| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 From the Sektion Visuelle Sensorik, Department of Neuro-ophthalmology, University Eye Clinic, Tübingen, Germany; the 2 Department of Optometry and Visual Science, City University, London, United Kingdom; and the 3 Institute for Brain Research IV, Human Neurobiology; University of Bremen, Germany.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
METHODS. The method was improved in two ways. First, the grain of the visual noise was increased toward the periphery of the visual field to accommodate the peripheral decrease in visual acuity. Second, the type of stimulus pattern was varied to include separate investigations of different visual components or functions (color, motion, temporal resolution, line orientation, stereoscopic depth, acuity, and figure–ground segmentation). To evaluate the reliability of the method, the visual fields were compared, as assessed by the new method, with those of conventional perimetry in 41 patients with neurologic disorders and 22 normal control subjects.
RESULTS. The results were encouraging. All patients with suprageniculate lesions subjectively experienced visual field defects in component perimetry. Sizes of visual field defects obtained with both methods corresponded qualitatively with each other, with a highly significant correlation. The specificity of component perimetry was higher than that of the original noise field campimetry.
CONCLUSIONS. This pilot study indicates that component perimetry is a subjective but relatively reliable method for detecting disorders of visual perception caused by lesions at different stages along the visual pathway, permitting fast screening of the visual field. In addition, this method seems to allow examination of the visual field, not only for defects in contrast sensitivity, as does conventional light perimetry, but also for the status of other components of vision such as color or motion perception. Further evaluation with larger patient cohorts is needed to allow exact assessment of the clinical usefulness of the method.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
In conventional light perimetry, defects are detected sequentially in the visual field by using point-by-point testing. This type of perimetry is time consuming and depends on the attentiveness of the patient. Aulhorn and Köst1 developed, to the best of our knowledge, the first method of truly simultaneous perimetry, termed noise field campimetry, which tests a large part of the visual field simultaneously. The noise field is an area filled with small black and white dots randomly distributed and flickering at high frequencies, similar to the snowstorm on a television screen after the end of transmission (Fig. 1A , pattern 1). When patients fixate a central spot in the noise field, scotomata are perceived subjectively as circumscribed areas with less or no flicker and sometimes with a brightness differing from the surround. The advantages of noise field perimetry are the simultaneous examination of the whole visual field and the subjective experience of visual field defects, allowing the development of a fast screening method for visual field defects. However, noise field campimetry has two disadvantages in comparison with conventional perimetric methods: First, it has somewhat low sensitivity for homonymous hemianopias caused by suprageniculate damage, which are either not perceived at all in the noise field or are perceived over a much smaller spatial extent1 2 3 4 ; and second, the method cannot quantitatively determine the exact size, location, and depth of visual field defects.
|
The concept of examining different visual functions simultaneously in the whole visual field is based on the finding that humans can discriminate between certain elementary stimulus features such as luminance, color, orientation, motion, and stereoscopic depth in parallel. In our opinion, the only way to achieve this feat is by using a large number of dedicated processors for the different components of visual perception in parallel, one for each visual component and field position.5 6 7 8 9 This concept is supported by both electrophysiological10 11 12 13 and neuroanatomical data.14 15 16 For many visual features, processors seem to be organized in a retinotopically ordered map, and each individual map can be tested individually and simultaneously by the appropriate perimetric pattern. There is also evidence from neuropsychological studies that visual perception depends on several relatively autonomous mechanisms, each of which contributes distinct components. Various studies report selective disturbances or even losses of visual functions after circumscribed cerebral damage in humans, resulting, for example, in isolated achromatopsia, loss of color vision, or impaired perception of motion.17 18 19 20 21
The disturbance of one elementary visual function, for example color perception, does not always lead to a pathologic visual field when tested with conventional light perimetry.18 22 23 Testing different visual components or functions such as color, motion, stereoscopic depth, orientation, visual acuity, or figure–ground segmentation is impossible with conventional perimeters and would be time consuming if done sequentially, as in conventional perimetry. Therefore, we developed new perimetric stimuli to reveal specific visual field defects. These stimuli selectively test different components or functions of visual perception such as color, motion, temporal resolution, line orientation, stereoscopic depth, acuity, and figure–ground segmentation (Fig. 1) . Moreover, element size of the (noise) pattern for those parts stimulating the peripheral visual field was progressively increased with eccentricity to achieve a stimulation equally far above resolution threshold at each position of the visual field.24 25 In this pilot study we assessed the ability of component perimetry as a screening test to detect (absolute) visual field defects caused by suprageniculate damage by testing patients who had homonymous hemianopia and, as control groups, patients with neurologic but no visual disorders and intact visual fields, as well as normal control subjects. In addition, we evaluated the reliability of the new method by comparing the results of component perimetry with those of conventional light perimetry.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
Types of Perimetric Stimuli
Thirteen perimetric stimuli served to test different submodalities
of vision (Fig. 1)
. Figures 1A
1B
1C
1D
1E
display the perimetric
patterns as static representations of one frame of the stimulus,
whereas the remaining perimetric stimuli (Figs. 1F
1G)
are shown
schematically. All patterns were generated by computer (Macintosh;
Apple Computer, Cupertino, CA) and presented on a 20-in. color monitor
(Trinitron Multiscan; Sony, Tokyo, Japan; 20se, 40 cm wide and
30 cm high). We examined a visual field area of approximately ±45°
horizontally and ±36° vertically at a viewing distance of 20 cm.
Correcting lenses were used if necessary.
Noise Field Perimetry: Contrast Noise Field.
Pattern 1 in Figure 1
shows a static representation of one frame of a
noise field stimulus. Black and white dots flickered independently and
at irregular intervals at high temporal frequencies. Each white or
black pattern element turned to black or white randomly after one or
two frames, so that not all pattern elements changed luminance at the
same time (see Appendix A). In the case of homogeneous dot size (Fig. 1A
, pattern 1), the stimulus resembled the flicker on a television
screen after the end of transmission, eliciting association with a
snowstorm. Pattern 2 in Figure 1A
shows a modification of the noise
field with element size increasing toward the periphery of the visual
field to compensate for the increasingly poorer resolution of the
periphery.
Noise Field Perimetry: Color Noise Field.
To test the color-sensitive mechanisms of the visual system, the noise
field was composed of either red and green or blue and yellow dots. The
two types of elements in these color noise fields were isoluminant, and
color processing was therefore required for the perception of flicker,
whereas a black-and-white system would have perceived a stationary gray
area. The red-and-green or blue-and-yellow noise fields with magnified
elements toward the periphery are shown as patterns 3 and 4 of Figure 1A
. (The point of isoluminance was determined in a way analogous to
flicker photometry. A red-and-green or blue-and-yellow random dot
pattern was presented flickering at a frequency of 12 Hz. The pattern
had the same size as the patterns used in component perimetry.
Observers adjusted the relative intensity of one of the two colors
until the subjective sensation of flicker was
minimal.26
27
Because the point of isoluminance varies
slightly between the fovea and the periphery, strict isoluminance was
restricted to the central visual field in our investigation.)
Color Perimetry: Colored Dots.
The colored dots represent another stimulus to test color perception.
The dots were isoluminant with the background—that is, the background
differed from the dots in color, but not in luminance. Either red (Fig. 1B
, pattern 5) or yellow dots appeared and disappeared on a green or
blue background. The time course of this color modulation followed a
sine wave function, and the frequency of the flicker was relatively low
(
0.01 Hz; Appendix A).
Acuity Perimetry: Rotating Landolt Cs.
Stationary Landolt Cs are a classic tool for visual acuity tests. To
test visual acuity over the whole visual field, the Landolt Cs (Fig. 1C
, pattern 6) rotated around their centers at a constant speed. The
rotation could be perceived only if the gap in the Landolt C was being
resolved. The gaps started from random orientations to prevent or
complicate potential filling-in processes. The size of the individual
rotating Landolt Cs increased toward the periphery, because visual
acuity is directly proportional to the cortical magnification
factor,28
29
(i.e., to the extent of area in the
primary visual cortex devoted to each part of the visual field).
Orientation Perimetry: Rotating Lines.
The rotating lines served to test motion perception in combination with
orientation perception. Differently oriented lines (Fig. 1D , pattern 7)
rotated at identical angular speed around their center axes. If
orientation and/or motion perception was defective in parts of the
visual field, the perception of this rotation should be impaired in
those areas. Similar to the gaps of the rotating Landolt Cs, the lines
started rotating from random positions.
Filling-in Perimetry: Interrupted Lines.
Lines interrupted at random positions were arranged to resemble sun
rays around the central fixation point (Fig. 1E
, pattern 8). To prevent
local adaptation, we presented the interrupted lines with a
counterphase flicker, so that line elements appeared at positions
previously held by interruptions, and vice versa. The design of this
pattern is based on the following observation: If the gap between two
aligned line elements is projected exactly onto the blind spot, we have
the impression of a straight line without any gap; the gap is
subjectively filled in by the visual system. If such a filling-in
mechanism is also active in cases of cortical scotomata, patients
should have the impression of stationary straight lines rather than
interrupted lines within defective areas of the visual field.
Motion Perimetry: Coherent Motion.
This pattern contained a large number of randomly distributed moving
dots, with only half of these dots moving coherently in the same
direction and the remainder moving in the opposite direction (Fig. 1F
,
pattern 9). Observers had the impression that the dots moved on two
different planes: Fifty percent of the dots seemed to move on a
transparent plane, in front of the background plane defined by the
remaining 50%. This effect of depth separation could be achieved only
by integrating motion information over (parts of) the visual field.
With similar coherent motion stimuli, selective impairment of motion
perception was found in primates with lesions of the middle temporal
area.30
Patients who have a unilateral lesion and/or one
that does not affect the whole middle temporal area should have such a
motion perception deficit in part of the visual field only. In these
parts, the impression of depth should disappear.
Segmentation Perimetry.
Our visual system can divide a pattern consisting of many small
elements into figure and ground if the local elements within the figure
differ in features such as luminance, color, depth, orientation,
temporal information, texture, or motion.31
32
33
34
It is
supposed that this segmentation is achieved in parallel at an early
stage of the visual system.6
7
35
36
Therefore,
simultaneous testing of figure–ground segmentation in the whole visual
field should be possible. The following perimetric stimuli are based on
the concept that an extended (global) stimulus can be segmented into
figure and ground only if the visual processing mechanism (e.g., motion
perception) for the component defining the global pattern is intact.
If, however, the corresponding mechanism contains a defect within at
least part of the visual field, a patient should only see randomly
distributed local elements within the area of defect without detecting
any global structure.
Motion-Defined Checkerboard.
A checkerboard was defined by motion information: Points in neighboring
checkerboard areas moved in directions at right angles (Fig. 1G
,
pattern 10). Although there were no line borders between the areas,
observers had the impression of clear boundaries separating the fields
of the checks. These boundaries were purely motion defined.
Time-Defined Checkerboard.
Neighboring checks differed in the presentation time of the small
points displayed within each check (Fig. 1G
, pattern 11). This stimulus
corresponded to a counterphase flicker checkerboard without differences
between the luminance of the checks (i.e., there were no black or white
checks). Only patients with a sufficiently precise temporal resolution
perceived a checkerboard over all the stimulated visual field.
Depth-Defined Checkerboard.
In this stimulus, the only difference between the two types of checks
relied on stereoscopic depth (Fig. 1G
, pattern 12). Points in
neighboring areas were presented in a different plane, rather than with
a different luminance, as in a conventional checkerboard. With intact
stereoscopic vision, observers had the impression of a checkerboard
consisting of checks located at two different planes in depth.
Orientation-Defined Checkerboard.
This checkerboard was defined by the orientation of its line elements.
The only difference between the checks of this stimulus was in the
orientation of the short lines within each area (Fig. 1G , pattern 13).
If patients were not able to analyze line orientation or to group
elements of the same orientation, they could not recognize the global
checkerboard pattern.
Stimulus Parameters
Luminance, contrast, size of pattern elements, increase of
elements toward the periphery, temporal frequency, and further
characteristics of the stimuli are listed in Appendix A. Contrast
(C) is always expressed as Michelson contrast37
and
is expressed as a percentage:
 |
For the tests with depth-defined checkerboards, the visual fields of both eyes were separated by means of spectacles with LCD shutters that alternately opened and obstructed the field of view of both eyes in counterphase at a frequency of 75 Hz. Special software and hardware doubled the number of frames presented per second and produced 2 x 75 images on the monitor screen, with small disparities between the images displayed to the two eyes, thus creating a stereoscopic depth impression.
All patterns could be displayed in two ways: with a homogeneous element size of checkerboard or pattern elements (Fig. 1 , pattern 1) and with an increasing size of checkerboard or pattern elements (Fig. 1 , patterns 2 through 8). Central element size was clearly above threshold even for older observers, and size increased with eccentricity roughly according to the magnification factor28 39 of the human visual system, to achieve a stimulation equally far above threshold at each position of the visual field (Appendix A).
Patients and Controls
Component perimetry was tested in 41 patients with
neuro-ophthalmologic disorders and 22 normal control subjects.
Inclusion criteria for patient selection were a good general state of
physical and mental health, normal or corrected-to-normal visual
acuity, no cataract or glaucoma, no disease of the retina or optic
nerve, no severe oculomotor disorders (e.g., inability to maintain
fixation), no visual neglect or other deficits of attention, and no
sedative medication. Clinical details of the patients and/or results of
the different standard tests for basic visual functions (conventional
light perimetry, visual acuity, color vision, and stereoscopic vision)
are provided in Table 1
. An automatic perimeter (Tübinger Automatic Perimeter
[TAP]; Oculus, Wetzlar, Germany) was used for conventional light
perimetry (either before or after component perimetry testing), and
visual acuity was checked with Landolt Cs. Central color vision was
tested using Ishihara plates, and central stereo vision tests was
tested using the Lang and Titmus stereo tests (Oculus). The anatomic
analyses were based on computed tomographic or magnetic resonance
imaging scans.
|
In all 22 subjects of the normal control group basic visual functions were also tested with the different standard tests described earlier (Table 2) , and the visual field was determined to be normal by conventional perimetry (Tübinger Automatic Perimeter; Oculus). Inclusion criteria for controls were normal or corrected-to-normal visual acuity, normal color and stereo vision, an intact visual field, and no ophthalmic disease.
|
Data Analysis
To compare the results of component and conventional light
perimetry, we determined the size of visual field defects revealed by
the different perimetric methods. As a first step, the margins of
visual field defects, as they appeared on the patients’ drawings after
component perimetry, and the printouts of conventional light perimetry
were fed into a computer by means of a digitizing tablet and
transformed to a common format. To obtain the binocular visual field as
revealed by conventional perimetry, we superimposed the printouts of
both eyes to determine the margins of the homonymous visual field
defect. Figure 2
illustrates the differences between the testing situations and the
scaling of length in the printouts. Test stimuli are presented in a
sphere in conventional perimetry and, in contrast, on an almost flat
monitor in component perimetry. In a sphere the length of line
a' seen under a given visual angle is independent of
eccentricity, whereas on a monitor the length of line a seen
under the same visual angle increases with eccentricity. After
transformation, the length of line a', seen on the printouts
under a given visual angle, was also independent of eccentricity.
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
|
Healthy Control Subjects and Control Patients
As a control, 22 normal observers (Fig. 3D)
and 12
patients (Fig. 3C)
with brain lesions not affecting the visual pathway
were examined with component perimetry. Some controls described areas
where perception deviated (indicated schematically at the bottom of
Fig. 3 ). Those areas counted as pathologic (Figs. 3C
3D
; dark bars) if
they extended asymmetrically around the fixation point, that is, if
they were similar to the results described by the patients who had
visual field defects. However, those areas counted as physiological
changes of visual performance across the visual field, if they lay
symmetrically around the fixation point (Figs. 3C
3D
; hatched
bars).
In the patient control group, false-positive results were mostly observed with the classic noise field (pattern 1), whereas no false-positive results were obtained with patterns 8 (filling-in perimetry), 12, and 13 (depth-defined and orientation-defined segmentation perimetry, respectively). Physiological changes were experienced with all patterns of component perimetry except pattern 8 (filling-in perimetry; Figs. 3C 3D ). The normal control group showed false-positive results only for the classic noise field (pattern 1), the rotating lines (pattern 7), and motion-defined segmentation perimetry (pattern 10; Fig. 3D ). On average, normal control subjects showed fewer false-positive and physiological results than the patient control group. Most of the controls, patients, and normal subjects perceived homogenous patterns without local differences and were therefore classified as having negative results (Figs. 3B 3C 3D ; light bars).
Statistics for False-Positive and False-Negative Results
To estimate the number of false-positive and
false-negative results using conventional perimetry as the gold
standard, all results from all subjects (patient groups 1 and 2,
patient controls, and normal control subjects) were pooled together in
Table 3
as follows:
|
Type B results
Type C results
Because conventional perimetry served as the standard method, results of five patients in group 2 showing visual field defects in conventional perimetry but an intact visual field for some functions tested with component perimetry were classified as false negatives, although this discrepancy was probably caused by selective visual field defects based on lesions of higher visual cortical areas not affecting detection of luminance differences as tested in conventional perimetry.
We calculated the sensitivity and specificity of component perimetry using conventional perimetry as the gold standard as follows (Table 3) :
Types of Results
Testing the visual field with component perimetry can lead to
three different types of visual field defect, as demonstrated in Figure 4 . For comparison, the corresponding visual field revealed by
conventional light perimetry is shown. When looking at the stimuli of
component perimetry, most patients outlined areas where they did not
experience any visual stimulation. Either sharp boundaries (Fig. 4
,
Type A) or transitional areas (Fig. 4 , Type B) separated these blind
areas from the intact visual field. Within the transition area,
patients always had the impression of degraded visual perception. Other
patients, particularly those with relative visual field defects,
perceived their whole visual field defect as an area of degraded vision
without any blind portions (Fig. 4
, Type C).
|
|
|
Conventional Perimetry versus Component Perimetry
To evaluate how well the results of component and conventional
perimetry agree, we calculated the correlation between the size of
corresponding visual field defects by component and conventional light
perimetry. The correlation was calculated twice: physiological results
were classified either as false positive (i.e., visual field defects;
Fig. 6 , dark bars) or as negative results (i.e., intact visual field; Fig. 6
,
hatched bars). (Because the data might not have been
distributed normally, we selected a nonparametric measurement of
association. The correlation coefficient r is given as
Spearman’s rho [rank order correlation] indicating relative rather
than absolute correlations.)
|
Moreover, we estimated the size and the degree of spatial overlap between corresponding visual field defects in component and conventional light perimetry. In Figure 7A the size of the visual field defects in component perimetry and the size of overlapping areas are represented as the proportion of the corresponding size of the visual field defects in conventional perimetry. The normalized average size of visual field defects (Fig. 7 , light bars) and the normalized average size of overlapping areas (Fig. 7 , dark bars) are shown for each perimetric pattern except pattern 6 (insufficient data to calculate paired t-tests). The difference between the light and dark bars corresponds to the nonoverlapping areas of the visual field defects.
|
For each component perimetry stimulus the proportion of results for the different categories of overlap is shown in Figure 7B . Except for pattern 10 (motion-defined checkerboard) there was at least a small overlap between all corresponding visual field defects. Unilateral suprageniculate lesions of the visual system cause visual defects only in the contralesional visual hemifield. A small overlap between corresponding visual field defects—and that comprises both hemifields—usually constitutes an unsatisfactory result. Testing the visual field with stimulus 12 (depth-defined checkerboard) led to a small overlap in approximately 50% of cases. With all other stimuli of component perimetry, more than 80% of the results agreed adequately or well with the results of conventional light perimetry (i.e., comprised only corresponding hemifields). Also for the categories of adequate and good overlap, the average size of visual field defects revealed by component perimetry was usually smaller than the corresponding visual field defect revealed by conventional light perimetry (Fig. 7C) . Defects are significantly smaller (Fig 7 , *on the 5% level; **on the 1% level) for pattern 2 (black-and-white noise field with increasing elements), pattern 3 (red-and-green noise field), pattern 9 (coherent motion), and patterns 10 and 13 of segmentation perimetry (motion- and orientation-defined checkerboards).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
Patients with Homonymous Absolute Visual Field Defects
A fast screening test should fulfill at least two conditions.
First, all visual field defects should be detected, and, second,
false-positive diagnoses should be minimized. The first condition, high
sensitivity, is fulfilled for visual field defects caused by lesions
beyond the optic chiasm. All visual field defects were detected by
component perimetry. In contrast to previous noise field campimetry
studies1
2
3
4
all our patients with absolute homonymous
visual field defects caused by suprageniculate lesions subjectively
perceived their defects. This was true for all perimetric stimuli
tested, even for the classic noise field.
As outlined in the introduction, earlier studies using the classic noise field as a test stimulus1 2 3 4 yielded a relatively low sensitivity for lesions caused by suprageniculate lesions. There are three probable reasons for the difference between our results and those of Aulhorn and Köst,1 2 Schiefer et al.,3 and Kolb et al.4 Although the stimuli used for the classic noise field were virtually identical (we in addition presented the noise field with element size increasing toward the periphery), there were differences in the instructions to the patients. The first difference among the studies concerns the way patients were introduced to the task. We told them not only to look for local differences within their visual fields but also to pay attention to the boundaries of the visual field. Patients with lesions of the retina or the optic nerve usually have scotomata completely surrounded by areas of intact visual field. These types of scotomata are clearly perceived in the classic noise field.1 2 40 In most patients with suprageniculate visual field defects, however, the defective area is not surrounded by an intact visual field, but the outer boundaries of the visual field are shifted toward the fovea, so that the visual field simply becomes smaller—similar in some respects to the effect of closing one eye. Patients with narrower visual field boundaries seem to habituate to the shrunken visual field. They certainly do not perceive local differences within the remaining intact visual field. But as soon as these patients are asked pay attention to the boundaries of the visual field and to draw them on a monitor, defects become apparent.
The second and more relevant reason for the differences between the studies may be the increasing element size used in most of the new perimetric stimuli. Most patients reported that the visual field defect is subjectively much more pronounced when the size of pattern elements increased toward the periphery. The increased salience of the defect leads first to a better detection of peripheral elements. Second, the whole stimulus configuration appears as asymmetric in the case of a peripheral defect, because of the absence of larger elements within the defective visual field area. Therefore, the increasing element size not only enables a better discrimination of pattern elements in the periphery but also allows the patients to better perceive and judge the homonymous hemianopia.
A third possible reason for the difference is that we used a larger stimulus area.
Size of Visual Field Defects and Areas of Overlap
The ratio between visual field defects calculated by component
versus conventional perimetry tended to be smaller than 1, with the
exception of the results obtained with pattern 12, the stereoscopically
defined checkerboard (Fig. 7)
. This is to say that the field defect was
smaller for component than for conventional perimetry. It seems that
the stereoscopic visual field of patients with homonymous visual field
defect was affected to a larger degree than the visual fields for all
other components tested. Although the average ratio was smaller than 1
for all other patterns, the deviation from 1 is only significant for
patterns 2, 3, and 9 (Fig. 7A)
. One reason for the smaller defect size
may be fixation instability by the patients during the test, another
one filling-in. As mentioned, patients were asked to outline the
subjectively perceived visual field defects while maintaining fixation.
If fixation was unsteady, the borders of the visual field defect would
have moved back and forth. This effect could lead to perception of
visual field defects as smaller than they really are. Unsteady fixation
can also influence the degree of overlap between corresponding visual
field defects. As the subjectively perceived defect moves back and
forth, it is hard to outline its correct position.
The noncorresponding areas between visual field defects revealed by the two methods are probably also influenced by the test conditions. The patients are not allowed to fixate the pen while outlining the defect. Figure 8 illustrates these possible influences on the degree of overlap by presenting our worst case. In the case of small eccentric scotomata, the unsteady fixation and the motor error during drawing can lead to a visual field defect in component perimetry not overlapping with the one revealed by conventional perimetry.
|
Especially in the case of patterns 9 through 13, the visual system has to analyze local elements before a global structure can be extracted by integrating over parts of the visual field. This task is more complex than the detection of dynamic visual noise, and higher cortical visual areas are probably involved. A relative visual field defect, as detected in conventional perimetry, may not impair the perception of dynamic visual noise but may influence visual functions needing additional processing steps (e.g., figure–ground segregation).
Although the number of subjects in patient group 2 was small, the results indicated that component perimetry may relatively selectively reveal defects of different components or functions of visual perception. Three of the patients in this group had brain lesions of higher visual areas not including the visual pathway up to V1. These lesions probably led to selective visual dysfunction in parts of the visual field selective for some visual functions, whereas others remained spared. Clearly, far larger numbers of patients must be tested to further clarify the selectivity of the new method.
Control Groups
The control groups served to evaluate the specificity of component
perimetry through the number of false-positive results. Most but not
all the controls did not experience any inhomogeneities (i.e., defects
in the stimuli of component perimetry). There were no abnormalities or
deficits in the visual field as measured by component perimetry.
Physiological Results
Detection thresholds are not homogeneous over the whole visual
field but increase from the fovea toward the periphery. Even the outer
borders of the visual field vary with the visual function
tested.41
Some subjects perceived these functional
differences between the central and the peripheral parts when looking
at some of the perimetric patterns. Several control subjects even
indicated the boundaries of their binocular visual field when tested
with the depth-defined checkerboard stimulus. They experienced depth
only within this binocular field, but not in the monocular field parts.
These physiological results do not represent classic false-positive
results. They represent inhomogeneities of the normal visual field that
most subjects are not aware of. Therefore, Figure 6
includes two
evaluations, the physiological results were either counted as normal
(Fig. 6
, hatched bars) or as false positive (Fig. 6
, dark bars). We
concluded that component perimetry detected not only pathologic visual
fields but also physiological inhomogeneities in normal fields.
Patient Data
In contrast to the results of the control group (normal observers
and patients) none of the patients with visual field defects described
areas of deviating visual perception lying symmetrically around the
fixation point. Perhaps the degraded perception within the areas of
visual field defects is so prominent in relation to the physiological
changes that the latter were not perceived by the patients.
The prevalence of pathologic or false-positive results was higher in the patient control group than in normal control subjects (see Table 3 ). This could be caused by lesions of extrastriate visual areas that do not affect the visual field in conventional perimetry but affect the field in component perimetry. In this case, the false-positive results would not really be false positive but the result of a higher sensitivity of component perimetry for this type of defects. Another possible cause of false-positive results is a disturbed microcirculation in the patient’s retina. Erb et al.42 suppose that perturbations of retinal blood flow can lead to temporary visual field defects perceptible in the classic noise field. Indeed, in all subjects of the patient control group the cause of the neurologic disease was a circulatory disturbance (stroke or hemorrhage).
Evaluation
The good detection of scotomata and the low number of
false-positive results in component perimetry encourages the use of
this new method to screen for different types of visual field
disruptions in patients with neurologic disorders. Of course, component
perimetry by itself can only qualitatively detect visual field defects.
Although we found a correlation between the results of component and
conventional perimetry, component perimetry was unable to measure the
exact size and location of visual field defects. The size of the
scotoma detected by component perimetry is generally smaller than that
found by conventional perimetry. One reason may be a partial filling in
of defects in higher cortical areas. In addition, a complete
correlation would not be expected between the results of conventional
and component perimetry for defects caused by cortical lesions. In
these cases, defects detected by component perimetry may be even larger
than those obtained by conventional perimetry, because detection of
differences in luminance may still be intact, whereas more complex
functions such as motion detection are impaired.
To supplement component perimetry as a screening method, more quantitative perimetric methods, comparable to conventional light perimetry, must be developed for the different components or functions of visual perception. Although component perimetry is a subjective and therefore qualitative perimetric method, all patients in our study were able to perceive their own visual field defects, and the results of the tests were reproducible.
On the basis of this pilot study, we conclude that component perimetry is a relatively reliable and sensitive method for detecting visual field defects simultaneously over large parts of the visual field. Therefore, a rapid screening of the visual field appears to be possible, not only for defects in the detection of differences in luminance, as with conventional light perimetry, but for other submodalities of vision such as motion and color perception. This study shows that component perimetry provides an effective tool to detect visual field defects, especially in patients with neurologic disorders, at least on an exploratory basis for screening purposes. Subsequent studies with component perimetry should investigate whether the method is also a clinically useful tool to distinguish between different disorders of visual perception and whether the results indeed provide information about the localization of lesions in the visual system.
|
Appendix A |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
|
|
|
Appendix B |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
is usually given as the surface of a unit sphere—that
is, in units of sterad. The steradian can be transformed into units of
arcdeg2 by the following formula
 |
a by a may be
calculated as
 |
 |
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Submitted for publication December 10, 1997; revised August 21, 1998, April 6 and September 14, 1999, and January 24, 2000; accepted February 15, 2000.
Commercial relationships policy: N.
Corresponding author: Manfred Fahle, Institute for Brain Research IV, Human-Neurobiology; University of Bremen, Argonnenstraße 3, D-28211 Bremen, Germany. mfahle{at}uni-bremen.de
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix AAppendix BReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
. Because patients
drew their visual field defects directly on the monitor screen the
printouts of component perimetry correspond exactly to the situation on
the monitor screen. On the printouts the radius r' for
conventional perimetry (right) and r for
component perimetry correspond to the extent of the visual field area
examined. 
: Size of the visual field examined in arc degrees
squared. Ratio (c) between the radius r'
on the printout and the visual angle of the visual field area examined
served as a scaling factor for the transformation procedure.
