York Presbyterian Hospital, New York, NY 10021, USA, , 2 Department of Physiology, Medical Faculty, Erasmus University Rotterdam, 3000 DR Rotterdam, The Netherlands, , 3 Department of Neurology, Memorial Sloan–Kettering Cancer Center, New York, NY 10021 and , 4 Center for Neural Science, New York University, New York, NY 10003, USA
 |
Abstract |
|---|
 Top Abstract IntroductionClinical ProfileFunctional Imaging StudiesPsychophysical and Evoked...DiscussionAppendixReferences |
|---|
Functional neuroimaging, psychophysical and electrophysiological investigations were performed in a patient with non-decussating retinal–fugal fibre syndrome, an inborn achiasmatic state in which the retinal projections of each eye map entirely to the ipsilateral primary visual cortex. Functional magnetic resonance imaging (fMRI) studies showed that for monocularly presented simple visual stimuli, only the ipsilateral striate cortex was activated. Within each hemisphere's striate cortex, the representation of the two hemifields overlapped extensively. Despite this gross miswiring, visual functions that require precise geometrical information (such as vernier acuity) were normal, and there was no evidence for the confounding of visual information between the overlapping ipsi-lateral and contralateral representations. Contrast sensitivity and velocity judgments were abnormal, but their dependence on the orientation and velocity of the targets suggests that this deficit was due to ocular instabilities, rather than the miswiring per se. There were no asymmetries in performance observed in visual search, visual naming or illusory contour perception. fMRI analysis of the latter two tasks under monocular viewing conditions indicated extensive bilateral activation of striate and prestriate areas. Thus, the remarkably normal visual behavior achieved by this patient is a result of both the plasticity of visual pathways, and efficient transfer of information between the hemispheres.
|
Introduction |
|---|
|
TopAbstractIntroductionClinical ProfileFunctional Imaging StudiesPsychophysical and Evoked...DiscussionAppendixReferences |
|---|
Understanding the relationship of functional connections among brain areas and behavioral performance is a major goal of current neuroscience research. The visual system is a particularly useful model system to study this structure–function relationship because in the early stages of cortical visual processing information is represented as a series of retinotopic maps. We describe in detail a patient with an isolated achiasmatic state in which both temporal and nasal retinal output of each eye map to ipsilateral primary visual cortex. The overall pattern of relatively preserved visual behavior, coupled with markedly abnormal functional connectivity, provides insight into the profound plasticity of normal structure–function relationships.
The decussation of the nasal retinal fibers at the optic chiasm is essential to the normal mapping of visual information, ensuring that each hemisphere receives binocular information about the contralateral visual field. For each hemifield, the precise alignment of the retinotopic maps from each eye in the lateral geniculate nucleus (LGN) and the coincidence of the corresponding points in primary visual cortex are regarded as crucial for the organization of visual processing. Apkarian and colleagues (Apkarian et al., 1994a
, 1995) identified a previously undescribed congenital malformation, the absence of the optic chiasm, in two unrelated children with associated oculomotor instabilities. Misrouting of retinal fibers consistent with an absence of the optic chiasm was suggested by visual evoked potentials (VEPs), which showed activation of occipital cortex primarily by the ipsilateral eye. The complete absence of the optic chiasm was subsequently confirmed by cranial magnetic resonance image (MRI) scans, which showed a continuous separation of the optic nerves without other associated structural defects. Monocular visual fields were full. Thus, apparently, each side of the brain receives a nearly complete, but totally monocular, map of the whole visual field. As a consequence, the anatomical organization of visual processing is fundamentally altered. This adaptive mapping is in marked contrast to the primary visual pathway organization in acquired achiasmatic deficits due, for example, to tumor (Walsh, 1956; MacCarty et al., 1970) or trauma (Traquair et al., 1935; Østerberg, 1938). In acquired chiasmatic anomalies, the nasal fibers are lost rather than misrouted to the ipsilateral hemisphere, resulting in bitemporal hemianopsia rather than dual complete representations.
The identification (Williams et al., 1994) of a strain of mutant achiasmatic Belgian sheepdogs, also with a notable incidence of comparable oculomotor instabilities (Williams and Dell'Osso 1993), provided an animal model of the isolated chiasmal malformation. Electrophysiological exploration at the level of the LGN in the canine mutants demonstrated that retinal fibers terminated in anatomical layers that normally would receive input appropriate to the hemiretina, nasal or temporal. The aberrant nasal fibers from the ipsilateral eye formed a mirror-image map, in the LGN layers that ordinarily would receive nasal fibers from the contralateral eye.
The findings in the achiasmatic canines are comparable, albeit opposite to those in albino mammals, in which the fibers from the temporal retina near the vertical meridian erroneously decussate to the contralateral hemisphere (Lund, 1965; Guillery and Kaas, 1971; Gross and Hickey, 1980; Guillery et al., 1984). In albinism, a genetic abnormality present across species, the LGN receives aberrant crossed input from temporal fibers which replaces a portion of the normal ipsilateral temporal projections. Again, each hemiretina maps to the appropriate layer of the LGN; however, the abnormal segment represents the mirror symmetric portion of the visual field, disrupting the normal alignment from layer to layer (Guillery, 1986, 1990).
The effects of chiasmal abnormalities at the cortical level have been explored in Siamese cats, homozygous for an allele of the albino gene (Guillery and Casagrande, 1977; Stone et al., 1978; Guillery, 1996). In albino optic pathway misrouting, regardless of mammalian species, axons from each eye project to non-overlapping sites within the LGN, as in normal animals (Shatz, 1977; Cooper and Blasdel, 1980). Beyond the LGN, two patterns of anatomical organization of the geniculocortical pathways and respective cortical topography have been described. In one pattern (Hubel and Wiesel, 1971), termed ‘Boston’, primary visual cortex contains the normal hemifield representation as well as a portion of the ipsilateral temporal field, compressed in retinotopic order. The representation of the vertical meridian is displaced from its usual position at the 17/18 border. In the other pattern (Kaas and Guillery, 1973), termed ‘Midwestern’, the cortical areas which receive aberrant input from the contralateral temporal retina are primarily silent, and the cats apparently do not respond to behavioral stimuli presented to the temporal retina (Elekessy et al., 1973; Guillery and Casagrande, 1977). A recent review (Guillery, 1996), however, does not consider these patterns to represent mutually exclusive neural wiring strategies, and indeed the two distinct profiles of neural wiring may well coexist within a single animal (Cooper and Blasdel, 1980).
Regardless of the profile of adaptive neural organization, in both the albino and the achiasmat, the visual cortex must somehow incorporate expanded, albeit essentially monocular, information about the visual world. The achiasmat is the more extreme variant because of a total lack of input from the contralateral eye coincident with a full hemifield of aberrant input from the ipsilateral eye. For the achiasmat and as well as for the majority of albinos, the retinogeniculate miswiring is accompanied by oculomotor instabilities, the mechanism of which remains unclear (Collewijn et al., 1985; Abel, 1989; Leigh and Zee, 1991; Williams et al., 1994). The associated optic pathway misrouting and related oculomotor perturbations raise a variety of questions about subsequent visual processing at the cortical level. For example, how does the developmental programming of the visual cortex accommodate the profound retinogeniculate miswiring? How does the altered mapping and sensory input affect visual processing? Does convergence of monocular information occur, and if so, where? What are the effects of the anomalously organized visual system on the normal pattern of hemispheric lateralization and specialization? Previous attempts have been made to address some of these queries by studies of human albino visual function (Taylor, 1978; Loshin and Browning, 1983; Abadi et al., 1989; Abadi and Pascal, 1991). However, these studies are limited primarily because of the inherently underdeveloped albino foveae and associated reduced visual acuities (Taylor, 1975; Jacobson et al., 1984; Wilson et al., 1988; Abadi and Pascal, 1991; Pérez-Carpinell et al., 1992). Although visual behavioral studies in albino mammals also have been attempted (Creel and Sheridan, 1966; Elekessy et al., 1973; Loop and Frey, 1981; Schall et al., 1988), the latter are limited by the difficulty in assessing higher-order visual processing the absence of prominent hemispheric specialization.
The human achiasmats described by Apkarian and colleagues (Apkarian et al., 1994a, 1995; Apkarian, 1996) have normal fundi and no associated endocrinological, neurological and/or anatomical abnormalities, thus providing a unique opportunity to address various issues and queries concerning structure– function relationships. In the present study, we describe a series of investigations performed on one of the achiasmatic patients. To distinguish the isolated condition of an absent optic chiasm from various congenital absence of the chiasm associated with other inborn midline anomalies, e.g. septo-optic dysplasia (Kaufman et al., 1989; Roessmann, 1989), the naturally occurring, isolated inborn achiasmatic condition has been referred to as ‘non-decussating retinal–fugal fibre syndrome’.
The study described herein was specifically designed to assess the effects of the human achiasmatic state and associated eye movement instabilities on visual pathway organization, visual function and visual perception. Investigations were designed to focus on several overlapping issues. (i) Does the abnormal retinocortical mapping affect aspects of visual function that might depend critically on precise geometrical information, such as vernier acuity, texture perception and illusory contour formation, and does it lead to abnormal interactions between visual inputs in opposite hemifields? (ii) Are there abnormalities in visual functions that are likely to be sensitive to the ocular instabilities, such as contrast sensitivity or velocity discrimination? (iii) Given the ipsilateral-only projection pattern, do visual signals originating from the two eyes interact? (iv) What is the effect of the ipsilateral-only projection pattern on aspects of perceptual and cognitive function that involve hemispheric specialization or lateralization in normals, such as interactions of vision and language (as assayed by visual naming), visual search, attention (as assayed by an endogenous evoked potential, the P300) and illusory contour formation (Hirsch et al., 1995)?
To aid the reader's navigation of this multidisciplinary study, we begin with a clinical profile, including the achiasmat's clinical history, anatomical imaging, neurological evaluation and ophthalmological/oculomotor evaluation. The individual investigations are then described. For each of the major studies, the methods, results and brief comments are presented separately. For the minor studies, results are summarized in the body of the text, and methodological details are deferred to the Appendix. These investigations were performed in four laboratories over a period of 5 days when the patient was 18 years of age, with appropriate informed consent from the patient and her parents in accord with institutional guidelines.
|
Clinical Profile |
|---|
|
TopAbstractIntroductionClinical ProfileFunctional Imaging StudiesPsychophysical and Evoked...DiscussionAppendixReferences |
|---|
Case History
The patient, a fraternal twin (both females) was born at term by spontaneous delivery without complications following an uneventful third pregnancy. Birth weight was ~2500 g. Family members, including the twin sister, two older siblings (male and female) and the parents, tested negative for comparable ocular and/or optic pathway anomalies. Detailed pedigree analysis and a search for relatives with related symptoms also proved negative, suggesting a possible autosomal recessive genotype. The patient's direct clinical records report (i) ocular motor irregularities at 3 months of age, (ii) myopic refractive correction at 2 years, and (iii) nystagmus, head shudder, alternating esotropia and torticollis at 6 years. Also at the age of ~6 years, the patient presenting, in addition to ocular motor instabilities, with blond hair, fair skin and blue irides was tentatively diagnosed with possible albinism. At 9 years, the patient was referred for the VEP albino misrouting test, based primarily on the presenting ocular motor instabilities. The VEP misrouting test, implemented three times over a 5 year period, consistently revealed abnormal VEP cortical response topography. Rather than contralateral hemispheric VEP response dominance following monocular stimulation [a pathognomonic and highly reliable indication of albino optic pathway misrouting (Apkarian, 1991a,b)], the VEP response profile consistently showed ipsilateral response dominance. Repeated and appropriate visual pathway MRI imaging finally clarified the VEP results by documenting the isolated and complete absence of an optic chiasm structure (Apkarian et al., 1994a, 1995; Apkarian, 1996). CT scans and MRI imaging also documented as normal other midline, subcortical or cortical structures. Around the period of the first VEP test, visual acuity was reported as 20/100 OD (right eye) and OS (left eye); visual fields and color vision were normal. Stereoscopic vision was absent. Patient medical history also revealed normal growth and endocrine function. The patient has attended regular high school, participates and performs well in gymnastics and rides a bicycle without difficulty. She does report difficulty judging the speed of vehicles in traffic and that she performs poorly in sports activities that involve throwing, catching or hitting a fast-moving target. In the last few years there has been no noticeable change in the patient's vision, except for a slight improvement due to the use of contact lenses.
Anatomical Imaging
As previously reported (Apkarian et al., 1994a, 1995; Apkarian, 1996), a high-resolution MRI at age 15 demonstrated complete absence of the optic chiasm, without associated midline abnormalities. The anatomical MRI study documented the absence of auxiliary and/or accompanying midline and visual pathway abnormalities. The latter point is of particular importance to distinguish this patient's condition from other forms of chiasmal agenesis that are associated with other morphological perturbations (Maitland et al., 1982; Kaufman et al., 1989; Roessmann, 1989; Dutton, 1994; Menezes et al., 1996). Previous CT scans showed only a broad and hypodense cisterna chiasmatis and an increased dorsum sellae to pons distance but were non-diagnostic, as were lower-resolution MRI's (Apkarian et al., 1994a, 1995). High-resolution anatomical imaging, including axial and coronal scans, was pursued because of the repeated, consistent VEP evidence of optic pathway misrouting (Apkarian, 1991b).
Neurological Evaluation
The examination was carried out in the patient's native Dutch, with the aid of a translator. The patient was alert, oriented, cooperative and right-handed. No disorders of language were noted. There was an intermittent head tilt (torticollis) to the right, with a rapid titubation. There were no dysmorphic features, and no asymmetries in sizes of digits. Detailed evaluation of cranial nerves I, V and VII–XII were entirely normal. Visual fields were full to finger confrontation with single and double simultaneous stimulation. Pupils were 5–6 mm, briskly reactive to light from nasal or temporal fields, and with consensual response to light from either field. There was a prominent nystagmus in horizontal, vertical and torsional planes. The oculomotor instabilities exacerbated in upgaze, and dampened with extreme lateral gaze, especially to the left [more detailed description and analysis of visual and oculomotor function may be found below and elsewhere (Apkarian et al., 1994a,b, 1995, 1997, 1999; Apkarian, 1996)].
General motor evaluation revealed that fine finger movements (drawing circles with index finger) were slightly clumsy on the left. On first and second trials of tapping with the left foot, there was overflow to the right, but there was no overflow on repeated trials or overflow to the left with right foot tapping. The remainder of her neurological examination, including gross motor, sensory, coordination and gait testing, was normal. Additionally, simultaneous tactile stimuli presented to any two limbs were readily identified. A posture or movement passively imposed by the examiner on either hand was readily mimicked by movements of the other hand.
The patient's score on the Edinburgh Handedness Inventory (Oldfield, 1971) was +100, the maximal right-handed score.
Ophthalmologic/Oculomotor Evaluation
Fundus examination revealed normal foveal and macular reflexes. Fundus pigmentation was light but normal. Optic disks, slightly myopic, were otherwise normal. Retinal vasculature also was normal; optic media were clear. The patient's blue irides were not diaphanous. Optical refraction resulted in spherical and cylindrical corrections with OD: S = –5.25, C = –1.5 axis 50° and OS: S = –6.75. Visual acuity was dependent upon viewing conditions. Corrected Snellen acuity at 6 m was 20/60 OD viewing, 20/100 OS viewing and 20/60 OU (both eyes) viewing. At 3 m, with OU viewing, acuity improved to –20/50, and at 20 cm (reading distance) OU acuity could reach 20/20. Interocular alignment showed an alternating esotropia of 20–25 prism diopters with preference for right eye fixation and a highly variable hypotropia (2–5°) of the left eye also during right eye fixation. Color vision tested with Ishihara plates was normal. Stereopsis tested with Titmus plates, TNO and Julesz stereograms was absent.
As reported previously (Apkarian et al., 1995; Apkarian, 1996), static perimetry revealed normal visual fields; normal kinetic perimetry was confirmed during the present study. For kinetic perimetry testing, minor field instabilities at primary position were recorded due to the patient's oculomotor instabilities. However, for both static and kinetic perimetry, the blind spot was accurately plotted and the measured vertical, horizontal and oblique meridians demonstrated 100% function.
Multiplanar eye movements recorded (Apkarian, 1996; Apkarian et al., 1997, 1999) with highly accurate twoand three-dimensional scleral induction coil methods (Robinson, 1963; Ferman et al., 1987) revealed eye movement instabilities in the horizontal planes characteristic of classic congenital nystagmus (Yee et al., 1976; Daroff et al., 1978; Abadi and Dickinson, 1986), including well-defined foveation periods and typical congenital nystagmus waveforms. Oculomotor instabilities in the vertical and torsional planes also were present and followed profiles characteristic of classic see-saw nystagmus (Maddox, 1914), with intorsion and upward movement of one eye and simultaneous extorsion and downward movement of the fellow eye (Apkarian et al. 1999). Velocities depended strongly on viewing distance, target position and patient attention.
|
Functional Imaging Studies |
|---|
|
TopAbstractIntroductionClinical ProfileFunctional Imaging StudiesPsychophysical and Evoked...DiscussionAppendixReferences |
|---|
Motivation
The initial fMRI investigations were designed to confirm the ipsilateral occipital projection of the visual pathways and to establish whether there were corresponding anomalies in sensorimotor pathways. Subsequent fMRI investigations examined the pattern of activation in object naming and illusory contour perception. These more complex visual tasks were of particular interest because of the known hemispheric asymmetries such as left-hemisphere dominance for language, and (Hirsch et al., 1995) a tendency for right-hemisphere dominance for illusory contour completion.
Methods
Stimulation
Full-field flash visual stimulation was accomplished via red LED goggles (Grass Instruments Model S10VSB) set to flash at 8 Hz. Patterned visual stimulation was provided by back-projection onto a screen located ~205 cm from the subject's visual path in front of the scanner. The patient viewed the stimuli by looking up into a slanted mirror located above the head. She was instructed to fixate a continuously visible crosshair (0.8° x 0.8°) to help maintain a stable head and eye position. Full-field high-contrast checkerboard stimuli had a check size of 1° and a field size of 12° x 12°, and were contrast-reversed at 8 Hz. Hemifield checkerboards consisted of the left-most or right-most 5 x 12 subarray of checks, and foveal checkerboards consisted of the central 4 x 4 subarray of checks. Illusory contour stimuli consisted of Kanizsa squares (see Fig. 5) and related configurations that subtended ~4.5°, with the corner elements (inducers) subtending 2°. For language tasks, visual stimuli consisted of black-and-white line-drawings of objects taken from the Boston Naming Test (Kaplan et al., 1983), presented to occupy approximately an 8° x 8° field centered around fixation. Visual stimulation was either monocular or binocular, as specified below; the non-viewing eye was covered by an opaque occluder.
|
Sensorimotor activation consisted of self-paced finger–thumb tapping, with each hand tested separately.
Imaging
Images were acquired on a 1.5 T GE scanner equipped for echoplanar imaging. A standard quadrature head coil (GE) was employed for initial studies of monocular full-field reversing checkerboard stimulations, illusory contours, sensory/motor stimulation and internal speech (object recognition). A 5 in. surface coil centered on the occipital pole was employed for the other stimulation conditions to optimize sensitivity in the occipital lobe. The axial planes were set parallel to a reference line through the superior edge of the anterior commissure and the inferior edge of the inferior commissure (AC–PC line) (Talairach et al., 1988) on a T1-weighted midsagittal view acquired prior to each imaging session. A standard echoplanar sequence (TR = 3000, TE = 60, flip angle = 90°) was employed to acquire 16 contiguous slices. In the case of the global head coil studies slice thickness was 4.5 mm and covered the entire cortical area, whereas in the case of later surface coil studies slice thickness was 3.0 mm and the slices covered the occipital lobe. Otherwise all imaging methods were equivalent for both series. With these parameters, in-plane resolution was 1.6 x 1.6 mm.
Thirty images were acquired during each 90 s imaging run. The first 10 images (30 s) were acquired during a baseline period, followed by a stimulation or task period of 10 images (30 s), and a final (30 s) baseline period also consisted of 10 images. Two identical runs were performed in each condition. Prior to statistical analysis, all brain images for each session were computationally aligned for direct spatial comparisons between different tasks (Woods et al., 1993), and a two-dimensional Gaussian filter (~3 volume elements, voxels, at half-height) was applied except where the highest resolution was required and then no filter was applied. Significant signal changes were identified by a multi-stage statistical analysis which compared average baseline and stimulation signal intensities and required significant signal changes on two runs (coincidence) (Hirsch et al., 1995). For the criteria used in the images shown here, the fraction p of false-positive voxels, empirically determined from images of a spherical phantom filled with copper sulfate solution (General Electric standard), was <0.0005, unless otherwise noted.
Results
Figure 1 shows activations induced by 12° x 12° checkerboard stimulation of each eye. Activation in primary visual cortex was restricted to the ipsilateral hemisphere. Extrastriate activation was seen on the ipsilateral side with stimulation of either eye, and also in the contralateral hemisphere with right eye stimulation (Fig. 1B). In contrast, the sensorimotor task, as shown in Figure 2, elicited a crossed (i.e. predominantly contralateral) pattern of cortical activation, as expected from normal neuroanatomy (Truex and Carpenter, 1969) and consistent with findings in control subjects studied in our laboratory (J. Hirsch, M. Ruge, K.H.S. Kim, M.M. Souweidane and P. Gutin, unpublished data). Thus, we found no evidence of anomalies in sensorimotor pathways corresponding to, or induced by, the congenital miswiring of the visual pathways.
|
|
The subsequent imaging studies explored visual organization in greater detail. Figure 3A
shows responses to both left hemiretina (right hemifield) and right hemiretina (left hemifield) checkerboard stimulation OS, and illustrates the general observations for OS and OD. Voxels activated by each visual stimulation of each hemiretina occupied overlapping areas of primary visual cortex (orange), and neither hemifield detectably activated contralateral visual cortex or extrastriate areas. To determine the significance of this overlap, we performed the statistical analysis summarized in Figure 3B. The criteria in the multistage statistical analysis of each eye's image were allowed to vary, to provide a range of stringency levels (i.e. false-positive activations p). For each p-value (abscissa), the fraction of voxels activated by stimulation of left and right hemiretina (pL and pR) within the occipital region of interest was determined, as was the fraction of voxels activated by both eyes( pLR ). The region of interest for this calculation consisted of a volume of 1280 voxels (16 x 16 x 5) within the brain, located by anatomical criteria alone (aligned T1 images), without reference to the activation images. If joint activation were due to chance alone, then the probability of joint activation pLR would be approximated by the product of the probabilities of activation by either eye alone, pLpR. However, across more than a 1000-fold range of stringency levels, the number of common voxels observed was far in excess of the chance expectation. Thus we conclude that, to within the resolution of the imaging technique, stimulation of either hemifield activated overlapping regions of occipital cortex. Similar findings were present with right eye stimulation, but fewer voxels were activated.
|
Because the patient's oculomotor instabilities precluded a detailed analysis of the topography of the retinocortical mapping by conventional methods (Engel et al., 1997
), we used a customized, low-resolution strategy to demonstrate the presence of at least a rudimentary topographic map. We compared activations elicited by monocular viewing of a central 4° x 4° patch of a checkerboard (presumed to stimulate primarily the fovea and parafovea) with areas activated by full-field LED flashes (presumed to stimulate the entire retina, but to be a relatively less-effective stimulus for foveal and parafoveal retinal ganglion cells sensitive to spatial contrast). As shown in Figure 4, OS viewing of the central patch activated primary visual cortex predominantly along the medial bank inferiorly (slices 3–8), while OS viewing of the full-field flash activated primary visual cortex along the medial bank more superiorly (slices 10 and 11) and also along the outer convexity (posteriorly in slices 3–6). Areas of extrastriate cortex activated by OS viewing of these stimuli were also present (anterolaterally along the outer convexity in slices 3–11). With OD viewing, the areas of striate cortex activated by the full-field flash were also spatially distinct, and followed a pattern comparable to that obtained with OS viewing. With OD viewing, striate activation extended more superiorly and extrastriate activation was relatively less prominent, compared to OS viewing.
|
Viewing of illusory-contour stimuli with OD (Fig. 5
) elicited foci of activation in extrastriate areas of both hemispheres. However, the right-hemisphere foci were more extensive, and the focus of voxels activated in the left hemisphere was also activated by stimulation by the control (no illusory contour) stimulus, in which the corner elements occupied the same area of visual space but were rotated so that no illusory contour was present. Viewing of the illusory contour stimuli OS also produced bilateral foci of activation in extrastriate cortex. As was the case for OD viewing, the left-hemisphere focus was activated by the illusory contour stimulus as well as the control, while the right-hemisphere focus was activated only by the illusory contour. For the statistical criteria we used (P 
0.0005 to P 0.0001), the only cortical areas activated by the illusory contour stimulus but not by the control stimulus were in the right hemisphere (slice 5 for OD viewing, slice 6 for OS viewing, circled in Fig. 5). That is, with viewing either OD or OS, we identified a focus of right-hemisphere activation that was elicited by illusory contours but not by the control stimulus.
The observed patterns of activations induced by picture-naming tasks are complex, and are shown in Figure 6. With OS viewing (Fig. 6A), we observe areas of activation in ipsilateral striate (black circle) and extrastriate (black arc) cortex, and in the contralateral extrastriate (green arc) and striate (green circle) cortex. Activations were also seen more anteriorly in the left inferior frontal gyrus, putative Broca's area, and also in the corresponding portions of the right hemisphere. With OD viewing (Fig. 6B), a similar pattern of activation of both ipsilateral and contralateral striate and extrastriate cortices, as well the anterior areas described above, was seen. The activation of both Broca's area and its mirror image in this task has frequently been observed in control subjects in our lab while performing this task (Hirsch, Ruge, Kim, Souweidane and Gutin, unpublished data) and related tasks (Kim et al., 1997).
|
Comments
The fMRI responses to simple visual stimulation (Fig. 1) confirm the ipsilateral projection pathways of each eye to primary visual cortex, as expected from the anatomy of the patient's visual pathways and the analysis of the VEP responses (Apkarian et al., 1994a, 1995). As shown by statistical analysis of the activated voxels, areas of primary visual cortex activated by left and right hemifield stimulation of the same eye were overlapping, to within the resolution (1.6 x 1.6 mm in-plane) of the imaging technique (Fig. 3). Although the subject's eye movements precluded a detailed analysis of the topography of the retino-cortical map, the spatial separation of responses to central and full-field stimulation (Fig. 4) indicates that at least some topographic specificity is present. Moreover, the demonstration of distinct areas of activation by a 4° x 4° central checkerboard and a 12° x 12° patch (Fig. 4) demonstrate that eye movements alone cannot be responsible for the nearly complete overlap of the hemifield responses (Fig. 3). Sensorimotor pathways, as assessed by our functional imaging technique, showed the normal ‘crossed’ organization.
Illusory contour completion is part of the visual mechanisms for segmentation of figure from ground (Petry and Meyer, 1987), but is also a crucial step for visual object recognition. Previous work (Hirsch et al., 1995; Ffytche and Zeki, 1996; Mendola et al., 1997) indicates that extrastriate occipital cortex in humans is activated by illusory contour stimuli, and has suggested a right-hemisphere specialization for this process (Hirsch et al., 1995). The present results (Fig. 5) demonstrate in the achiasmat the presence of an extrastriate focus of activation induced by illusory contours. Moreover, cortical activation specific to the presence of the illusory contour is found predominantly in the right hemisphere, regardless of the viewing eye (and hence, regardless of side in which primary visual cortex is activated).
In some tasks, activation of contralateral visual areas was observed. This contralateral activation was progressively more prominent as the complexity of the task increased: for passive viewing of a checkerboard, the only contralateral activation observed was in extrastriate cortex, with OD viewing (Fig. 1). For illusory contour stimuli (Fig. 5), viewing with either eye elicited contralateral activation, but confined to extrastriate cortex. In the picture naming task (Fig. 6), viewing with either eye elicited contralateral activation not only of extrastriate cortex but of striate cortex as well.
|
Psychophysical and Evoked Potential Studies |
|---|
|
TopAbstractIntroductionClinical ProfileFunctional Imaging StudiesPsychophysical and Evoked...DiscussionAppendixReferences |
|---|
Vernier Acuity
Motivation
As demonstrated above, each striate cortex in this patient receives projections from both ipsilateral and contralateral halves of space. Since vernier acuity requires precise geometrical information, elevated thresholds would be expected if such information were not preserved by this abnormal mapping. Furthermore, since each striate cortex represents a visual field of twice the normal extent, elevation of vernier acuity thresholds out of proportion to acuity loss might be expected from amblyopia-like ‘crowding’ (Levi and Klein, 1985).
Methods
Stimuli consisted of abutting horizontal (8 x 64 min) bars at a contrast of 0.25, presented for 1 s on a mean luminance background of 150 cd/m2. Control signals for the visual stimulator (Tektronix 608) were produced by electronics modified from the design of Milkman et al. (Milkman et al., 1980). Viewing distance was 57 cm and the display size was an 8° x 8° square. In each trial, one of the two bars appeared vertically displaced. The patient was asked to indicate whether the left or the right bar was higher. A staircase algorithm (initial step size, 0.3 log units; final step size, 0.1 log units) controlled the bar displacements, and a 71% correct threshold was estimated from each staircase as the geometric mean of the final eight reversals. A final estimate of the 71% correct threshold for each viewing condition (OS, OD and OU) was taken as the geometric mean of six runs for each condition, presented in counterbalanced order.
Results and Comments
Displacement thresholds were: OS, 0.38 min; OD, 0.21 min; OU, 0.25 min. The difference between the thresholds for OS judgments and OD or OU judgments was not statistically significant (t-test, P > 0.05). These thresholds were in the hyperacuity range, and directly comparable to an OU threshold mean of 0.28 min obtained from six normal subjects (aged 20–38) studied under similar stimulus conditions. Thus, no abnormality in the precision of the retinocortical map or the ability of the patient to access precise geometrical information was identified.
Completion Tasks: Illusory Contour Formation and the Street Test
Illusory contour formation, along with vernier acuity, makes use of precise geometrical information. Additionally, illusory contour formation is considered to be a critical first step in segmentation of the visual field into objects (Petry and Meyer, 1987). To study illusory contour formation, we used the Kanizsa square stimuli similar to those used in the fMRI studies, and also a more complex task, the Street test (Street, 1932) of ‘Gestalt completion’ (Wasserstein et al., 1987). Performance on the Street test appears to depend in part upon the same mechanism as that studied with illusory contours, because completion of many of the incomplete contours in the Street figures involves illusory contour formation. However, the Street test also requires placing these completed contours in context as well as object recognition. On both tests (see Appendix for details), performance was normal with either eye.
Contrast Sensitivity and Velocity Discrimination for Moving Gratings
Motivation
Ocular instabilities might directly affect contrast sensitivity (as a result of image blur), or velocity discrimination (as a result of retinal image motion). Additionally, the patient's sole visual complaint was of difficulty judging velocities. To determine whether these deficits were present, and to separate the effects of elevated detection thresholds on velocity discrimination, we measured detection thresholds for moving gratings and velocity discrimination thresholds at a fixed multiple of the measured detection thresholds.
Methods
To determine the detection threshold for drifting gratings, the patient was asked to determine which interval of two 1.25 s interval presentations contained a sinusoidal luminance grating (1 c/°, 4°/s drift velocity); in the other interval, the display remained at the mean luminance of 150 cd/m2. Viewing distance was 57 cm and the display size was an 8° square. A two-alternative, forced-choice staircase algorithm (initial step size, 0.3 log units; final step size, 0.1 log units) controlled the contrast of the grating. (Contrast is defined as the Michelson contrast [(Imax – Imin)/(Imax + Imin)].) In this study, during the monocular viewing conditions, the non-viewing eye was lightly patched with a translucent gauze.
A 71% correct contrast threshold was estimated as the geometric mean of the final eight reversal points on each staircase. For each viewing condition (OS, OD and OU), and direction of drift (upwards, downwards, leftwards and rightwards), geometric means from two staircases formed the final threshold estimate. A similar procedure was used to determine contrast thresholds for 4 c/°, 4 Hz flickering gratings.
Velocity discrimination was measured via a two-interval forced-choice method adapted from Taub et al. (Taub et al., 1997). Stimuli consisted of pairs of sequentially presented drifting gratings, each chosen from one of five velocities (2.0, 2.8, 4.0, 5.6 and 8.0°/s), and sharing the same drift direction. The patient was asked to determine which interval contained the faster grating. Grating spatial phase was randomized, and spatial frequency and presentation time were jittered by 10% to prevent the use of cues other than velocity. The contrast was set to equal 10x the threshold determined in the grating-detection experiment. A block of trials consisted of 75 such presentations for a single drift direction. On each of two days, one such block of trials was run for each viewing condition (OS, OD and OU) and each of the four drift directions. Blocks were run in an order that counterbalanced learning effects. For each condition, velocity discrimination was summarized by a Weber fraction (i.e. the fractional change in velocities necessary for 75% correct discrimination), derived from the staircase data by the method of Taub et al. (Taub et al., 1997).
Results
Contrast thresholds (mean of two staircase determinations) for detection of drifting 1 c/° gratings are shown in Table 1. For vertically drifting (horizontally oriented) gratings, monocular and binocular thresholds were similar, and within the range of normal subjects in our laboratory. For horizontally drifting (vertically oriented) gratings, thresholds were ~2-fold higher than for horizontally oriented gratings when viewed OD and OU, and 3-fold higher when viewed OS.
|
For velocity discrimination, there were no consistent differences between leftward and rightward drifting gratings, or between upward and downward drifting gratings, so these pairs of conditions were pooled. As summarized in Table 2
, there were large and consistent differences between horizontal and vertical drifting gratings. First consider the velocity discrimination Weber fractions obtained from the full range (2–8°/s) of stimuli. For vertical drift, Weber fractions (0.35–0.37) were similar under the three viewing conditions (OS, OD, OU), but were approximately twice the velocity discrimination Weber fraction of 0.17 obtained for three normal subjects (aged 20–38) studied under similar conditions (Taub et al., 1997). For horizontal drift, there was also little dependence on viewing condition, but the Weber fractions (0.62–0.81) were ~4-fold higher than the values obtained in normal subjects.
|
Examination of the psychometric surfaces across velocities (Fig. 7
) reveals important information that is not evident from the overall Weber fractions. These contour surfaces indicate the fraction of times that one stimulus is judged as faster, as a function of the pair of velocities being compared. The more closely spaced the contour lines, the greater precision of judgment (because a smaller ratio of velocities is required to lead to consistent judgments). Figure 7 thus indicates that velocity discrimination is poorer at low velocities than at high velocities, in that there is a greater spread of the contour lines from the main diagonal. This is most dramatic for the horizontal-drift conditions. With OD viewing, velocity comparisons of 2.0 and 2.8°/s gratings are at chance (no contour lines separating this point from the main diagonal), while velocity comparisons of 5.6 and 8.0°/s gratings are >70% correct (two contour lines' separation from the diagonal). With OS or OU viewing, velocity comparisons of 2.0 and 2.8°/s gratings are ~60% correct, while velocity comparisons of 5.6 and 8.0°/s gratings are >80% correct.
|
Weber fractions for velocity discrimination calculated for low velocities (2.0–4.0°/s, Table 2
) and high velocities (4.0–8.0°/s, Table 2) confirm these observations. For low velocities, impairment is remarkably large. Weber fractions are 3–4x normal for vertical drift. For horizontal drift, the lowest Weber fraction measured is 0.94 (OD). This is 6x normal, and indicates that velocity discrimination requires a 2-fold velocity difference. For OS and OU viewing of low-velocity horizontal drift, discrimination of even a 2-fold velocity difference is close to chance, so that a precise value of the Weber fraction cannot be reliably estimated (though one can be sure that it is >1, and thus at least 6x normal). In contrast, performance for high velocities (Table 2) is more nearly normal, with Weber fractions only 1.5x normal for vertical drift and 2.5x normal for horizontal drift. Comments
Contrast sensitivity for horizontally oriented gratings was within the normal range, but contrast sensitivity for vertically oriented gratings was markedly reduced, consistent with the effects of greater retinal image blur for vertical targets because of ocular instability. Velocity discrimination was also much more impaired for vertical targets than for horizontal targets. Although the patient's fixation behavior was highly abnormal, the positional accuracy of the eyes was sufficient to perform the task, in that she had well-defined foveation periods with a target accuracy of at least ±30 min in both horizontal and vertical planes (Apkarian et al., 1999). Near-chance psychophysical performance was specific to horizontal drift, and occurred only for slow-velocity stimuli, which were randomly intermixed with the high-velocity stimuli. Neither acuity nor contrast sensitivity can account for these findings, since each stimulus was presented at a contrast set to 10x its measured detection threshold.
Although eye movements were not recorded during this task, the patient's oculomotor profile allows analysis of the extent to which performance may be accounted for by the intrinsic eye movement instabilities. The velocity range in which psycho-physical performance fell to chance (<4°/s) corresponds closely to the minimum average eye velocities during foveation periods (Conte et al., 1998, Apkarian et al., 1999). In congenital nystagmus patients, non-fixational viewing also has been shown to contribute to judgements of flicker (Waugh and Bedell, 1992), based on the normal temporal contrast sensitivity functions. Such non-fixational viewing strategies may account for the worse performance for horizontal motion in this patient, since horizontal eye velocities were greater than vertical eye velocities during non-foveation periods, but not during foveation periods (Conte et al., 1998; Apkarian et al., 1999). Additionally, horizontal eye movement velocities across recording epochs are larger with binocular viewing than with either monocular viewing condition. These results may correspond to the psychophysical observation (Table 2) that, at low velocities, horizontal drift judgements are at chance with OU viewing, even though they are above chance with OD viewing.
Kinetic Depth Effect
In the kinetic depth effect (KDE) (Wallach and O'Connell, 1953), the percept of a three-dimensional object is evoked by appropriately moving dots. In response to the KDE display, the patient reported a vivid shape percept (the normal response), and described the depicted three-dimensional shape accurately. The normal perception of a three-dimensional shape in the KDE indicates that the achiasmat's absence of stereopsis (see Clinical Profile) relates to the source of the three-dimensional information, rather than the ability to generate three-dimensional percepts per se. That is, brain mechanisms for extraction and representation of three-dimensional information appear to be relatively normal, even in the absence of binocular cues.
The velocities of the dots used in this study were in the range in which the patient's ability to discriminate velocities was impaired. This does not necessarily mean that velocity discrimination and implicit use of velocity differences are dissociated, since velocity difference thresholds for the KDE effect were not measured.
Component and Plaid Motion
We sought to determine whether visual stimuli placed in corresponding areas of left and right hemifields interacted, and whether the patient would attend preferentially to visual input from the ipsilateral or contralateral hemifields when these inputs were in conflict. We reasoned that a superimposition of visual inputs from drifting gratings in corresponding points of the hemifields might lead to a combination of the gratings to form plaids (Adelson and Movshon, 1982), even if the anomalous superimposition were not in perfect spatial registry. As detailed in the Appendix, this study provided no evidence for anomalous interaction of inputs from the two halves of visual space.
Visual Evoked Potentials
Motivation
Monocular full-field VEPs have been shown effective in demonstrating the misrouting of visual projections in albinos and achiasmats (Apkarian, 1991a,b; Apkarian et al., 1995), but some visual stimuli (high-contrast transient pattern-onset checkerboards) demonstrate the misrouting more effectively than others. Checkerboard stimuli yield more robust VEP responses than comparably scaled stripes or grating patterns both in normal subjects, primarily due to the greater density of contours (Spekreijse et al., 1973; Van Der Tweel, 1979), and in patients with oculomotor instabilities and/or optical aberrations such as astigmatism (Spekreijse, 1980; Apkarian, 1994). On a more neurophysiological level, the early portions of pattern-onset VEPs have a localized topography that typically is well described by one or two dipoles (Maier et al., 1987; Van Dijk and Spekreijse, 1989), while stimuli such as flash and pattern-reversal VEP's generate responses from multiple areas and populations and require multiple dipoles to account for their scalp surface topography (Wood and Allison, 1981; Kraut et al., 1985; Ducati et al., 1988).
Thus, the observation that the VEP evidence of asymmetry is stimulus dependent reflects the multiple functional pathways by which visual information is transmitted to the brain and the fact that the contribution of these pathways depends on the visual image. To further investigate this dependence, we used VEP methods [isodipole texture interchange stimuli (Victor and Zemon, 1985; Victor, 1986) and windmill-dartboard stimuli (Zemon and Ratliff, 1982)] that separate responses related to processing of visual form from responses driven by luminance and/or contrast. These methods compare responses elicited by stimuli that are matched in spatial frequency content, and thus any resulting differences in the scalp topography must be due to neural processing, rather than optical or oculomotor factors.
Methods
Visual stimuli were presented on a background of 150 cd/m2, and subtended an 8° square at the viewing distance of 57 cm. Isodipole textures (Victor and Zemon, 1985; Victor, 1986) were presented at a contrast of 0.4, a temporal frequency of 4.22 Hz and a check size of 16 min. Windmill-dartboard stimuli (Zemon and Ratliff, 1982) were presented at a contrast of 0.3. The VEP was recorded via standard gold cup electrodes placed on the scalp in a chain at 2.5 cm intervals across the occiput, from 5 cm to the left of Oz (Jasper, 1958) to 5 cm to the right of Oz, with recordings referred to a midline frontal lead placed at Fz. Signals were filtered (0.1–100 Hz), digitized at the frame rate of 270.3 Hz and averaged, with sweeps that contained artifacts (recognized as large voltage excursions) excluded. For each stimulus, three trials of 30 s were obtained. Fourier components were calculated from each of these trials, and confidence limits were obtained from the three trials via the Tcirc2 statistic (Victor and Mast, 1991). Each trial was run in three viewing conditions (OS, OD, OU).
Results and Comments
The fundamental (first harmonic) response elicited by isodipole interchange isolates intracortical processes related to extraction of visual form (Victor and Zemon, 1985; Victor, 1986). Its scalp topography, shown in Figure 8, demonstrated an ipsilateral projection pattern. OD viewing produced the largest response over the right occiput; OS viewing produced the largest response over the left occiput. Contralateral responses were small (<0.5 µV), thus precluding evaluation of interhemispheric phase shifts. These results suggest that this intracortically generated VEP response (Victor and Zemon, 1985; Victor, 1986; Victor and Conte, 1991), is driven by a purely ipsilateral geniculocalcarine projection.
|
The second-harmonic responses to interchange between members of identical isodipole texture ensembles is dominated by responses to local contrast, and does not require intracortical processing (Victor and Zemon, 1985
; Victor, 1986). As shown in Figure 9, this response is largest over the right hemisphere for all viewing conditions, and does not demonstrate the ipsilateral-only projection pattern. Similar findings were obtained for the windmill-dartboard stimulus (data not shown), which generates responses due to local contrast interactions (Zemon and Ratliff, 1982). For this stimulus, responses were sufficiently robust across the electrode chain to examine interhemispheric differences in response timing via the phases of the Fourier components. As seen in Figure 9, OS responses occur earlier (i.e. have a phase shift that is less negative) over the right hemisphere than over the left hemisphere. OD responses occur earlier over the left hemisphere than over the right hemisphere. The phase lead of the contralateral response is ~0.4
radians (OS) and ~0.25 radians (OD). This corresponds to an effective latency reduction (at 2f = 8.45 Hz) of 24 ms (OS) and 15 ms (OD).
|
In summary, these findings indicate that the VEP correlates of intracortical pattern processing show an ipsilateral-only pattern (Fig. 8
), while VEP signals that include contributions from contrast and luminance are distributed bilaterally (Fig. 9). The most direct interpretation of these findings is that, in addition to the ipsilateral-only geniculocalcarine projections that support pattern extraction (Fig. 8), there also are contralateral (or perhaps bilateral) projections that convey contrast and/or luminance information but do not contain sufficient detail (or are not used) to support the extraction of pattern. But it is also possible that the source structure is entirely ipsilateral in both cases, and that contrast and luminance signals arise from a more complex pattern of dipole sources than pattern-processing signals and thus produce a bilateral scalp field (Shagass et al., 1976; Barrett et al., 1976; Skrandies and Lehmann, 1982; Maier et al., 1987). Whatever the basis of their distinct topographies, the more widely distributed contrast and luminance responses have a shorter latency than the pattern-specific responses, both in normals (Victor and Zemon, 1985) and in this patient (Fig. 9). Thus, callosal transfer from an ipsilateral-only retinocortical projection pattern could not account for our results, since the second-harmonic responses (Fig. 9) over the contralateral hemisphere appear earlier than over the ipsilateral hemisphere.
Responses during binocular (OU) viewing were similar to responses elicited through OD alone, rather than what would be predicted by vector summation of OS and OD responses. This electrophysiological indication of suppression of OS signals was seen for the isodipole responses (Figs 8 and 9) as well as for the windmill-dartboard responses (not shown). Quantitative analysis of the non-additivity of monocular contributions (not shown) indicated that it was primarily present over the left hemisphere. In primary gaze, the patient has a horizontal misalignment of 20–25 prism diopters, and preferentially fixates with the right eye. If this misalignment is manifest under the viewing conditions for the VEP (8° field size at 57 cm), then the stimulus might well be viewed by the perip