ACTA O P H T H A L M O L O G I C A VOL. 55 1977
The Department of Experimental Ophthalmology (Head: C . E. T. Krakau), University Eye Clinic, Lund, Sweden
A NOTE ON FIXATION DURING PERIMETRY BY
A. HEIJL and C. E. T. KRAKAU
Small eye movements in the direction of the disappearing stimulus can be observed a t kinetic perimetry when the test object enters a scotomatous area. These may be responsible for the fact that several field defects, missed a t routine manual perimetry, are spotted a t automatic perimetry. In the latter case the patient cannot predict the position of next test light, since these are illuminated a t random, whereas at kinetic perimetry they are exposed in systematic order. By using two specially designed logics in automatic perimetry, it was shown that there may be a reduction of the scotoma size if the lights are exposed in a n ordered sequence. This effect is avoided by using a logic with randomly exposed stimuli. It is most likely that the difference can be attributed to ‘malfixation”.
K e y words: perimetry - kinetic perimetry
-
automatic perimetry - fixation.
In a recent study on glaucoma screening (Heijl 1976) it was shown that many visual field defects, previously missed at routine perimetry, kinetic or the Armaly technique (Armaly 1972) were detected by automatic perimetry. In a previous paper on a n automatic perimeter (Heijl & Krakau 197513) a few glaucomatous field defects were used to illustrate the performance of this instrument. I t was then observed that the defects sometimes seemed larger and/or deeper when plotted by the automatic device than with kinetic perimetry. Received March 29, 1977.
854
Fixation During Perimetry
A difference between the automatic and the manual test, which perhaps, at least partly, might explain these discrepancies, is the fact that the object appears at random locations during automatic perimetry but in a more or less ordered sequence in manual perimetry. Thus the patient is most often able to predict the position of the test object before it is seen and it might be tempting for him to change his fixation slightly when not observing the stimulus. In fact, a close look at the patients' fixation during manual perimetry reveals that fairly often, when a test object passes the border of a scotoma from a seeing to a non-seeing area, fast, small eye movements in the direction of the disappearing stimulus can be seen in the "telescope" of the Goldmann perimeter. As far as we know this phenomenon has not been explicitly described or analysed in the literature. The aim of the present note is to demonstrate the existence of this phenomenon by the recording of eye movements during kinetic perimetry and also to show how the phenomenon is provoked or avoided in automatic perimetry.
A. Kinetic Perimetry Material
Eleven patients with verified or suspected glaucoma were tested. Their ages ranged from 43 to 79 years (mean age: 68 years). Patients who were very trained perimetric subjects were avoided as were patients known to cooperate badly in perimetry. Nine pathological and two normal visual fields were examined. Some patients were chosen because their field defects had been missed by manual perimetry but detected by automatic perimetry. Only one eye was examined in each patient. Methods
Recording of eye movements. Two disposable AgAgCl electrodes (Medicotest A-15 M) were used. They were placed lateral to the outer canthus of each eye. A neutral electrode was placed above the left wrist. The temple electrodes were connected to a small preamplifier, the output of which was fed to an ink recorder (Mingograf 24B). The amplification of the recorder was so adjusted as to give a deviation of 10-20 mm in the ink-written curve for a horizontal eye movement of 30'. A small marker switch was connected to the second channel of the recorder. When pressed it produced markings on the ink-written line of this channel. Control of perimeter stimulus. The pantograph lever of the Goldmann perimeter was connected to a string driven by a n electrical motor controlled from a drive 855 55*
A . Heijl and C. E. T . Krakau
unit. Through this arrangement the pantograph lever could be made to move along different straight lines on the visual field chart at even speed. Experiment
The patient was placed in the Goldmann perimeter and provided with correction for ametropia and near vision. The electrodes were connected. A filled-in chart of the examined eye was put into the chart holder of the perimeter. The stimulus was positioned at an eccentricity of about 35”. The stimulus intensity was so adjusted as to be slightly supraliminal in this position. The patient was instructed to keep his gaze steady at the fixation target of the perimeter and to signal if the stimulus disappeared. The ink recorder and the movement of the stimulus were then started simultaneously and the stimulus travelled centripetally towards the fixation target of the perimeter a t constant speed (usually 3’/sec). A n assistant pressed the switch of the marker unit when the stimulus crossed the 30O-circle and the fixation target. A t least two meridians were tested - each five times - in every subject. One meridian crossed the blind spot or a known glaucomatous scotoma, the other passed only through seeing areas. The assistant was further instructed to press the button of the marker unit when any scotoma border was crossed. The patient’s fixation was constantly supervised in the “telescope” of the Goldmann perimeter. Results
Three patients demonstrated steady fixation in all tracings and one patient showed poor fixation in most tracings. In the remaining seven patients there was a distinct difference in fixation when the stimulus passed through seeing and non-seeing areas. These patients usually maintained good fixation when the stimulus moved in seeing areas, but when the stimulus passed into a scotoma they often, obviously unconsciously, changed their fixation in the direction of the stimulus, recognized the stimulus and quickly resumed fixation again (Fig. 1). Thus they never “lost” the stimulus, and did not signal its disappearance. These “scotoma-induced malfixations” were not only recorded by the ink recorder, but were also readily seen in the fixation-monitoring “telescope” of the perimeter. They were 7”-15” when the stimulus passed into the blind spot. Thus the patients sometimes changed their fixation all the way to the test object, but in some cases no more than to let the stimulus reach a seeing part of the visual field. Four out of these seven patients never recognized the stimulus as missing and mostly demonstrated malfixation each time the stimulus entered a scotoma. Three patients did not realize that the stimulus was missing in non-seeing areas during the first trial, but later found out that the stimulus 856
Fixation During Perimetry
b
U
30' circle
J-____c
malfixotion
C Fix. target
\-
30'
circle
J 4
Fix. target
\
I I
I
-z
-
Blind spot
H
1 sec Fig. 1.
Recordings of eye movements (upper tracings). Markings (lower tracings) a. Calibration. 20' horizontal eye movement. b. Stimulus moving through seeing areas. No eye movements occur. c. Stimulus moving through blind spot. Malfixation occurs when stimulus disappears into the blind spot area.
disappeared. When understanding this they showed proper fixation and signalled the disappearance of the stimulus. In four of the patients circumscript pathological scotomata were tested, in the other cases the blind spot was used. Malfixations of the same type were elicited in both cases.
B. Automatic Perimetry Materlal
Ten patients with verified or suspected glaucoma were tested. Their ages ranged from 43 to 79 years (mean age: 66 years). The same criteria for the selection of patients were used as described under "Kinetic perimetry". Seven of these patients were included in the material of "Kinetic perimetry" (see above). Left eyes were preferred because when using the non-randomized logic described below, the stimulus sequence is easier to predict in the blind spot area of the left than of the right eye (compare Fig. 2). Methods
A n automatic perimeter described by Heijl & Krakau (1975b) was used. It is a computer-controlled perimeter with 64 static stimuli, 56 of which are located at 5 O , 10' and 15' of eccentricity. The stimuli can be illuminated at 16 intensity levels, the ratio between two consecutive levels is 1:2. Two different perimetric test logics were used: 1. One test logic was identical to that used for glaucoma visual field screening (Heijl 1976), though the projected stimulus of the blind spot area was omitted.
857
A . Heiil and C. E . T . Krakau M 90'
M 0"
M 270'
Fig. 2. Test point pattern in the automatic perimetry. I.,ach test point is marked by a circle. In test logic 2 the stimuli are shown in ordered sequence beginning at point 1 and ending at point 56.
The threshold was first determined by a repeated up-and-down staircase method a t four points a t 10' eccentricity. T h e test point chosen among these four was decided by a random generator in the computer. Using these thresholds the computer calculated supraliminal stimulus intensities to be used for the testing a t the remaining 52 paracentral test points. The testing in these points was then carried out with supraliminal stimuli. During this testing the stimulus was exposed at random, i. e. the position of the stimulus to be exposed was chosen by a random generator in the computer among the test points not yet ready-tested. 2. The second test logic used the thresholds obtained in the four points at 10' eccentricity with the first logic. All 56 paracentral test points were then tested in sequence beginning at point 1 and ending at point 56 (Fig. 2 ) . The test points at 20' and 25' eccentricity were not used. The test procedure at
each point was identical to that in the first logic and the same intensity levels were used. Experiment
The patients were instructed to look steadily a t the red fixation light of the perimeter during the whole test and to press the button whenever a stimulus was visible. 858
Fixation During Perimetry
The test were carried out at a background luminance of 1.0 or 0.1 cd/m?. All subjects were corrected for ametropia and for near vision. The examination using the first test logic preceded the examination with the second test logic. Results
The measured threshold was zero (meaning no answer to the strongest intensity level of the perimeter) in a number of points in the field charts - denoting scotomatous areas. I n four patients there were very small differences between the results obtained by test logics 1 and 2. In the remaining six patients there was a difference of 1 to 6 in the number of points with threshold zero when the charts obtained with test logic 1 were compared to those of logic 2 and the field defects were smaller or even unrecognizable when the second (nonrandomized) logic was used (Compare Fig. 3). In no case was the opposite tendency found. The correspondence between the results in the seven patients, both in series A and B was satisfying. I n one of them the kinetic perimetry showed no ”malfixation” and there was no difference between automatic tests l and 2; in four there was “malfixation” in A and reduction of visual field defects in experiment B 2. In one case there was no “malfixation” in the recordings of eye movements
Fig. 3. Pathological visual field. a. Test logic 1 (randomized).
b. Test logic 2 (non-randomized). Visual defects appears smaller with logic 2 than with logic 1. The measured threshold is zero in eleven test points in Fig. a and in five test points in Fig. b. Intensity level numbers are marked along the 135’ meridian of Fig. a.
859
A . Heijl and C . E . T. Krakau
but the non-randomized logic showed reduction of scotomata compared with the randomized one; the opposite, “malfixation” but no reduction of scotoma size, was found in one case. Thus the frequencies of “malfixation” in A and of reduction of scotomata in B were similar.
Discussion The experiments with kinetic perimetry demonstrate that patients may exhibit a scotoma-induced “malfixation”, i. e. they may change their fixation as soon as the stimulus enters a non-seeing area, although they maintain reliable fixation as long as the stimulus is seen. This means that scotomata of blind spot size or even larger are easily missed a t kinetic perimetry. It might be pointed out that “malfixation” could also be elicited by switching off the stimulus, but then the patients indicated the disappearance of the stimulus. T h e situation with a test object moving at a constant speed in kinetic perimetry has no exact counterpart in static perimetry. The best approximation in our automatic perimeter may be to show the test points in ordered sequence along the three circles of test points (Fig. 2). The experiments with automatic perimetry indicate that visual field defects are more often missed when static stimuli are presented in sequence, i. e. at predictable places, than when they are exposed in random order. This is in agreement with earlier results obtained when comparing a randomized and a non-randomized test logic for automatic perimetry (Heijl & Krakau 1975a). The analogous results of the kinetic (A) and the automatic (B) test in the patients subjected to both support the assumption that the “scotoma-induced malfixation” comes to the fore in both situations. By randomization of the locations of consecutive stimuli such eye movements are largely prevented and this praxis permits us to detect visual field defects by automatic perimetry which have been overlooked in routine perimetry. I t also seems likely that the larger scotomata often recorded a t automatic perimetry are nearer to the true size than those obtained a t kinetic perimetry. As for manual perimetry, especially kinetic, the importance of extremely careful monitoring of the patient’s fixation during testing is evident.
References Armaly M. F. (1972) Selective perimetry for glaucomatous defects in ocular hypertension. Arch. Ophthal. (Chicago) 87, 518-524. Heijl A. & Krakau C. E. T. (1975a) An automatic static perimeter, design and pilot study. Acta ophthal. (Kbh.) 53, 293-310. 860
Fixation During Perirnelry Heijl A. & Krakau C. E. T. (1975b) An automatic perimeter for glaucoma visual field screening and control. Construction and clinical cases. Albrecht v. Graefes Arch. Ophthal. 197, 13-23. Heijl A. (1976) Automatic perimetry in glaucoma visual field screening. A clinical study. Albrecht v . Graefes Arch. Ophthal. 200, 21-37.
Author’s address: Anders Heijl, M. D., Department of Experimental Ophthalmology, University Eye Clinic. S-221 85 Lund, Sweden.
861
The Department of Experimental Ophthalmology (Head: C . E. T. Krakau), University Eye Clinic, Lund, Sweden
A NOTE ON FIXATION DURING PERIMETRY BY
A. HEIJL and C. E. T. KRAKAU
Small eye movements in the direction of the disappearing stimulus can be observed a t kinetic perimetry when the test object enters a scotomatous area. These may be responsible for the fact that several field defects, missed a t routine manual perimetry, are spotted a t automatic perimetry. In the latter case the patient cannot predict the position of next test light, since these are illuminated a t random, whereas at kinetic perimetry they are exposed in systematic order. By using two specially designed logics in automatic perimetry, it was shown that there may be a reduction of the scotoma size if the lights are exposed in a n ordered sequence. This effect is avoided by using a logic with randomly exposed stimuli. It is most likely that the difference can be attributed to ‘malfixation”.
K e y words: perimetry - kinetic perimetry
-
automatic perimetry - fixation.
In a recent study on glaucoma screening (Heijl 1976) it was shown that many visual field defects, previously missed at routine perimetry, kinetic or the Armaly technique (Armaly 1972) were detected by automatic perimetry. In a previous paper on a n automatic perimeter (Heijl & Krakau 197513) a few glaucomatous field defects were used to illustrate the performance of this instrument. I t was then observed that the defects sometimes seemed larger and/or deeper when plotted by the automatic device than with kinetic perimetry. Received March 29, 1977.
854
Fixation During Perimetry
A difference between the automatic and the manual test, which perhaps, at least partly, might explain these discrepancies, is the fact that the object appears at random locations during automatic perimetry but in a more or less ordered sequence in manual perimetry. Thus the patient is most often able to predict the position of the test object before it is seen and it might be tempting for him to change his fixation slightly when not observing the stimulus. In fact, a close look at the patients' fixation during manual perimetry reveals that fairly often, when a test object passes the border of a scotoma from a seeing to a non-seeing area, fast, small eye movements in the direction of the disappearing stimulus can be seen in the "telescope" of the Goldmann perimeter. As far as we know this phenomenon has not been explicitly described or analysed in the literature. The aim of the present note is to demonstrate the existence of this phenomenon by the recording of eye movements during kinetic perimetry and also to show how the phenomenon is provoked or avoided in automatic perimetry.
A. Kinetic Perimetry Material
Eleven patients with verified or suspected glaucoma were tested. Their ages ranged from 43 to 79 years (mean age: 68 years). Patients who were very trained perimetric subjects were avoided as were patients known to cooperate badly in perimetry. Nine pathological and two normal visual fields were examined. Some patients were chosen because their field defects had been missed by manual perimetry but detected by automatic perimetry. Only one eye was examined in each patient. Methods
Recording of eye movements. Two disposable AgAgCl electrodes (Medicotest A-15 M) were used. They were placed lateral to the outer canthus of each eye. A neutral electrode was placed above the left wrist. The temple electrodes were connected to a small preamplifier, the output of which was fed to an ink recorder (Mingograf 24B). The amplification of the recorder was so adjusted as to give a deviation of 10-20 mm in the ink-written curve for a horizontal eye movement of 30'. A small marker switch was connected to the second channel of the recorder. When pressed it produced markings on the ink-written line of this channel. Control of perimeter stimulus. The pantograph lever of the Goldmann perimeter was connected to a string driven by a n electrical motor controlled from a drive 855 55*
A . Heijl and C. E. T . Krakau
unit. Through this arrangement the pantograph lever could be made to move along different straight lines on the visual field chart at even speed. Experiment
The patient was placed in the Goldmann perimeter and provided with correction for ametropia and near vision. The electrodes were connected. A filled-in chart of the examined eye was put into the chart holder of the perimeter. The stimulus was positioned at an eccentricity of about 35”. The stimulus intensity was so adjusted as to be slightly supraliminal in this position. The patient was instructed to keep his gaze steady at the fixation target of the perimeter and to signal if the stimulus disappeared. The ink recorder and the movement of the stimulus were then started simultaneously and the stimulus travelled centripetally towards the fixation target of the perimeter a t constant speed (usually 3’/sec). A n assistant pressed the switch of the marker unit when the stimulus crossed the 30O-circle and the fixation target. A t least two meridians were tested - each five times - in every subject. One meridian crossed the blind spot or a known glaucomatous scotoma, the other passed only through seeing areas. The assistant was further instructed to press the button of the marker unit when any scotoma border was crossed. The patient’s fixation was constantly supervised in the “telescope” of the Goldmann perimeter. Results
Three patients demonstrated steady fixation in all tracings and one patient showed poor fixation in most tracings. In the remaining seven patients there was a distinct difference in fixation when the stimulus passed through seeing and non-seeing areas. These patients usually maintained good fixation when the stimulus moved in seeing areas, but when the stimulus passed into a scotoma they often, obviously unconsciously, changed their fixation in the direction of the stimulus, recognized the stimulus and quickly resumed fixation again (Fig. 1). Thus they never “lost” the stimulus, and did not signal its disappearance. These “scotoma-induced malfixations” were not only recorded by the ink recorder, but were also readily seen in the fixation-monitoring “telescope” of the perimeter. They were 7”-15” when the stimulus passed into the blind spot. Thus the patients sometimes changed their fixation all the way to the test object, but in some cases no more than to let the stimulus reach a seeing part of the visual field. Four out of these seven patients never recognized the stimulus as missing and mostly demonstrated malfixation each time the stimulus entered a scotoma. Three patients did not realize that the stimulus was missing in non-seeing areas during the first trial, but later found out that the stimulus 856
Fixation During Perimetry
b
U
30' circle
J-____c
malfixotion
C Fix. target
\-
30'
circle
J 4
Fix. target
\
I I
I
-z
-
Blind spot
H
1 sec Fig. 1.
Recordings of eye movements (upper tracings). Markings (lower tracings) a. Calibration. 20' horizontal eye movement. b. Stimulus moving through seeing areas. No eye movements occur. c. Stimulus moving through blind spot. Malfixation occurs when stimulus disappears into the blind spot area.
disappeared. When understanding this they showed proper fixation and signalled the disappearance of the stimulus. In four of the patients circumscript pathological scotomata were tested, in the other cases the blind spot was used. Malfixations of the same type were elicited in both cases.
B. Automatic Perimetry Materlal
Ten patients with verified or suspected glaucoma were tested. Their ages ranged from 43 to 79 years (mean age: 66 years). The same criteria for the selection of patients were used as described under "Kinetic perimetry". Seven of these patients were included in the material of "Kinetic perimetry" (see above). Left eyes were preferred because when using the non-randomized logic described below, the stimulus sequence is easier to predict in the blind spot area of the left than of the right eye (compare Fig. 2). Methods
A n automatic perimeter described by Heijl & Krakau (1975b) was used. It is a computer-controlled perimeter with 64 static stimuli, 56 of which are located at 5 O , 10' and 15' of eccentricity. The stimuli can be illuminated at 16 intensity levels, the ratio between two consecutive levels is 1:2. Two different perimetric test logics were used: 1. One test logic was identical to that used for glaucoma visual field screening (Heijl 1976), though the projected stimulus of the blind spot area was omitted.
857
A . Heiil and C. E . T . Krakau M 90'
M 0"
M 270'
Fig. 2. Test point pattern in the automatic perimetry. I.,ach test point is marked by a circle. In test logic 2 the stimuli are shown in ordered sequence beginning at point 1 and ending at point 56.
The threshold was first determined by a repeated up-and-down staircase method a t four points a t 10' eccentricity. T h e test point chosen among these four was decided by a random generator in the computer. Using these thresholds the computer calculated supraliminal stimulus intensities to be used for the testing a t the remaining 52 paracentral test points. The testing in these points was then carried out with supraliminal stimuli. During this testing the stimulus was exposed at random, i. e. the position of the stimulus to be exposed was chosen by a random generator in the computer among the test points not yet ready-tested. 2. The second test logic used the thresholds obtained in the four points at 10' eccentricity with the first logic. All 56 paracentral test points were then tested in sequence beginning at point 1 and ending at point 56 (Fig. 2 ) . The test points at 20' and 25' eccentricity were not used. The test procedure at
each point was identical to that in the first logic and the same intensity levels were used. Experiment
The patients were instructed to look steadily a t the red fixation light of the perimeter during the whole test and to press the button whenever a stimulus was visible. 858
Fixation During Perimetry
The test were carried out at a background luminance of 1.0 or 0.1 cd/m?. All subjects were corrected for ametropia and for near vision. The examination using the first test logic preceded the examination with the second test logic. Results
The measured threshold was zero (meaning no answer to the strongest intensity level of the perimeter) in a number of points in the field charts - denoting scotomatous areas. I n four patients there were very small differences between the results obtained by test logics 1 and 2. In the remaining six patients there was a difference of 1 to 6 in the number of points with threshold zero when the charts obtained with test logic 1 were compared to those of logic 2 and the field defects were smaller or even unrecognizable when the second (nonrandomized) logic was used (Compare Fig. 3). In no case was the opposite tendency found. The correspondence between the results in the seven patients, both in series A and B was satisfying. I n one of them the kinetic perimetry showed no ”malfixation” and there was no difference between automatic tests l and 2; in four there was “malfixation” in A and reduction of visual field defects in experiment B 2. In one case there was no “malfixation” in the recordings of eye movements
Fig. 3. Pathological visual field. a. Test logic 1 (randomized).
b. Test logic 2 (non-randomized). Visual defects appears smaller with logic 2 than with logic 1. The measured threshold is zero in eleven test points in Fig. a and in five test points in Fig. b. Intensity level numbers are marked along the 135’ meridian of Fig. a.
859
A . Heijl and C . E . T. Krakau
but the non-randomized logic showed reduction of scotomata compared with the randomized one; the opposite, “malfixation” but no reduction of scotoma size, was found in one case. Thus the frequencies of “malfixation” in A and of reduction of scotomata in B were similar.
Discussion The experiments with kinetic perimetry demonstrate that patients may exhibit a scotoma-induced “malfixation”, i. e. they may change their fixation as soon as the stimulus enters a non-seeing area, although they maintain reliable fixation as long as the stimulus is seen. This means that scotomata of blind spot size or even larger are easily missed a t kinetic perimetry. It might be pointed out that “malfixation” could also be elicited by switching off the stimulus, but then the patients indicated the disappearance of the stimulus. T h e situation with a test object moving at a constant speed in kinetic perimetry has no exact counterpart in static perimetry. The best approximation in our automatic perimeter may be to show the test points in ordered sequence along the three circles of test points (Fig. 2). The experiments with automatic perimetry indicate that visual field defects are more often missed when static stimuli are presented in sequence, i. e. at predictable places, than when they are exposed in random order. This is in agreement with earlier results obtained when comparing a randomized and a non-randomized test logic for automatic perimetry (Heijl & Krakau 1975a). The analogous results of the kinetic (A) and the automatic (B) test in the patients subjected to both support the assumption that the “scotoma-induced malfixation” comes to the fore in both situations. By randomization of the locations of consecutive stimuli such eye movements are largely prevented and this praxis permits us to detect visual field defects by automatic perimetry which have been overlooked in routine perimetry. I t also seems likely that the larger scotomata often recorded a t automatic perimetry are nearer to the true size than those obtained a t kinetic perimetry. As for manual perimetry, especially kinetic, the importance of extremely careful monitoring of the patient’s fixation during testing is evident.
References Armaly M. F. (1972) Selective perimetry for glaucomatous defects in ocular hypertension. Arch. Ophthal. (Chicago) 87, 518-524. Heijl A. & Krakau C. E. T. (1975a) An automatic static perimeter, design and pilot study. Acta ophthal. (Kbh.) 53, 293-310. 860
Fixation During Perirnelry Heijl A. & Krakau C. E. T. (1975b) An automatic perimeter for glaucoma visual field screening and control. Construction and clinical cases. Albrecht v. Graefes Arch. Ophthal. 197, 13-23. Heijl A. (1976) Automatic perimetry in glaucoma visual field screening. A clinical study. Albrecht v . Graefes Arch. Ophthal. 200, 21-37.
Author’s address: Anders Heijl, M. D., Department of Experimental Ophthalmology, University Eye Clinic. S-221 85 Lund, Sweden.
861
