Dynamic infrared pupillometry was undertaken on 19 newborn infants while they fixated a uniform background upon which a 0.1 c/deg sine wave grating was ...
Visual Acuity and Pupillary Responses to Spatial Structure in Infants Kenneth D. Cocker,* MerrickJ. Moseley,* Jeffrey G. Bissenden,f and Alistair R. Fielder*
Purpose. To determine the age of onset of the pupil grating response (PGR). To compare estimates of resolution acuity obtained by pupillometric and behavioral methods in early infancy. Methods. Dynamic infrared pupillometry was undertaken on 19 newborn infants while they fixated a uniform background upon which a 0.1 c/deg sine wave grating was briefly presented. Pupillary responses were also recorded to an increment in luminance of a spatially homogeneous target. Longitudinal measurements of PGRs were obtained from a subset of eight infants between 3.5 and 38 weeks of age. In this group, behavioral estimates of visual resolution obtained using the acuity card procedure were compared with the highest spatial frequency grating to elicit a PGR. Results. When presented with the pattern stimulus, newborn infants did not show any pupil reaction indicative of a PGR. This finding could not be attributed to immaturity of pupillomotor function: All infants showed marked pupillary constriction to diffuse light stimulation. By 1 month of age, pupillary responses to pattern stimuli were reliably present. For these and older infants, the spatial frequency of the finest grating to elicit a PGR was comparable to the behaviorally determined resolution threshold: mean difference (±95% confidence interval) = 0.28 ± 0.23 octaves. Conclusions. A PGR could not be detected in newborn infants. From 1 month of age, responses to spatial structure can provide objective estimates of visual acuity comparable to those determined by established methods. Invest Ophthalmol Vis Sci. 1994;35:2620-2625.
A he most familiar and widely studied pupil responses are those contingent upon changes in the retinal light flux. However, it is now evident that the pupil reacts to many attributes of a visual stimulus, including its spatial structure, chromaticity, and motion.12 The pupillary response to the appearance of a grating stimulus against an equiluminant background, the so-called pupil grating response (PGR), exhibits a spatial frequency dependence grossly similar to the psychophysical contrast sensitivity function.1'3 This has led some workers to suggest that the neural mechanisms underlying the contrast sensitivity function are also involved in generating the PGR.34 Although the
From the *Department of Ophthalmology, University of Birmingham, and the •\Neonatal Unit, Dudley Road Hospital, Birmingham, United Kingdom. Supported by The Wellcome Trust. Submitted for publication September 10, 1993; revised November 22, 1993; accepted November 23, 1993. Proprietary interest category: N. Reprint requests: Dr. Merrick Moseley, Academic Unit, The Birmingham and Midland Eye Hospital, Church Street, Birmingham, B3 2NS, United Kingdom.
2620
neural pathways mediating this response remain obscure, these small magnitude constrictions (typically ~ 5 % change in area) have been resolved into transient and sustained components that may reflect the activity of neurons functionally similar to those found in the magno- and parvo-cellular layers of the LGN.5 A PGR cannot be recorded for stimuli presented within the blind hemifield of patients with homonymous hemianopias, who otherwise demonstrate near normal pupillary reaction to light.1'6 This dissociation of light and pattern responses suggests that the latter reflect predominantly the activity of cortical visual mechanisms. Several investigators, noting the spatial frequency dependence exhibited by the PGR and other pupil responses to spatial structure, have compared estimates of visual acuity extracted from pupillograms with psychophysically determined resolution and recognition acuity.3'4'7'8 These studies have yielded very high levels of correlation in adult subjects (Table 1). Although it is intriguing that the pupillomotor system and its associated neural pathways can provide such a precise
Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5 Copyright © Association for Research in Vision and Ophthalmology
Pupil and Behavioral Acuity in Infants
TABLE l.
2621
Comparisons of Pupillometric and Conventional Psychophysical Acuities Diagnostic Category
Reference
Pupillometry Stimulus
Psychophysical Test
Slooter and van Norren7 (1980) Slooter8 (1981)
Checker board reversal
Resolution (checkerboard)
n
r = 0.96
Checker board reversal
Resolution (checkerboard)
a P a, p, o a, p n
r r r r r
n, c, a
ICC = 0.98
n, c, a
ICC = 0.98
Barbur and Thompson3 (1987) Cocker and Moseley4 (1992)
Sine wave gratings Sine wave gratings
Recognition (letter) Recognition (letter) Extrapolation from CSF (sine wave gratings) Resolution (sine wave gratings) Recognition (letter)
Correlation
= 0.91 = 0.97 = 0.81 = 0.91 = 0.93
n = normals; a = amblyopes; c = cataracts; o = optic atrophy/metamorphosia; p = other pathologies; r = product moment correlation coefficient; ICC = intraclass correlation coefficient.
gauge of visual perception, such findings, at least for adults, remain of limited application. Pupillometry with yoked stimulus presentation is a time-consuming procedure that requires expensive instrumentation, whereas visual acuity can, under most circumstances, be accurately determined with minimal equipment and expertise. However, in infants, who cannot provide appropriate verbal responses, measurement of acuity by pupillometric means would seem apposite. Here it would combine the advantages of established, but mostly subjective, behavioral techniques, which are noninvasive, with those of objective electrophysiological methods, which require electrode placement. Against this, pupillometry in infants presents considerable potential difficulties insofar as these subjects cannot be instructed to maintain the steady head and eye fixation generally necessary for this procedure. One further reservation stems from reports that certain components of visual function in early infancy are subserved by subcortical mechanisms.9"11 If this is so, given that the pupil grating response appears to reflect the activity of central visual processes, might it be attenuated or even absent in the newborn? In this study, our first aim was to determine the age of onset of the PGR. Then, using a previously described method4 requiring a minimal number of individual pupillograms, our second aim was to obtain objective measurements of visual acuity in a group of healthy infants and compare the acuities so obtained with those from an established behavioral method. METHODS Nineteen healthy, full-term neonates (gestational age 40 weeks ±16 days) were tested within 3 days of birth. Our study followed the tenets of the Declaration of Helsinki, with local ethical committee approval and the full, informed consent of parents.
Binocular visual acuity was determined behaviorally using the acuity card procedure,12 a rapid, preferential looking-based technique now finding widespread use in clinical practice.13 In this test, cardmounted, high-contrast, square wave gratings are presented at a distance of 50 cm. Stimuli range from 0.1 to 38 c/deg spaced at 0.5 octave intervals. The tester observes the infant's looking behavior through a peephole in the center of the card and judges whether the infant shows a preference (i.e., appropriate eye and/or head movements) for the grating, which is displaced to the right or left of the card. The tester records the resolution acuity as the highest spatial frequency to which the infant reliably responds. Pupillometry was undertaken using a P_Scan 100 infrared (IR) pupillometer (Circuit Solve, UK Ltd.), details of which have been published elsewhere.14 This instrument extracts pupil size using a circle-fitting algorithm with spatial and temporal resolutions of 0.01 mm and 20 ms, respectively. In common with most instruments of this kind, head restraint and proximally located cameras and illumination sources are normally required. We adapted the pupillometer for use with infants by adjustment to the imaging systems and the use of a remote IR source such that it was possible to undertake pupillometry without rigid head restraint (see Fig. 1). The immediate test environment of the modified pupillometer is similar to that required for the acuity card procedure: A grey curtain surrounds the stimulus screen, below which an IR light source and IR sensitive camera are positioned. With visual feedback from a videomonitor, the experimenter locates the infant at the position where the focal plane of the camera coincides with the IR light beam in front of the stimulus screen. This distance (d) can be adjusted depending on the required subtense and spatial frequency of the test targets to ensure that pupillometry is only undertaken when the infant is correctly posi-
Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5
2622
STIMULUS DISPLAY COMPUTER
IR SOURCE
FIGURE 1. Experimental configuration for infant pupillometry. The infant is held in the arms of the operator, who locates the infant's head within the field of view of the camera. Feedback is provided by the monitor shown on the right. tioned and fixating the display. The IR source remained at a constant distance (30 cm) independent of the actual viewing distance. Camera field of view subtended 2.5 cm horizontally by 1.7 cm vertically. By continually monitoring the pupil image and maintaining image focus, it was possible to hold the infant to within ±1 cm of the appropriate viewing distance. The maximum error in recorded pupil diameter within this tolerance bandwidth was ±0.5%. Infants could be held in a variety of holds, dependent upon their age and individual preference. At the start of the experimental session, infants were screened to confirm the presence of qualitatively normal pupillary responses to flash-light illumination. After acuity card assessment, recordings were undertaken during which a homogenous gray background (1.3 log cd.m"2) was exchanged for 500 ms with an achromatic 0.1 c/deg sine wave grating of identical space average luminance. The grating subtended 22° with 80% Michelson contrast. Recording commenced 500 ms before stimulus presentation and continued for 2 seconds after stimulus offset. Stimuli were viewed binocularly on a 19-inch Aydin high-resolution monitor at a distance of 50 cm. Twelve pupillograms were recorded. These were digitally filtered to eliminate instrumental noise and artifacts caused by subject movement and blinking and were incorporated into an average trace. The trace was analyzed4 to determine objectively if any activity indicative of a PGR was present, i.e., if any post-stimulus constriction was significantly greater than the background pupillomotor activity. At the end of the session, which lasted approximately 1 hour, infants who remained in an appropriate state of attentiveness underwent quantitative measurement of their pupillary light reflex. These responses
were assessed by a half-second, 0.9 log unit increment in the luminance of a 1.3 log cd.m"2 homogeneous grey square subtending 22°. Of the original 19 infants, we repeated testing on eight infants at intervals of approximately 1, 2, 3, and 7 to 9 months of age, though none achieved 100% attendance (Table 2). At each of the follow-up sessions, acuity was again determined with acuity cards. Twelve pupillographic records were taken of responses to sine wave gratings of a spatial frequency equal to the behaviorally determined acuity and again were analyzed for the presence of a PGR. Further pupillograms were recorded in response to a doubling (if PGR present) or halving (if PGR absent) of the spatial frequency of the test stimulus. This was repeated until a reversal occurred, then a final stimulus was presented, one-half octave less than or greater than that of the reversal frequency. The pupil acuity was recorded as the highest spatial frequency grating from which an identifiable PGR was elicited. For test spatial frequencies greater than 3.2 c/deg and up to 6.5 cy/ deg, viewing distance was increased to 1 meter (11° field); for spatial frequencies greater than 6.5 c/deg, viewing distance was 2 meters (5.5° field).
RESULTS A comparison of pupillograms elicited by the luminance and grating stimuli in the neonatal period are shown for a single infant in Figure 2. No pupillograms recorded from any neonate showed evidence of a pupil grating response to the 0.1 c/deg grating stimulus. Quantitative measurements of light responses were obtained from nine of the newborn infants. Mean (± SE) percentage change in area and latencies were 24.6% ± 2.6% and 349 ± 53 ms, respectively. All infants had a measurable behavioral resolution acuity: mean (± SE) = 0.471 ± 0.025 c/deg. Beyond the neonatal period, it proved possible to identify the presence of a PGR (Fig. 3), and estimates of pupil acuities were compared with those obtained behaviorally (Table 3). The level of agreement between TABLE
2. Infant ages at test
Patient no.
/ mo
2 mo
1 2 3 4 5 6 7 8
3.5
7.7 9.7 7.7 9.1 10.1
3.7 6.0 5.3 4.5 5.1 5.7
Values are age (weeks) at test.
3 mo
7-9 mo
13.7 38.0 15.4 14.0 19.8
28.4 31.0
Pupil and Behavioral Acuity in Infants
2623
3. Pupillometric Versus Behavioral Resolution Acuities
TABLE 5.0-
E, 0)
4.8"
\
c5 E n
5 '5. 0.
Patient
Age
No.
(wk)
Pupil Acuity (cy/deg)
Behavioral Acuity (cy/deg)
i
1
3.4
2
7.7 13.7
%
4.6-
\
9.7
\
3
4.4•..»'
38.0 7.7 3.7 9.1 6.0 5.3
4
4.2" 1000
2000
3000
5-
Time (ms)
FIGURE 2. Representative pupillograms recorded from a newborn infant to the appearance of a 0.1 c/deg sine wave grating (solid line) to a 0.83 log unit increment in the luminance of a homogenous grey square (broken line). Note: The 3-second recording period is not sufficient to show the recovery of the pupil to its pre-stimulus diameter. Hatched bar indicates stimulus onset and offset.
1.8 4.8 1.8 4.8 3.6 0.9 2.4
0.86 0.86
0.97 0.97
10.1
1.3
1.8
4.6
0.86
0.96
7
15.4 28.4 31.0
3.2 9.8 9.8
4.8 9.8 9.8
5.1
0.32
0.65
14.0
1.6
2.4
5.7
0.64 0.86
0.64
19.8
(octaves*)
0.97
1.3 1.6 3.2 2.4 3.2 1.6 1.3 2.4
6
8
Difference
0.422 -0.170 -0.585 0.415 -0.585 -1.170 0.422 0.000 -0.174 -0.174 -0.469 -0.159 -0.585 0.000 0.000 -1.022 -0.585 0.000 -0.896
1.6
Log;, (pupil acuity) — log., (behavioral acuity).
3.2 cy/deg
1000
each individual pair of measurements is given in the final column as the octave (log2) difference between each individual's scores. A scatterplot of card and pupil acuities is given in Figure 4. The mean octave difference (log2 pupil acuity — log2 card acuity: ±95% confidence interval) between the two methods was -0.28 ± 0.23 octaves. The most appropriate measure of correlation for method comparison studies is the Intraclass Correlation Coefficient.15 For these data, ICC = 0.909 with a 95% confidence interval of 0.297 to 0.973. Card and pupil acuities are plotted as a function of age in Figure 5.
2000
Time (ms) Q)
FIGURE 3. Pupillograms of responses to grating stimuli in a single infant. In the analysis method,4 the largest post-stimulus constriction (upward arrow) and the preceding maxima (downward arrow) are identified. The latter is denned as the onset of the putative PGR. A least-squares regression line is fitted to the data from the onset of recording until this point and extrapolated forward to a point coincidental in time with the occurrence of the maximum post-stimulus constriction. A 95% prediction interval is calculated about this point, as indicated by the vertical bars. If the response lies outside (below) this interval, it is significantly greater than the random fluctuations that preceded it, and its identity as a PGR is confirmed. In the pupillograms shown, PGRs are present in response to gratings of 2.3 and 3.2 c/deg, but not of 4.2 c/deg. Hatched bar indicates stimulus onset and offset.
•o o
I O GL
Q.
0.25 0.25
0.5
1.0
2.0
4.0
8.0
16.0
Behavioral Grating Acuity (cy/deg) FIGURE 4. Pupil grating versus behavioral grating acuity. The equation for the regression line is: Pupil acuity = 0.94 X behavioral acuity — 0.22.
2624
Investigative Ophthalmology & Visual Science, April 1994, Vol. 35, No. 5
mation of 0.08 octaves found in a previous study4 on adults using similar procedures. In estimating acuity from pupillometric data, one has to deal with residual noise present in the pupillary system originating from nonvisual, psychologic, and somatic sources. This residual noise is attenuated by averaging many recordings to a given stimulus; however, the limited number of averages available from any individual infant may elevate the threshold for detecting the response. Despite the excellent levels of agreement found between pupil acuities and those measured behav0.25 Neo nates 10 20 iorally, it is judicious to assess the practical conseAge (weeks) quences of this finding. At present, pupil acuity measurement is a laborious procedure requiring expensive FIGURE 5. Visual acuities by pupillometric and behavioral and specialized instrumentation, which clearly limits methods. Binocular acuity card age norms are indicated by solid lines. (From Teller Acuity Cards Handbook, Vistech Con- its application. However, a parallel may be drawn with the development of preferential looking-based techsultants Inc.) Open squares: behavioral acuities. Open cirniques. In their earliest incarnations, these were of cles: pupil acuities. Note: No pupil acuities were determined during the neonatal period (see text). similar complexity to the pupillometry system described here requiring computer-controlled display monitors and the necessity of many trials to meet the DISCUSSION requirements of the psychophysical method. Gradually, it became apparent that less sophisticated techAlthough neonates showed a clear behavioral preferniques could provide comparable accuracy, thus preence for low-frequency gratings (acuity cards), we cipitating their routine clinical application. Although were unable to detect pupil responses to stimuli of dynamic pupillometry yoked to stimulus display is equivalent or lower spatial frequency. This deficit canlikely to remain technically sophisticated, the developnot be attributed to immaturity of pupillomotor funcment of handheld instrumentation and real-time analytion: All subjects showed qualitatively normal light resis is undoubtedly feasible. We do not argue that pusponses that were confirmed quantitatively in nine inpillometry is superior to current methods of measurfants. Possibly, cortical visual mechanisms that ing infant acuity, rather, that it is complementary, mediate pupil responses to spatial structure are, like combining the objectivity of electrophysiological techother components of early visual function,9"11 undeveloped at birth. We cannot rule out, however, that patniques, such as the sweep VEP, 16 with the noninvasivetern responses are in fact present in neonates but are ness of most behavioral tests. so grossly attenuated as to be undetectable by our current methods. Key Words For infants tested at 1 month of age and older, a pupil, acuity, infant, visual development PGR was consistently present and a high level of correlation was found between estimates of grating References acuity measured behaviorally with those extracted from the PGRs. In particular, the results show a 95% 1. Barbur JL, Keenleyside MS, Thomson WD. Investigaconfidence interval for between-test agreement of aption of central visual processing by means of pupillomproximately one-quarter of an octave. The coarse resoetry. In: Kulikowski KK, Dickinson CM, Murray IJ, eds. Seeing Contour and Colour. Oxford: Pergalution (0.5 octaves) of the two methods will tend to mon;1989:431-451. limit the spread of marginal outliers and thus will im2. Barbur JL, Harlow AJ, Sahraie A. Pupillary responses prove agreement. Intraclass correlation, however, reto stimulus structure, colour and movement. Ophthalmains high. It is clear that there is comparable developmol Physiol Opt. 1992;12:137-141. ment of pupil acuity and behavioral resolution acuity 3. Barbur JL, Thomson WD. Pupil response as an objecbetween 1 and 9 months of age. Though the pupil tive measure of visual acuity. Ophthalmol Physiol Opt. grating stimuli were viewed at three different distances 1987;7:425-429. as a function of spatial frequency (see Methods), there 4. Cocker KD, Moseley MJ. Visual acuity and the pupil is no suggestion that this had any effect upon the level grating response. Clin Vision Sci. 1992;7:143-146. of agreement. 5. Young RSL, Kennish J. Transient and sustained comPupil acuity underestimates behavioral acuity by ponents of the pupil response evoked by spatial patterns. Vision Res. 1993;33:2239-2252. 0.28 octaves or, on average, just more than one-half 6. Barbur JL, Forsyth PM. Can the pupil response be the test's resolution. This is higher than an underesti-
Pupil and Behavioral Acuity in Infants
7. 8. 9. 10. 11. 12.
used as a measure of visual input associated with the geniculo-striate pathway? Clin Vision Sci. 1986;l:107111. Slooter J, van Norren D. Visual acuity measured with pupil responses to checkerboard stimuli. Invest Ophthalmol Vis Sci. 1980;19:105-108. Slooter J. Clinical use of visual acuity measured with pupil responses. Doc Ophthalmol 1981;50:389-399. Bronson G. The postnatal growth of visual capacity. Child Dev. 1974;45:873-890. Atkinson J. Early visual development: Differential functioning of parvocellular and magnocellular pathways. Eye. 1992;6:129-135. Fielder AR, Evans NM. Is the geniculostriate system a prerequisite for nystagmus? Eye. 1988;2:628-635. McDonald MA, Dobson V, Sebris SL, Baitch L,
2625
13. 14.
15.
16.
Varner D, Teller DY. The acuity card procedure: A rapid test of infant acuity. Invest Ophthalmol Vis Sci. 1985;26:1158-1162. Fielder AR, Dobson V, Moseley MJ, Mayer DL. Preferential looking: Clinical lessons. Ophthalmol Paed Genet. 1992;13:101-110. Barbur JL, Thomson WD, Forsyth PM. A new system for the simultaneous measurement of pupil size and two-dimensional eye movements. Clin Vision Sci. 1987;2:131-142. Lee J, Koh D, Ong CN. Statistical evaluation of agreement between two methods for measuring a quantitative variable. Comput Biol Med. 1989;19:71-70. Norcia AM, Tyler CW. Spatial frequency sweep VEP: Visual acuity during the first year of life. Vision Res. 1985;25:1399-1408.
