Chapter 4
Biometry of the Eye in Infancy and Childhood
Donald O. Mutti and Karla Zadnik
Refractive error is actually an unusual finding in a
general population of young children, although it
may seem to be a common occurrence in clinical
practice. The prevalence of refractive error at the
age of 6 is approximately 8%, with 2% of children
being myopic (£-0.50 D) and 6% hyperopic
(>+1.50 D) [1]. Yet, by the age of 15 years, nearly
15% of children are myopic, whereas the preva-
lence of hyperopia remains at about 6% [1]. Nearly
75% of the adult population of the United States is
myopic [2].
If most myopia develops in childhood, why
study ocular growth in infancy? Although juvenile
myopia is classically associated with axial elonga-
tion, most eye growth and change in the power of
the eye’s refractive surfaces take place in infancy.
The fact that most of this eyc growth occurs without
creating refractive error, in fact more often reduc-
ing neonatal refractive error (emmetropization),
makes infant eye growth an important model for ap-
propriate and normal ocular development.
Refractive error and ocular component develop-
ent have fascinated clinicians and vision scientists
+1ik= for hundreds of years. Tools such as retin-
cvcloplegia, keratometry, and phakometry
- available since the last century for the
Jf refractive error and the ocular compo-
iribute to it. Recent technologic de-
i trends in the delivery of eye care
cial opportunities for assessing oc-
For example. the advent of refrac-
« created a renaissance of interest in
corneal topography. The availability of charge-
coupled device (CCD), or “chip” cameras, inter-
faced with personal computers has allowed for the
rapid capture and digitization of video-based infor-
mation generated by these topographers. Increases
in computer power and speed in the last decade
have allowed for processing and storage of these
large video image files. The increased demand for
cataract surgery from our active aging population
has stimulated continual improvement of modern
ultrasonography units, resulting in the availability
of more efficient, compact, and powerful machines.
BIOMETRIC TECHNIQUES
Keratometry and Corneal Topography
Obviously, the chief obstacle in obtaining biomet-
ric data in infants is their lack of cooperation. Two
Laws of Infant Examination, attributed to Howard
Howland of Cornell University, govern infant ex-
amination: (1) infants wiggle; and (2) a lot. Con-
ventional keratometry is therefore difficult but has
been successfully performed in newborns with the
help of an eyelid speculum [3, 4]. Several photo-
graphic methods have been used successfully. Man-
dell [5] has used a 35-mm camera equipped with a
tube extending from a lens that has clear rings at set
intervals. When the tube is illuminated by the flash
ring around it, these circles form mires that make
concentric reflections on the cornea, serving as a
small and portable corneoscope. Howland and
Sayles [6] have used a 35-mm camera with a lens
surrounded by eight optical fibers connected to the
flash. This serves as a keratometer that works si-
multaneously in four meridians. The examiner can
have the infant sit on the parents lap and attract the
baby’s attention in the direction of the camera.
Ultrasonography
Measurement of the eye’s axial dimensions presents
special challenges in infants. Since the ultrasound
probe must come into contact with the cornea or
eyelid, infants tend to object more strongly to it
than to other, less invasive procedures. The dimen-
sions obtained from ultrasonography that are im-
portant in refractive development arc anterior
chamber depth (ACD), lens thickness, and vitreous
chamber depth (VCD). The sum of all three is the
axial length of the eye. Because of the difficulty in
examining awake infants, much ultrasonography
data have been obtained on sedated or anesthetized
infants [7, 8], although cooperative infants and new-
borns may tolerate the procedure while awake [8,
9]. Recently, it has been shown that performing A-
scan ultrasonography with the probe over the closed
eyelid, similar to the procedure used for B-scan, is a
feasible method for infant ultrasonography [10].
Measures on sleeping infants or cooperative tod-
dlers may thus be accomplished without sedation or
use of a lid speculum.
Phakometry
One of the least studied ocular components con-
{ributing to refraction in the infant eye is the crys-
talline lens. This is unfortunate, as the lens is
critical to the refractive development of the eye,
second only to axial length in importance. The lens
. undergoes a greater dioptric change (20 D), on av-
erage, than the cornea (5 D), and it continues to de-
velop with respect to power and curvature from
infancy through childhood. On the other hand, the
cornea is nearly fully developed by the age of 2
years. Therefore, compensation for axial growth to
maintain emmetropia is more likely to come from
the crystalline lens. Technologic limitations have
hampered previous investigators, limiting our cur-
rent knowledge. The first descriptions of crystalline
lens radii in the newborn came in 1909, in which
the estimates were made from frozen sections of
seven cadaver eyes [11]. Up until the present, these
were the only available data for use in creating
schematic eyes and modeling of infant dioptrics,
Phakometric techniques have been known, however,
since the time of Tscherning [12]. He described
how viewing a pair of Purkinje images could yield
the equivalent radius of curvature of a lens surface,
since the separation of the two images was propor-
tional to the radius of curvature. Figure 4.1 depicts
how incoming, parallel light forms images after re-
flection from the cornea and the anterior surface of
the crystalline lens. These images form at the focal
points of the respective mirror surfaces, or 1/2. By
similar triangles, the heights of the images will be
proportional to the radius of curvature of the mir-
rors as viewed in air. By knowing the position of the
mirror surfaces from ultrasound data, ray tracing
equations can be applied to find the true radius for
each lens surface within the eye.
Phakometry has been performed in several stud-
ies of children in conjunction with 35-mm photog-
raphy. This format has several disadvantages,
however. The number of images collected is small
because of the time needed to take a picture, the
camera flash may disturb or surprise photophobic
children, and a child’s motion during measurement
may result in out-of-focus or poorly aligned images.
Given that film development takes time, an exam-
iner may not be aware of the poor data quality until
long after the measurement session. Since phakom-
etry generally requires pupillary dilation, repeating
the measurement to obtain complete data becomes
an added burden to the child. Advances in video and
computer technology address many of these issues.
Using a video format means that no flash is required
and the camera may change focus to follow the pa-
tient’s movement more easily. More important, the
clinician can know that acceptable images are being
recorded at the time of data collection without de-
lays for processing. Computer hardware can grab
multiple frames of video, increasing the number of
samples obtained from each subject and improving
the repeatability of the video technique [13].
We have recently built a video-based phakome-
ter suitable for use in infants (Figure 4.2). The unit
consists of a CCD camera equipped with a fixed-
focus lens mounted on an extension tube. The
working distance from the infant is set at 25 cm,
ch to obtain adequate magnification and
n. but far enough away to be out of the
h or to have the baby taking a bottle
ocking the phakometer. The light sources
diodes (LEDs). The room lights can therefore be
i without dazzling the baby with bright lights.
CCD cameras have excellent sensitivity in the in-
frared once the infrared filter is removed. The
kometry LEDs are arranged in two pairs, one to
measure the vertical meridian and one for the hori-
zontal meridian. Additionally, three rings of eight
dimmer keratometry lights are mounted on the face-
plate of the unit in order to measure infant corneal
toricity and topography.
REFRACTIVE ERROR IN INFANCY
The typical newborn is hyperopic under cycloplegia
(Table 4.1). Cycloplegia is a valuable tool in refract-
ing the newborn to reveal any latent hyperopia.
Many infants who are myopic by noncycloplegic
retinoscopy can be hyperopic after instillation of a
cycloplegic, such as atropine [14]. Mean refractive
error is also less hyperopic when near retinoscopy is
used compared with cycloplegic retinoscopy [15].
The distribution of newborn refractive error also
lacks the sharply peaked, leptokurtic distribution
typical in children and adults. On the other hand,
myopia is not uncommon in newborns, with preva-
lences ranging from 0 to 25% (see Table 4.1) [16].
Figure 4.3 shows the normal distribution skewed to-
ward hyperopia, which is typical of neonatal refrac-
tive error. During ocular growth, the process of
emumetropization results in a reduction in the aver-
age amount of hyperopia in infancy, as well as a re-
duction in the variance of refractive errors. This
process is responsible for transforming the distribu-
tion of refractive errors from normal to narrow. Re-
cent evidence suggests that the highest rates of
emmetropization seen during the first 12-17 months
of life occur in infants with the highest initial
ametropia, whether hyperopic [17, 18] or myopic
[19]. The mechanism underlying emmetropization
is unknown. Based on considerable work in animal
models of refractive error, it has been hypothesized
that the eye senses the sign and magnitude of its re-
active error, then modulates its rate of growth
through an active. visual feedback mechanism in
order to reduce that error [20]. Eyes myopic at birth
would reduce their refractive error by growing
JJowly while the cornea and lens lost power during
development. and neonatal hyperopic eyes would re-
duce that error by growing more rapidly.
An interesting feature of infant refractive error
is the high prevalence of against-the-rule astigma-
tism (Table 4.2). Anywhere from 17% [21] to 63%
[22] of infants may have astigmatism of more than
1.00 DC. Most reports place the orientation as
against-the-rule in 40-100% of cases. There are
several potential sources for this astigmatism, such
as a large-angle lambda [23], off-axis peripheral re-
fractive astigmatism, or a change in accommodation
between measurement of the two principal meridi-
ans during noncycloplegic retinoscopy [24]. From
photokeratometric measures, Howland and Sayles
[6] showed that this astigmatism is primarily
corneal, especially in infants younger than 1 year
old. This astigmatism is also transient. The preva-
lence of refractive astigmatism 21.00 DC decreases
rapidly during the first year of life, reaching levels
found in childhood by the age of 18 months [25] to
2 years [26]. Interestingly, this is similar to the time
when the corneal radius of curvature has stabilized.
OCULAR COMPONENT DEVELOPMENT
IN INFANCY
The infant eye presents an interesting puzzle in oc-
ular development. The average eye of a newborn is
about 17 mm in length [7-9, 27-29], with a corneal
power of about 49 D (Tables 4.3 and 4.4) [3-5, 30].
During the first 5 years of life, the eye grows an-
other 4 mm in length on average. If uncompensated
by changes in the other ocular components, this in-
crease in length would result in a refractive error of
—-18 D. Obviously, other optical components must
change so that myopia can be avoided. The cornea
loses only approximately 5 D of power during those
years. The crystalline lens is the component that un-
dergoes the greatest dioptric change to create and
preserve emmetropia. During the first 5 years of
life, it loses an average of 20 D of power. The sur-
prising feature of this phase of ocular development
is that despite this substantial rate of growth, the
prevalence of myopia is low by age 6 years (about
2%), lower perhaps even than at birth. These
changes in ocular dimensions occur rapidly. Both
axial length and corneal curvature undergo the ma-
jority of their development in the first 2 years of life
(Figure 4.4). Although the overall power of the eye
decreases from 90 D at birth to 75 D at age 12
months [7, 31], studies have shown little residual
spherical equivalent refractive error by age 12
months, averaging approximately +1.00 D, with stan-
dard deviations of less than 2.00 D [14, 32-35].
Ocular growth is much slower in childhood, in
contrast to the rapid changes occurring during the
first 2 years of life. During the age period of 6-14
years, the eye grows only about 1 mm, the lens loses
another 3.00 D of power, and the cornea remains vir-
tually unchanged (flattening by 0.10 D per year)
[36]. It is puzzling that during this period of slow
growth when the increases in axial length should
present little challenge to the compensating mecha-
nisms of the eye, the prevalence of myopia increases
from 2% at age 6 to 15% by age 15 years [1]. Is the
eye growing more rapidly than it was during in-
fancy? Inspection of Figure 4.4 shows that this is
probably not the case, but clearly the eye is growing
more rapidly than is appropriate—that is. more
rapidly than the ability of the other components to
compensate. An alternate possibility is that some
mechanism that keeps the axial length coordinated
with the focal length of the eye breaks down.
PREMATURITY AND REFRACTIVE ERROR
Several factors may interfere with normal refractive
and ocular component development, such as low
birth weight or pathology that deprives the eye of
normal, high-contrast form vision (e.g., cataract).
Lorenz et al. [37] studied the impact of unilateral
and bilateral congenital cataract surgery followed
by contact lens correction in the first year of life.
Refraction was measured by retinoscopy; corneal
radius was inferred from the base curve of contact
lenses worn by the infants; and the axial length was
measured by A-scan ultrasound. Eyes with a unilat-
eral cataract were longer at the time of surgery, then
appeared to grow at a normal rate compared with
unaffected children. Corneal radius also flattencd at
a rate similar to that for normal infants. Bilateral
cataract patients tended to have initially shorter eyes
than expected for their age, accompanied by a
higher initial aphakic refractive error (approxi-
mately +35 D, compared with approximately +29
D for the unilateral cataract group). The bilateral
cataract patients continued to display slow rates of
growth, ending up with shorter eyes than expected
at age 5-8 years. Their corneas did not appear to
flatten during eye growth. The net impact on re-
fractive error of these two trends was that both
groups decreased in hyperopia by about 13 D dur-
ing the first 3 years of life in a nearly linear man-
ner. The bilateral group was always more hyperopic
than the unilateral group. These rates are similar to
those found by Moore [38] for infants with unilat-
eral cataracts operated on in the first 6 months of
life and corrected with contact lenses. Moore found
that a polynomial provided a better fit to the rate of
decrease than a line—refractive error = 31.897 —
(0.753 – age in months) + (0.013 – [age in months?)
— (0.00005556 – [age in months |).
The finding that changes in both axial length and
corneal radius of curvature occur slowly in infants
who have undergone surgery for bilateral cataracts
is consistent with the strong relationship that exists
between these two components in infancy. The cor-
relation between corneal radius and axial length in
infants is 0.69, indicating that flatter radii are typi-
cally associated with longer eyes [39]. This corre-
lation also continues into childhood and beyond but
at a reduced level of correlation of 0.29-0.31 [40,
41]. The correlation between birth weight and axial
length is also very significant at 0.66, consistent
with the eyes of premature babies being smaller
than those born at term [9, 31]. Tucker ct al. [42]
also found high correlations between birth weight
and axial length (r = 0.866), as well as axial length
and postconceptual age (r = 0.906). Strangely, the
correlation between axial length and refractive
error, which is so significant in childhood and adult-
hood (r = —0.76) [40], is either not seen in infancy
(r = =0.132) [39], or is somewhat reduced (7 =
~0.49) [9]. This may be due to the higher degree of
correlation between axial length and corneal radius
described above.
Despite the shorter axial length of the premature
infant eye, it is typically less hyperopic than that of
an infant born at term. Dobson et al. [43] found that
the average refraction of 146 premature infants was
-0.55 D, which is much more myopic than expected
for full-term newborns. In a study of 380 children
between the ages of 6 months and 3.5 years who
had birth weights of less than 2,000 g and no
retinopathy of prematurity (ROP), Shapiro et al.
[44] found that 5% displayed myopia. The correla-
tion between cycloplegic (cyclopentolate) refractive
error and birth weight was low (r <0.28). Scharf et
al. [45] found that 43% of babies weighing less than
2,500 g were myopic. Refractions fell into a bi-
modal distribution, with peaks centered on low hy-
peropia and moderate myopia. The use of a milder
cycloplegic agent in this study, mydriaticum, may
explain the greater prevalence of myopia. Again, no
association was found between the prevalence of
myopia and birth weight.
CRYSTALLINE LENS DEVELOPMENT —
PHAKOMETRIC RESULTS
The crystalline lens appears to be more important
than the cornea for offsetting the potential for my-
opia posed by axial elongation. The lens undergoes a
greater amount of dioptric change than the cornea
by a factor of 4-5, and its development continues
during the period when the majority of juvenile my-
opia has its onset. The cornea appears to have its
adult radius of curvature by the age of 2. Despite the
importance of the crystalline lens. relatively little is
known about its development. Because phakometry
has not been applied to the developing crystalline
lens in vivo, knowledge of the infant lens comes
from data taken nearly 90 years ago on a small series
of frozen cadaver eyes [11]. The infant schematic
eye of Lotmar is based on these limited data [48].
We measured the crystalline lens radii of curvature
of 27 infants ranging in age from 3 months to 18
months using the video-phakometric technique [49].
Fifteen of the infants were boys, and 12 were girls.
There were 25 white and two black infants.
Measurements were taken using cyclopentolate
1% cycloplegia (1 drop) following 1 drop of 0.5%
proparacaine. The infants were refracted by spot
retinoscopy 20 minutes after drop instillation, with
each meridian neutralized separately for an estimate
of refractive astigmatism. Corneal and lens curva-
tures were measured using a portable, hand-held
video-based keratophakometer similar to the one
described above. This unit consists of a series of in-
frared LEDs mounted on a clear, plastic faceplate.
Two pairs of bright phakometric LEDs are mounted
on the right side and at the top of the faceplate. The
angle between the phakometric LEDs and the center
of the disc was 40 degrees at the infant’s eye, with
17-degree separation between each member of the
pair. Dimmer infrared LEDs are arranged in three
concentric circles to measure corneal radius of cur-
vature in eight meridians. Results are reported for
the horizontal meridian only.
The infrared LEDs in the instrument must have
a margin of safety with respect to cataract forma-
tion. The infrared radiance of the keratometric and
phakometric LEDs was determined using a Photo-
dyne 44XL radiometer. The radiance of a single
keratometric LED was 0.01 W/cm®sr, and that of
a single phakometric LED was 0.166 W/cm?sr.
The irradiance of the 880-nm infrared sources at
the cornea was calculated at 0.34 mW/cm? for the
24 keratometric LEDs and 0.47 mW/cm? for the
two phakometer LEDs. The maximum permissible
exposure is less than 10 mW/cm?® using the 1988
American Conference of Government Industrial
Hygienists’ standard for infrared radiation beyond
770 nm [50]. This is also below British Standard
7192 of less than 200 mW/cm? for exposures more
than 10 seconds. The margin of safety was. there-
fore at least a factor of 20 compared with pub-
lished standards.
As previously mentioned, ultrasonography is dif-
ficult to perform on awake infants. At the time of
this study, we had not yet used the technique of per-
forming ultrasonography with the probe over the
closed lid. In this analysis, we used age- and sex-
appropriate values for ACD, lens thickness (LT), and
VCD from Larsen [28, 29, 51] to obtain lens radii
and refractive index from the phakometric data.
The median cycloplegic refractive error in the
horizontal meridian was +1.50 D—in good agree-
ment with expected values for infants in this age
range [15, 32-35]. There is a significant trend to-
ward reduction of this hyperopia with age (Figure
4.5A, r=-047, p=.043).
The median radius of corneal curvature in the
horizontal meridian of the infants in this study was
7.76 mm (43.5 D), ranging from 7.35 to 8.46 mm
(45.9-39.9 D). Interestingly, this was somewhat
flatter than expected considering previously re-
ported values (see Table 4.3), yet there was also no
correlation between the corneal radius of curvature
and age (Figure 4.5B, r= 0.00, p = .99).
The infant lens is 3.4-4.0 mm thick at birth (see
Table 4.4). Examination of changes in lens wet
weight indicates that the lens grows most rapidly
during the first 2 years of life, doubling in wet
weight during the first 8-10 years of life. Lens
growth in later childhood through adulthood is
much slower, with a further doubling in weight
again during the course of adult life (Figure 4.6)
[52]. Interestingly, despite this rapid increase in
lens substance in infancy, the thickness of the lens
is fairly stable throughout this time, remaining at
an average of 3.6 mm at the age of 6 years. From 6
to 10 years of age, the lens actually thins by about
0.2 mm [29, 53]. After age 10, lens thickness is
again constant on average in childhood and may
begin to increase in the early teens. Although it is
a textbook truism that the lens grows throughout
life, it appears that it does not always increase in
axial thickness. The lens may lay down new fibers
throughout life, but forces discussed below must
be at work to distribute these new fibers into a
thinner profile.
In our study, the median anterior and posterior
lens radii of curvature in the horizontal meridian
were 8.7 and 5.6 mm, respectively, much flatter
than the 5.0- and 3.7-mm values given by Lotmar
for the infant schematic eye [48]. No age-related
trends were observed for either anterior or posterior
lens radius over the range of ages tested in this
study at this sample size (see Figure 4.5C, r= 0.25,
p =.30; see Figure 4.5D, r = 0.00, p = .99).
If the infant lens had the equivalent refractive
index of 1.43 provided by Lotmar but had the sur-
face curvatures found in this study, it would
clearly have too little power and would result in
more than the +1.50 D of hyperopia seen in these
infants. To obtain the correct match of lens di-
mensions and measured refractive error, the
equivalent index must be higher, namely 1.49.
This index of refraction has a significant negative
correlation with increasing age (see Figure 4.5E,
=-0.73, p <.0001). Lens power also decreases
with age (see Figure 4.5F, r = 0.93, p <.0001).
Since curvatures are changing substantially with
age, the major contributor to this decrease in lens
power must be the decrease in equivalent index.
The lens power calculated from the median cur-
vatures and equivalent index in Table 4.5 is 44.8
D in infancy and 25.0 D at age 6 years. If the
equivalent refractive index remained at 1.49
throughout infancy, the decrease in lens power
due to curvature changes alone would be 49D,
or only 25% of the 19.7 D of power loss over
those years. The decrease in equivalent index
from 1.49 to 1.431 accounts for the remaining
14.8 D, or 75% of the power change.
The flatness of crystalline lens curvatures indi-
cates a degree of maturity in infant lens shape re-
sembling that of a child. Gwiazda et al. [54] have
shown that refractive error in later childhood may
be associated with the noncycloplegic retinoscopic
result at the age of 1 year. Perhaps this maturity of
lens shape in infancy may provide some basis for
the similarity in refractive error between infancy
and childhood.
POSSIBLE ROLE FOR THE CRYSTALLINE
LENS IN EMMETROPIZATION
AND MYOPIA
Since refractive error is in some sense a structural
error—that is, the eye is either too long or too short
for its focal length—understanding the development
of the eye as a structural entity is useful. At best, it
may provide some insight into the etiology of re-
fractive error itself. In the previous section, we saw
how the radii of the infant crystalline lens are much
closer to those found in childhood than previously
thought, how critical lens power is to compensation
for axial growth, how equivalent index plays a
major role in these power reductions, and how the
crystalline lens maintains a nearly constant thick-
ness in infancy despite rapid growth, then thins in
childhood. Reductions in lens equivalent index most
likely indicate changes in the equatorial gradient
index profile, rather than a decrease in core or in-
crease in superficial cortical indices. The respon-
siveness of the equatorial gradient profile may be
an important characteristic of the process in which
lens power keeps pace with axial elongation
throughout childhood to maintain emmetropia and
to avoid myopia.
If one tries to envision how the eye might grow
to produce reductions in lens power through both
flattening of lens radii of curvature and changes in
the equatorial gradient index, as well as crystalline
lens thinning, it could be pictured as overall
growth in both the axial and equatorial directions
of the eye. Equatorial ocular growth might result
in crystalline lens stretching in the equatorial
plane, which in turn flattens lens radii, changes the
equatorial gradient index, and thins the lens. The
matching of axial and focal lengths to produce em-
metropia across species [55, 56] and evidence of
increased frequency of myopia and higher refrac-
tive error variability when normal visual input is
disrupted [57, 58] have prompted others to assume
that eye growth and the process of emmetropiza-
tion are controlled by visually guided feedback
loops [40, 59]. Compensation by the chick eye for
the defocus imposed by spectacle lenses is consis-
tent with refractive error arising from visually me-
diated processes [60, 61]. Although clear visual
input free from deprivation is obviously required
for normal refractive development, we propose
that mechanical effects from the connection be-
tween globe axis, equator, and crystalline lens
may create an additional simple, mechanical feed-
back loop. If rapid rates of ocular growth are
accompanied by similarly rapid rates of lens flat-
tening or changes in the distribution of the gradi-
ent of the refractive index of the crystalline lens
due to lens stretch, the focal length of the eye may
keep pace with the physical length of the eye.
This type of physically and optically coordinated
growth has been proposed by several previous in-
vestigators [40, 62-64].
This model may also have implications for the
etiology of myopia. In contrast to the traditional
view that eye growth only occurs in the axial
dimension, the eye probably grows in three dimen-
sions, displaying equatorial as well as axial expan-
sion. In infancy, the growth of the lens is well
matched to the growth of the eye, allowing for the
proportional growth that maintains emmetropia at
any eye size. In childhood years, this pattern is
maintained in emmetropic patients, in whom reduc-
tion in crystalline lens power still occurs in concert
with the eye’s increasing axial length. In myopic
patients, the lens may no longer be able to meet the
challenge of the continued growth of the eye, per-
haps because it has reached some physical limit,
such as an inability to thin further. Figure 4.7 shows
the thickness of the lens as a function of refractive
error status and age. In all ages above 7, myopic pa-
tients have thinner lenses than emmetropic patients,
and emmetropic patients have thinner lenses than
hyperopic patients [53]. In myopia, the crystalline
lens may not be able to make compensating power
changes in the larger eye sizes of myopic eyes. It
may have reached some intrinsic physical limit or
be constrained by anatomic limits on equatorial ex-
pansion. As the eye at risk for myopia continues to
grow and the axial length starts to exceed the focal
length, the crystalline lens may be unable to further
thin or to further decrease its power in order to in-
crease the eye’s focal length. The result is the ex-
cessive, uncompensated axial length characteristic
of myopia.
It is hoped that improved biometric techniques
will be used in the future to study infant refractive
and ocular development in longitudinal studies.
Such studies will document the course of eye
growth that maintains emmetropia, thereby provid-
ing a contrast with the patterns of eye growth that
produce myopia in children.
Chapter 7
Electrodiagnostics, Ultrasound,
Neuroimaging, and Photorefraction
Deborah Orel-Bixler
Technological advances have made noninvasive as-
sessment of the human visual system possible
through the use of electrodiagnostics, neuroimag-
ing. ultrasound, and photorefraction. This chapter
reviews these specialized techniques and their clin-
ical application for the diagnosis of vision disorders
and eye disease in the pediatric population. Electro-
diagnostics includes the electroretinogram (ERG),
electro-oculogram (EOG), and the visual evoked
potential (VEP) to assess function of the retina,
retinal pigment epithelium, and visual cortex, re-
spectively. Neuroimaging provides in vivo depiction
of normal and pathologic central nervous system
(CNS) structures. Diagnostic tests in neuroimaging
include computed tomography (CT), magnetic res-
onance imaging (MRI), and positron emission to-
mography (PET). A-scan and B-scan ultrasound
provide imaging of the orbit and measurement of
ocular structures. The chapter concludes with a re-
view of photorefraction, a photographic technique
that has received much interest as a means of vision
screening for amblyogenic factors in infants and
preverbal children.
ELECTRODIAGNOSTICS
Electrodiagnostics serve primarily as an adjunct to the
clinical impression of an ocular or visual system dis-
order. The ERG and EOG can provide information
that aids in the diagnosis of disturbance of retinal func-
“on. The VEP can provide information to aid in the
diagnosis of disturbance of the visual pathway up to
the level of the visual cortex. Often, electrodiagnostic
test results are abnormal before ophthalmoscopic
changes are noted. Furthermore, electrodiagnostics
can be used to verily the patient’s subjective account
of a vision problem as well as to assess visual func-
tion in patients from whom a subjective account is not
possible (e.g., infants, nonverbal children, and patients
with multiple disabilities).
Electroretinogram
The ERG is a measure of the bioelectrical response
generated in the retina in response to stimulation by
light. The clinical ERG is recorded with a contact
lens recording electrode under both light-adapted
(photopic) and dark-adapted (scotopic) viewing
conditions using single and multiple flash presenta-
tions. The amplitude (size of the response) and la-
tency (time for the response to occur) of the ERG
components are compared with normative data to
establish the clinical diagnosis.
Clinical Paradigm for the Electroretinogram
The components needed for recording the ERG
are a light source, recording electrodes, an ampli-
fication system, and a device for registering the
amplified response. [For more information on inter-
national standards for ERG protocol and technical
specifications, see references 1 and 2.]
The standard ERG recording electrode is the Burian-Allen corneal electrode, which is a ring of stainless steel (active electrode) surrounding a cen- tral polymethylmethacrylate core [3]. The attached lid speculum has a conductive coating of silver to serve as the reference electrode. An additional ground electrode is typically placed on the forehead or earlobe. The ERG recording electrodes and placement are illustrated in Figure 7.1. A topica. ophthalmic anesthetic and a methylcellulose cush- ioning solution are needed with contact lens elec- trodes. The pupils are maximally dilated to control retinal illumination. 18
The Burian-Allen electrode is available in sizes appropriate for adults as well as premature infants [3]. The ERG can also be recorded using skin elec- trodes rather than a corneal electrode. This is an al- ternative when testing noncooperative patients such as infants and children. However, the amplitude of the ERG measured with skin electrodes is reduced 10-100x that measured with corneal electrodes [4] and therefore may not be comparable to established norms. In one study, ERGS measured with infraor- bital electrodes were one-eighth the size of the flash ERG obtained with corneal electrodes; however, re- liable ERGS could be obtained when signal averag-
ing was used [5]. Although corneal electrodes re- main the preferred recording device, even in infants and children, recording from sedated or anesthetized young patients is not recommended as the first line of approach, since anesthesia or sedation may alter the electrophysiologic responses [6].
The typical light source used to elicit the ERG is a xenon-arc photo stimulator presented as a full field in a Ganzfeld dome (diffusing sphere) (Figure 7.2.) The ERG protocol states that the procedure should begin with a period of dark adaptation [1, 2]. The first response measured after dark adaptation is the rod response to either a dim white flash or blue light (equated to the white standard). Next, the max- imal response elicited by a single white standard flash is recorded in the dark-adapted eye. After a 10-minute period of light adaptation to a back- ground luminance that serves as a rod-suppressing background, the single-flash cone response is recorded. The cone-generated response to flicker concludes the testing.
Electroretinogram Components and Interpretation
The components of the ERG vary depending on the adaptation state, light level, and stimulus parame-ters used to elicit the ERG. These components include the early receptor potential, a-wave, b-wave, oscillatory potentials (OP), c-wave, d-wave, and flicker response.
The a-wave is a negative wave with a photopic or cone component (al) and a scotopic or rod compo- nent (a2). It originates in the retinal photoreceptors. The b-wave is a positive wave with a photopic com- ponent (b1) ascribed to cone activity and a scotopic component (b2) ascribed to rod activity. It originates in the bipolar layer, mostly in the cells of Muell Under scotopic, high-intensity flash conditions, O. 19 appearing as 3-5 wavelets on the ascending limb of a b-wave may be recorded using a high-pass filter setting. The OP originates in the inner nuclear layer (bipolar/amacrine cells) and is thought to be sensi- tive to oxygen deprivation. It may have limited value since it is difficult to elicit in normals. The 30-Hz flicker response is mediated by cones since rods do not follow flicker past 20 Hz.
The five basic ERG responses include a maxi- mal response in the dark-adapted eye, a response developed by rods (in the dark-adapted eye), oscil- latory potentials, a response developed by cones, and responses obtained to flicker (Figure 7.3). In- terpretation of the ERG and comparison with norms are based on the amplitude and implicit time of the ERG components. Amplitude of the a-wave is mea- sured from the baseline voltage to the maximum voltage of the a-wave. Amplitude of the b-wave is measured from the a-wave maximum to the b-wave maximum voltage. Implicit time is measured from stimulus onset to the corresponding waveform peak (see Figure 7.3).
Table 7.1 lists the “normal” range of ERG values for amplitude and implicit time under maximum- intensity stimuli conditions. Each electrodiagnostic laboratory needs to establish its own norms based on stimulus intensity [7].
Value of the Electroretinogram in the Infant
Developmental Changes in the Electroretino- gram. An ERG can be recorded from infants within a few hours after birth. Implicit times are pro- longed compared with adult values [8]. By 2 months of age, the photopic component is comparable to adult values, but the dark-adapted b-wave is about half the amplitude of an adult’s [8]. The dark- adapted b-wave amplitude reaches adult values by 1 year of age [9]. The largest developmental changes in the ERG occur in the first year. In general, the in- terpretation of electrophysiologic responses depends upon comparisons to age-matched norms. It may be difficult to interpret the results when an infant has clinically abnormal vision yet a recordable ERG. Often serial testing will be necessary for a conclu- sive diagnosis [10].
Clinical Value of the Electroretinogram in In-
fants. The ERG is particularly useful in the differ-
ential diagnosis of four disorders of the retina that
produce vision impairment and nystagmus in an in-
fant without producing obvious retinal or eyeground
changes. These disorders include (1) achromatopsia,
(2) albinism, (3) congenital stationary night blind-
ness, and (4) Leber’s congenital amaurosis.
ACHROMATOPSTA. Achromatopsia, or rod mono-
chromacy, has an autosomal recessive inheritance
pattern and an incidence of 0.003% [11, 12). Tt is
characterized by extremely reduced or absent cone
function resulting in reduced visual acuity (20/160
on average) [13], very little if any color vision,
nystagmus, and significant photophobia [12, 14].
The fundus appearance is normal. :
The diagnosis of achromatopsia can be confirmed
by an ERG that shows normal or near-normal dark-
adapted (scotopic) responses but extremely reduced
or absent light-adapted (photopic) responses (Fig-
ure 7.4) [15]. =
There is no medical treatment for achromatopsia
per se. Appropriate refractive error correction and
use of low vision aids [16] and adaptations to com-
pensate for the photophobia, such as wearing sun-
glasses with very low transmission and squinting by
partial closure of the eyelids, will help improve the
person’s ability to function [17]. Standard gray sun-
glass tints generally do not provide adequate relief
from the photophobia since they are rarely made
dark enough (<1% transmission). Individuals with
achromatopsia are effectively light blind due to rod
saturation in bright lights. Since rods are very in-
censitive to long-wavelength radiation (red light),
red-tinted lenses with side shields turn daylight into
dusk and allow affected individuals to maintain op-
timal vision function even in bright daylight (Plate
7.1). Red lenses have anecdotally been reported to
be an effective intervention for photophobia in
adults [17]. In children with achromatopsia, carly
intervention with red lenses corresponded to a more
rapid improvement in visual acuity and earlier al-
tainment of adult-like levels [18].
Because cone function is virtually absent in
achromatopsia, the condition is of particular inter-
est in vision research since the rod visual system
can be studied independent of cone intrusion. The
development of visual acuity of the rod visual
system has been studied by measuring visual acuity
development in achromats. The rod-mediated vi-
sual acuity in achromats is not fully developed at
birth and improves during the first three years of
life. Rod-mediated visual acuity shows a similarly
shaped developmental time course to cone-
mediated visual acuity but it is delayed in time
compared to cone acuity [19].
Blue cone monochromatism should be consid-
ered a possibility in males with nystagmus, poor vi-
sion, and photophobia that is less severe than that
of rod monochromats. Blue cone monochromatism
is an X-linked recessive disorder. Affected males
have normal rod and blue (short-wavelength sensi-
tive) cone function [14] but dichromat color vision
at low photopic light levels, which can be revealed
with the Berson test for color vision [20]. The ERG
findings are identical to those in the rod mono-
chromat. The differential diagnosis for blue-cone
versus rod monochromacy is accomplished through
mode of inheritance and psychophysical testing in-
cluding measures of spectral sensitivity and the
Berson test. Magenta-tinted lenses are recom-
mended to protect the rods from saturation and pro-
vide relief from the photophobia but to also allow
retention of dichromat color vision by permitting
transmission of blue light [21].
ALBINISM. Albinism is a heterogeneous group of
congenital hypomelanotic disorders with at least 10
distinct forms of oculocutancous albinism and four
types of ocular albinism [22]. Mild forms of al-
binism are inherited as an autosomal dominant trait,
severe forms are inherited as autosomal recessive
traits, and ocular albinism is typically an X-linked
recessive trait. In ocular albinism, the hypopigmen-
tation of the iris and retinal pigment epithelium
manifests clinically as iris transillumination, macu-
lar hypoplasia, and chorioretinal hypopigmentation.
The extent of ocular involvement varies greatly but
nystagmus, reduced vision, and photophobia are
usually present. The reduced visual acuity in al-
binism (20/70 or worse) is due to foveal hypopla-
sia. Rods are present in the fovea and cones are
distributed away from the fovea [23]. Mild ocular
or oculocutaneous albinism have often been misdi-
agnosed as idiopathic congenital nystagmus [24].
Electrodiagnostic testing can aid in the differen-
tial diagnosis of albinism. The amplitude of the sco-
topic ERG exceeds the normal range in all forms of
albinism. The most sensitive and specific test to di-
agnose albinism, however, is the VEP. There is an
asymmetry in the amplitude of the VEP recorded
from the cerebral hemispheres [25]. This asymme-
try reflects the underlying anatomic miswiring in al-
binism. Axons from the temporal retina (within 20
degrees of the vertical midline) erroneously cross at
the chiasm, resulting in an abnormal layering of the
lateral geniculate nucleus and an abnormal visual
pathway to the visual cortex.
The appropriate interventions for ocular albinism
include tinted lenses for photophobia and low vi-
sion aids [23]. Since color vision is normal in al-
binism, red-tinted lenses are inappropriate, since
this tint reduces color discrimination abilities.
CONGENITAL STATIONARY NIGHT BLINDNESS. There
are several forms of congenital stationary night
blindness: autosomal recessive, autosomal domi-
nant, and X-linked inheritance types. The X-linked
disorder is the most difficult to diagnose due to the
normal fundus appearance. Clinical manifestations
of congenital stationary night blindness (CSNB) in-
clude poor night vision. myopia in the range of
—4.00 to —8.00 D, nystagmus. and reduced visual
acuity ranging from 20/40 to 20/200 [14]. Periph-
eral visual fields are normal. and the dark adapta-
tion curve shows no abnormalities of the cone
photoreceptors in children able to complete the task.
The disorder is nonprogressive.
The major abnormality shown in the ERG,
which is also diagnostic, is the absence of a posi-
tive response (b-wave) in the scotopic ERG [26,
27]. The initial negative response {a-wavc) is nor-
mal. The presumed abnormality is one of neural
transmission in the bipolar cell layer [28].
LEBER’S CONGENITAL AMAUROSIS. Leber’s con-
genital amaurosis [29] is an autosomal recessive
disorder characterized by blindness at birth. the
development of roving eye movements or nystag-
mus during the first few months after birth, slug-
gish or absent pupillary responses, high ametropia
(more commonly high hyperopia) [30], and a rela-
tively normal fundus during infancy. In one study
of 45 children with Leber’s congenital amaurosis,
75% had a normal fundus appearance during in-
fancy [31]. By 8 years, diffuse pigmentary stip-
pling may be seen, although up to 50% have a
normal pigmentary pattern and some have macu-
lar dysplasia and pigmentary abnormalities of the
peripheral retina. Pigmentary retinopathy with
bony spicules, attenuated retinal arterioles, and
optic atrophy may emerge gradually [32]. It is be-
lieved that the attenuated arterioles are present at
birth but rarely detected.
The visual acuity of children with Leber’s
ranges from no light perception to 20/200. It is rare
that it is as good as 20/200, however [31]. Persis-
tent pressing on the eye with the finger or fist is
known as the oculodigital reflex and is characteris-
tic of children with Leber’s and other severe reti-
nal disorders [33]. It has been suggested by Jan et
al. [34] that this pressure stimulates the visual cor-
tex by mechanically triggering ganglion cell action
potentials, thus producing phosphenes or entoptic
flashes of light in the retina. This persistent eye
pressing should be discouraged for eye health. Eye
pressing disperses the surrounding orbital fat and
causes the eyes to appear sunken with time as well.
Tt is estimated that Leber’s accounts for 10-18%
of childhood blindness [32] and may be isolated or
associated with systemic conditions such as Jou-
bert’s syndrome, which is characterized by a spe-
cific malformation of the cerebellum, cerebral
vermis hypoplasia, and oculomotor anomalies and
respiratory problems in the neonatal period [35].
The ERG is the definitive test for establishing
the diagnosis of Leber’s congenital amaurosis in an
infant with visual impairment and nystagmus but
normal ocular findings. Both the light-adapted
(photopic) and dark-adapted (scotopic) ERG com-
ponents are markedly reduced or absent [33].
Differential Diagnosis for Vision Impairment in
Infancy. Congenital retinal disorders are usually
accompanied by nystagmus, habitual pressing of the
eyes, sluggish pupillary responses or pupillary con-
striction in darkness (CSNB), and highly myopic
(CSNB) or hyperopic (Leber’s) refractive errors.
The ERG aids in the early differential diagnosis of
these disorders.
An ERG is not necessary in other disorders that
cause nystagmus and vision impairment in infants
when abnormal ophthalmoscopy findings are pres-
ent. Nystagmus and visual impairment in infancy
may be associated with ocular abnormalities or
media opacities such as bilateral congenital
cataracts or corneal opacities, macular scarring, vit-
reous hemorrhage, retinopathy of prematurity,
optic nerve hypoplasia, and optic nerve atrophy.
With careful examination, the anomalies of the
optic disc are detectable. Fundus signs and nystag-
mus are absent in infants with vision impairment
due to cortical visual impairment and delayed vi-
sual maturation. In cases of suspected cortical vi–
sual impairment, delayed visual maturation, and
optic nerve anomalies little information is gained
from an ERG. Instead, a referral for VEP testing
and neuroimaging would be recommended (for
more information on these procedures, see the sec-
tions on neuroimaging and VEP).
Value of the Electroretinogram in Children
Most infantile retinal dystrophies display severe rod
and cone involvement from onset. Retinal dystro-
phies with an onset in childhood involve predomi-
nantly the rods or the concs at presentation. The
differential diagnosis for retinal dystrophies with an
onset in childhood is obtained by comparison of the
photopic and scotopic ERG. Fundus abnormalities
are often not present early in the disease.
Retinitis Pigmentosa. Retinitis pigmentosa (RP),
a progressive rod-cone dystrophy, is a genetically
heterogeneous group of disorders characterized by
night blindness. visual field loss, and an abnormal
or extinguished ERG [36]. RP inherited in the X-
linked or autosomal recessive mode tends to have
an earlier onset and more severe involvement than
the autosomal dominant type. The dystrophic
process primarily affects rods.
Although children with RP may initially present
with night blindness or visual field loss. the appear-
ance of the fundus in the early stages of RP is vari-
able and abnormalities may be very subtle in young
children. Mild pigment epithelial atrophy in the
midperiphery appears early and is followed by pig-
ment deposition in the equatorial retina with nar-
rowing of the retinal arterioles and a waxy, pallid
appearance of the optic disc. In more advanced dis-
ease, bone spicule pigmentation is seen with optic
disc pallor and retinal arteriole attenuation. Visual
acuity may decline due to associated posterior sub-
capsular opacities, macular edema, or macular in-
volvement in the dystrophic process.
Children with RP often lack ophthalmoscopic
findings early in the disease; however, the scotopic
ERG is absent or reduced in amplitude at an early
stage. The ERG photopic responses may be re-
duced in amplitude but the flicker response may be
preserved [36].
Usher’s Syndrome. Usher’s syndrome is an au-
tosomal recessive disorder characterized by RP and
a severe, congenital neurosensory hearing loss [37].
In type I Usher’s, the scotopic ERG is usually ab-
sent. Profound deafness and an absent or abnormal
vestibular response may also be accompanied by
mental retardation, ataxia, or psychosis. In type II
Ushers, a scotopic ERG of reduced amplitude may
be recorded. Hearing loss is variable. Some patients
develop intelligible speech and normal vestibular
responses. Other neurologic problems are rare [38].
Included in the differential diagnosis for Usher’s is
congenital rubella with its accompanying pigmen-
tary retinopathy and profound deafness. The ERG
is normal in congenital rubella syndrome [39],
whereas the scotopic ERG is absent or reduced in
amplitude early in Usher’s syndrome.
Other Causes of Pigmentary Retinopathy. The
differential diagnosis for pigmentary retinopathy
may be aided by ERG testing. In congenital rubella
syndrome, the most common ocular abnormality is
pigmentary retinopathy, which is found in up to
40% of cases [40]. The ERG findings are normal
[39], however, and no visual impairment accompa-
nies the disorder. Pigmentary retinopathy may be
associated with infectious retinopathy from con-
genital syphilis and in severe cases may result in ex-
tensive pigmentary changes resembling RP. This
form of chorioretinitis is referred to as pseudo-
retinitis pigmentosa. In the chorioretinitis associated
with congenital syphilis, diffuse chorioretinitis as-
sociated with virus inclusion, and Leber’s congeni-
tal amaurosis, the ERG is abnormal and visual
impairment accompanies the disorder. Attenuated
retinal arterioles may be the sole finding in an in-
fant with syphilis or Leber’s congenital amaurosis.
The ERG distinguishes between these two condi-
tions. In Leber’s, the ERG is severely reduced; in
syphilis, the ERG is moderately reduced.
Cone-Rod Dystrophy. Cone-rod dystrophy is a
hereditary, degenerative retinal disease that primar-
ily affects cones. Unlike achromatopsia, which pre-
sents in infancy and is stationary, the progressive
cone dystrophies are not usually symptomatic until
late childhood or early adulthood and progressive
vision loss accompanies the disorder [14]. Clinical
signs of cone-rod dystrophy include a bull’s eye
maculopathy with symptoms of photophobia and
progressive loss of central and color vision [41].
In cone-rod dystrophy, the photopic ERG is ab-
normal. The flicker and photopic b-wave are char-
acteristically absent and the response to a bright
flash is decreased. The scotopic b-wave amplitude
may be normal.
Electroretinogram Versus Neuroimaging
Neuroimaging, such as an MRI or CT scan, is not
usually indicated in children with retinal disorders
unless other neurologic abnormalities or develop-
mental delays are present [42]. The ERG is se-
verely reduced in the retinal disorder Leber’s
congenital amaurosis, which may be accompanied
by optic atrophy. Other causes of optic atrophy.
without any retinal disorder. have normal ERG
findings [43]. Therefore. neuroimaging studies are
vital in all infants with optic atrophy or optic nerve
hypoplasia [44, 45].
The ERG confirms the diagnosis in many retinal disorders associated with visual impairment or blindness and nystagmus during infancy. Leber’s congenital amaurosis, achromatopsia, albinism, and CSNB have a similar clinical presentation characterized by reduced vision and nystagmus, but the ERG can differentially diagnose these disor ders. In Leber’s congenital amaurosis, the photopic and scotopic ERG is extinguished. In achromatop- sia, the flicker and single flash photopic (cone) ERG is absent. In all forms of albinism, including oculocutaneous albinism, the scotopic ERG is su- pernormal, In CSNB, the scotopic b-wave is ab- sent. In retinal dystrophies with an onset in childhood (RP and cone-rod dystrophy), abnormal ERG findings may precede the ophthalmoscopic appearance of ocular anomalies.
Electro-Oculogram
The EOG is a measure of the relative electrical po- tential of the eye. The clinical EOG uses a compar- ison of the voltage difference between the cornea and the posterior pole of the eye in the light- adapted and dark-adapted state. Physiologically, the EOG is a measure of the slow changes in po- tential that occur as a result of alterations in the metabolism of the pigment epithelium [46]. Clini cally, the EOG is useful in the detection of ocular disorders, especially abnormalities of the retinal pigment epithelium.
Clinical Paradigm for Electro-Oculogram
The equipment required for recording the EOG in- cludes skin electrodes, a direct current or alternating
current amplifier for each eye, and a data storage device such as a strip chart recorder, storage oscil- loscope, or computer. The recording electrodes are placed on the skin close to the medial and lateral canthus. An additional recording electrode serving as the common ground is typically placed on the forehead. The silver-silver chloride or gold-disc skin electrodes are attached to the skin with double adhesive tape and a small amount of conducting electrode paste after the skin is cleansed with iso- propyl alcohol.
The EOG is recorded as the patient alternately fixates on two red light-emitting diodes (LEDs) which are typically presented in a Ganzfeld globe. The LEDs are horizontally separated so that the eyes rotate laterally through a constant angle of about 30 degrees. The EOG protocol requires the attention and cooperation of the patient, who must fixate on the alternately illuminated LEDs several times per minute, repeating these eye excursions at 1-minute intervals for up to 15 minutes in the dark followed by 12 minutes in the light [47]. The eye acts like an electrical dipole inducing an electrical field in the periocular tissue. The electrode in close proximity to the cornea of the eye registers a posi tive voltage while the electrode in close proximity to the fundus registers a negative voltage. The av erage amplitude of the EOG (maximum to mini- mum) during several eye excursions is calculated and plotted in microvolts as a function of time in
the recording session. In the EOG protocol, baseline recordings are obtained during a 5-minute adaptation period, with the background field of the Ganzfeld globe illuminated. The lights are extinguished and the saccadic eye movements producing the EOG are recorded at 1-minute intervals during 12-15 min- utes of dark adaptation. The amplitude of the EOG reaches a minimum within 8-9 minutes in the dark. This minimum is referred to as the firs dark trough. The background field is then illumi- nated and the EOG is recorded at 1-minute inter- vals during 12 minutes in the light. The EOG reaches a maximum amplitude after 8-9 minutes in the light. This maximum is referred to as the first light peak.
The EOG is interpreted by means of the Arden ratio [47], which is calculated using the ratio of the voliage for the light peak (LP) and the dark trough (DT):
Arden ratio = LP/DT x 100%
An Arden ratio of more than 180% is considered
normal, a ratio of 165-180% is marginally subnor- mal, and a ratio of less than 165% indicates pathol- ogy of the retinal pigment epithehum 17, 47]. Figure 7.5 shows the amplitude of the standing potential of the eye (in microvolts) for the right eye (circles) and left eye (squares) plotted as a function of time (in minutes) for the pre-, dark, and light adaption peri- ods in the EOG recording session. The dark trough
and light peak are indicated by arrows. The calcu-
lated Arden ratios are normal.
In general, interpretation of electrophysiologic
responses depends on comparison to age-matched
norms. No age-related changes in Arden ratios have
been found [10]; however, duc to variations in
recording equipment and artifactual electrical noise,
each laboratory needs to establish its own norms.
Children older than 5 years of age can generally
cooperate sufficiently to complete the EOG proto-
col. Infants have been tested by inducing passive
eye movements via vestibular reflexes [48]. Pa-
tients with less than 20/200 acuity may not be able
to see the fixation lights but could be requested to
make consistent, full excursion eye movements
during testing.
Ocular Diseases Associated with an Abnormal
Electro-Oculogram
To obtain a normal light peak in the EOG, rods and
the retinal pigment epithelium must be functioning,
the retina and the retinal pigment epithelium must
be in contact, and an adequate choroidal blood sup-
ply must be present. An abnormal EOG is found in
ocular disorders in which one of these conditions is
not fulfilled [14]. These ocular disorders include
rod-cone degenerations, cone degenerations, achro-
matopsia, CSNB, Leber’s congenital amaurosis,
albinism, toxic retinopathy, diabetic retinopathy,
retinal detachment, and vitelliform macular degen-
eration (Best’s disease).
Value of the Electro-Oculogram in Infants
The ERG is more useful than the EOG in the dif-
ferential diagnosis of retinal disorders in infancy.
The ERG findings in achromatopsia, albinism,
CSNB, and Leber’s congenital amaurosis may be
supplemented by BOG findings, which may con-
tribute supplemental information for the diagnosis.
The EOG is usually abnormal in progressive cone
” degenerations and normal in nonprogressive cone
disorders such as achromatopsia. The ERG is more
diagnostic for achromatopsia due to an absent
flicker and photopic ERG. Supernormal values are
obtained when the EOG is recorded in albinism;
these results are similar to those obtained with the
ERG. The combination of ERG and EOG results is
useful since the incomplete universal and ocular
forms of albinism are often difficult to recognize
clinically. In CSNB, the EOG is abnormal in the au-
tosomal dominant condition but appears normal in
the X-linked recessive and autosomal recessive
variants. The absence of the ERG scotopic b-wave
is most diagnostic for CSNB. The EOG may be pre-
dictive of the visual prognosis in Leber’s congenital
amaurosis. In this cone-rod disorder. the ERG is
typically extinguished; however, the EOG is abnor-
mal when the disorder is progressive and near nor-
mal in stationary forms of the disease.
Value of the Electro-Oculogram in Children
Vitelliform macular degeneration, or Best’s dis-
ease, is an autosomal dominant, pleomorphic, pro-
gressive, retinal pigment epithelium disease usually
manifesting in the second decade of life [49]. The
classic ophthalmoscopic finding is a well-defined,
egg yolk-like, yellow lesion beneath the pigment
epithelium in the macular area. There is no loss of
visual acuity during the early stages of the disease.
The clinical presentation of the macular lesion is
variable; it may include partially reabsorbed vitel-
Jiform cysts or appear normal. Best’s can usually
be diagnosed on the basis of the clinical evaluation
and examination of family members and inheri-
tance pattern. The BOG is particularly diagnostic,
since it always yields abnormal results regardless
of the severity of Best’s disease [50]. The light rise
in the EOG is abnormal, whereas the ERG is usu-
ally normal. This dichotomy between the ERG and
HOG is not typical and therefore is diagnostic for
Best’s disease (Figure 7.6) [28]. EOG can also con-
firm if a nonsymptomatic individual is a carrier of
the disease [51]. :
Value of Electro-Oculogram in Retinal Disorders
The EQG is abnormal in diffuse rod-cone degener-
ations such as RP, diffuse choroidal sclerosis, and
choroideremia [36]; however, the ERG is a more
sensitive and reliable indicator of early discase. The
EOG shows moderate to severe abnormalities in ad-
vanced stages of the following diseases, which are
classified as the flecked retina syndromes: Star-
gardt’s disease, fundus flavimaculatus, and fundus
albipunctatus [52]. Both the EOG and ERG may be
abnormal in toxic retinopathy in patients after pro-
longed use of chloroquine for the treatment of
rheumatoid arthritis or as an antimalarial drug.
However, the tests are not predictive of early reti-
nal toxicity [53]. Fortunately, the introduction of a
less toxic drug, hydroxychloroquine, has reduced
the occurrence of toxic retinopathy [54]. In a study
of diabetes, the EOG was abnormal before the
retinopathy was apparent clinically [55]. The degree
of retinal detachment is reflected in a progressively
abnormal EOG ratio as contact between the retina
and retinal pigment epithelium is compromised [7].
Summary
The EOG is an important adjunct test for the diag-
nosis of abnormalities of the retinal pigment epithe-
lium. It is particularly useful in the diagnosis of
Best’s disease, since the EOG is always abnormal in
the presence of this condition. Its application to the
infant or young child is limited, since consistent and
repetitive cye movements are required for testing.
Visual Evoked Potential
Definition
The VEP is an electrical signal generated in the oc-
tal region of (he cortex in response to visual
stimulation [56]. Also referred to as the visual
evoked response or the visual evoked cortical po-
tential, the VEP is a specific occipital lobe response
to visual stimuli. The VEP can be isolated from the
background electroencephalogram (EEG) by re-
cording the VEP in relation to a time-locked stimu-
lus presentation. The VEP is elicited at a designated
time after the presentation of the stimulus, whereas
the EEG is random in its timing with respect to the
visual stimulus. If the VEP to a specific number of
stimuli is recorded, the VEP will be continuously
added at equal and constant intervals in time,
whereas the background EEG noise will average
out to zero [56].
Clinical Paradigm for Recording the Visual
Evoked Potential
The VEP is recorded with EEG-type electrodes and
1-cm gold cup disks, which are adhered to the scalp
with an electroconductive paste or gel and gauze
(Figure 7.7). Before electrode placement, small re-
gions of the scalp arc cleansed with a mild abrasive
gel containing pumice to reduce electrical imped-
ance. As in other electrodiagnostic recordings, a
minimum of three electrodes are needed: the active,
reference, and ground electrodes. Due to placement
of the recording electrodes over the occipital cor-
tex, underlying cortical anatomy, and cortical mag-
nification, the VEP assesses primarily foveal
projections to the visual cortex. One millimeter of
tissue in the cortex is devoted to 2 minutes of visual
angle when the cortical projections originate from
the fovea [57]. Amplification and computer averag-
ing or analysis is required to isolate the VEP from
the background EEG.
The VEP stimulus may be a [lash from a xenon-
arc photostimulator or patterned stimuli (checker-
boards or gratings) generated on a video monitor.
There are two presentation modes for patterned
stimuli: pattern-reversal, in which the pattern is
phase-reversed at a specific rate, and pattern onset-
offset, in which the pattern appears briefly and then
is replaced with a blank field of the same mean lu-
minance as the pattern [58]. It is important to keep
the overall mean luminance constant during either
the pattern-reversal or onset-offset presentations in
order to eliminate contamination of the pattern VEP
by luminance components.
Temporal stimulation in VEP recordings is ei-
ther transient or steady state. When recording the
transient VEP, the stimulus is presented bricfly at
regular, slow intervals, usually with less than 6-10
presentations per second. Repeated stimulus pre-
sentations and signal averaging is required to iso-
late the VEP response from the background EEG
noise. Since the noise is reduced by the square root
of the number of stimulus presentations, typically
up to 100 stimulus repetitions are presented to re-
duce the noise by a factor of 10 [59]. The transient
VEP is a complex waveform consisting of a series
of negative and positive components when plotted
as amplitude versus time (Figure 7.8) [58, 60, 61].
The luminance flash VEP is recorded after very
brief strobe flashes (see Figure 7.8, upper panel).
The transient VEP recorded to pattern onset-offset
stimuli (see Figure 7.8, middle panel) has three
components, with positive, negative, and positive
peaks called CI, CII, and CIII, respectively [60].
The transient VEP recorded to pattern-reversal
stimuli (see Figure 7.8, lower panel) shows a posi-
tive peak at latency around 100 ms called P100,
which is preceded and followed by negative peaks
(N1 and N2, respectively) [58].
The steady-state VEP is recorded in responsc (0
a continuous, fast, temporal presentation, with the
local changes in the pattern producing a response
that is approximately sinusoidal [62]. The steady-
state VEP is analyzed in the frequency domain and
expressed as the amplitude of the VEP at the tem-
poral frequency of the stimulus used to elicit the
VEP. The primary advantage of the steady-state
VEP is that stimuli can be more rapidly presented
than in the transient paradigm and thus may be less
affected by changes in patient cooperation [63]. The
swept parameter VEP technique [63, 64] enables
presentation of up to 20 different patterned stimuli
during brief, 10-second recording trials. This allows
for a better sampling of the relationship between
VEP amplitude and changes in the visual stimulus.
he improved speed in testing with sweep VEP
ques has increased clinical applicability.
VEP testing in pediatric patients is usually per-
formed in a darkened room to decrease distraction.
Attention to the video monitor is achieved by dan-
sling small fixation toys on the video screen and
speaking or singing to the infant or child being
tested. A pause/resume remote control is used by
the examiner to start and stop the VEP recording
during fixation lapses. Electrical artifact caused by
excessive movements of the subject can be elimi-
nated with appropriate data analysis techniques.
With skill of the examiner and rapid recording par-
adigms such as the sweep VEP technique, sedation
or general anesthesia is unnecessary.
The latency and amplitude of the pattern VEP is
sensitive to changes in contour and edges of the
stimulus. The amplitude of the major positive com-
ponent of the VEP decreases as the contrast or an-
gular subtense of the visual stimulus is decreased.
It is this consistent relationship between VEP am-
plitude and the visual stimulus that enables objec-
tive measurement of visual thresholds such as visual
acuity and contrast sensitivity.
To measure visual acuity, the amplitude of the
transient VEP is measured for a series of grating
patterns. Grating patterns are defined in terms of
spatial frequency or the number of cycles (light and
dark bar pairs) of the grating per degree of visual
angle (Figure 7.9). The amplitude of the P100 com-
ponent is plotted versus the spatial frequency of the
grating used to elicit the VEP. A straight line is fit
from the peak of the amplitude versus spatial fre-
quency function and is extrapolated to 0 pV to
yield the acuity estimate in cycles/degree (see Fig-
ure 7.9).
Conversion of grating acuity (measured in cy-
cles/degree) to Snellen notation is done using the
following equivalency:
Grating with 30 cycles/degree = 1 minute of arc
of visual angle (or 20/20 visual acuity)
It is important to note that this mathematical trans-
lation of grating acuity into Snellen notation is not
appropriate in some eye disorders with accompany-
ing low vision.
Estimation of acuity by extrapolation techniques
= dependent upon the number and range of pattern
sizes presented [64]. Maximal sampling of the am-
plitude versus spatial frequency function is ideal but
time consuming; therefore, transient VEPs have
limited clinical applicability for threshold measures
of visual acuity. A more rapid technique for mea-
suring visual acuity is the swept spatial frequency
VEP paradigm [64]. To determine visual acuity
with the sweep VEP, the spatial frequency of the re-
versing grating target is linearly incremented over
a 20 to 1 range of spatial frequencies during a 10-
second trial. The amplitude and phase of the steady-
state VEP at the reversal frequency is determined
using a discrete Fourier transform [65]. Acuity is
determined by an extrapolation to O uV of the func-
tion relating VEP amplitude and linear spatial fre-
quency (Figure 7.10).
The primary advantage of the sweep VEP is the
increased sampling of the spatial frequency func-
tion, with 20 different spatial frequencies eliciting
a VEP amplitude. In addition, the VEP trials are
scored by a computer that uses phase information.
local noise estimation, and signal-to-noise informa-
tion in its scoring criteria. The time advantage of a
10-second trial for an acuity measure allows for re-
peated measures and monocular and binocular
recordings within the clinical setting.
In normal adults, visual thresholds determined
with extrapolation to 0 pV for VEP amplitude agree
with psychophysical measures of contrast threshold
[58, 66, 67], grating acuity [68, 69], stereoacuity
[70], and vernier acuity [71]. Good correlations be-
tween psychophysical and VEP thresholds, with
changes in luminance, optical blur [68, 72] and reti-
nal location [73] have also been reported.
Visual Evoked Potential Studies of Normal
Visual Development
Maturation of the Visual Evoked Potential Wave-
form. The VEP shows maturational changes dur-
ing the first 6 months of life. In young infants, the
pattern VEP consists of a single positive peak [74,
75). With age, the VEP waveform becomes more
complex, the amplitude increases, and the latency of
the main positive component decreases [75]:
Assessment of Visual Function in Infancy.
Within the past two decades, there has been consid-
erable interest in the assessment of visual function
in infancy [for a review, see reference 76]. The two
methods effective in quantifying the visual capabil-
ities of infants are assessment of the “looking” be-
havior of infants in response to visual stimuli with
preferential looking (PL) techniques and the
recording of the VEP.
VisuaL Acuity. Use of VEP has shown that visual
acuity develops rapidly during the first year of life
(67, 74, 77, 78] and continues to develop into late
childhood [69, 74]. Adult levels of grating acuity
are reached as early as 6-8 months of age [67,
77-80]. VEP grating acuity develops rapidly until
the age of 8 months and then develops more slowly
until 5-11 years [69, 74].
VEP measurement of visual acuity develop-
ment indicates much more rapid growth than be-
havioral measures with preferential looking (PL)
techniques do. The development of grating acuity
during the first few years of life has been investi-
gated by several groups using forced-choice pref-
erential looking (FPL) techniques [81, 82] operant
FPL techniques [83], FPL acuity cards [84, 85],
and the acuity card procedure [36]. Studies have
generally agreed that PL grating acuity develops
from approximately 1 cycle/degree at 1 month-of
age and improves to 6 cycles/degree by 6 months
of age; however, PL grating acuity does not reach
adult levels until approximately 3 years of age
[85]. The 99% confidence limits for the mean
acuity are on average a factor of £2.4 octaves (an
octave is a factor of two in MAR [minimum angle
of resolution]) over the first year of life [82, 87].
Additionally, the 99% confidence limits for in-
terocular differences range from 0.5 octaves [86]
to 3.2 octaves [85].
The sweep VEP technique [80] had relatively
small test-retest variability and narrow confidence
limits in comparison to previously reported data for
PL techniques. The 99% confidence limits and in-
terocular differences for the sweep VEP were 2-4
times smaller than PL measures. Likewise, monoc-
ular VEP grating acuities were higher than PL acu-
ity by a factor ranging from 1 to 5 octaves.
Therefore, the sweep VEP technique may have
higher sensitivity than PL techniques and therefore
would better detect amblyopia or visual acuity
deficits in infants (Table 7.2).
The validity of the VEP measurement of visual
acuity has been supported by two studies of normal
populations. Optotype acuity measured with Lan-
dolt C targets was similar to VEP acuity (recorded
to checkerboard stimuli) in subjects 3-71 years old
[74]. Acuity determined by sweep VEP (recorded to
grating stimuli) and letter acuity was in good agree-
ment in normally sighted adults and in children
5-11 years old [69].
CONTRAST SENSITIVITY. The development of con-
trast sensitivity has been studied with the VEP [88,
29]. In a study using the sweep VEP, contrast sensi-
tivity for low spatial frequency grating targets (1
cvcle/degree) developed rapidly between birth and
10-12 weeks of age. The contrast sensitivity for
coarse targets of a 10-week-old infant was only a
factor of two lower than an adult’s [89].
STEREOPSIS. Petrig et al. [90] measured VEPs
elicited by dynamic random dot stereograms and
correlograms in infants. Stereoscopically evoked
potentials could be recorded at the age of 10-19
weeks, suggesting that the onset of cortical binocu-
larity precedes stereopsis. Braddick et al. [91]
recorded VEPs to the onset and offset of binocular
correlation in a large-screen dynamic random dot
display and reported that infants have a functional
binocular visual cortex by 3 months of age, with
some individuals showing cortical binocularity at
an earlier age.
CoLOR VISION. VEP studies indicate that infants
as young as 2 weeks of age have functional
medium- and long-wavelength—sensitive cones and
postreceptoral circuits, which relay information to
the visual cortex [92]. Infants as young as 5 weeks
of age have functional short-wavelength—sensitive
cones [93]. These findings do not imply. however,
that neonates have mature color vision because be-
havioral measures of color discrimination indicate
that development occurs later [94]. A complete
spectral sensitivity function for infants has been
recorded using the VEP to heterochromatic flicker
photometry [95]. The infant heterochromatic
flicker photometry functions had an elevation in
sensitivity at short wavelengths.
Clinical Applications for the Visual
Evoked Potential
Determination of Refractive Error. A small
amount of optical blur can attenuate the amplitude of
the VEP to centrally fixated fine checks [72]. Al-
though it is possible to estimate refractive error based
on the attenuation of VEP amplitude with increasing
dioptric blur, it is impractical and less accurate than
retinoscopy [96]. The dependence of VEP amplitude
on optical blur underscores the importance of cor-
recting the underlying refractive error when estimat-
ing visual acuity using VEP amplitudes.
Visual Evoked Potentials in Amblyopia: Con-
trast Sensitivity. Abnormal contrast responses
across a wide range of spatial frequencies have
been found using the pattern-reversal VEP in am-
blyopia [97]. These results were consistent with
psychophysical measures of contrast sensitivity
functions in individuals with strabismus and
anisometropia [98]. Differences in the contrast
sensitivity functions between strabismic and an-
isometropic amblyopia have also been measured
with the pattern-reversal VEP [99]. Individuals
with strabismic amblyopia showed an abnormal
contrast sensitivity function only in the high spatial
frequency range, whereas those with anisometropic
amblyopia showed an abnormal contrast sensitiv-
ity function both in the low and high spatial fre-
quency range.
Visual Evoked Potentials in Amblyopia: Visual
Acuity. Several studies have reported an abnor-
mal pattern VEP in patients with amblyopia
[100-102]. The pattern-reversal VEP of the ambly-
opic eye has smaller amplitudes than the fellow eye
[103] and smaller signal-to-noise ratios [97]. The
determination of which eye is amblyopic could be
made in 66% of the adults tested on the basis of the
interocular VEP amplitude difference; however, no
acuities were reported, so the magnitude of the acu-
ity deficit could not be inferred from the interocular
VEP amplitude difference [103].
A few studies have evaluated the usefulness of
the VEP in the diagnosis of amblyopia in pediatric
patients. Sokol et al. [104] calculated interocular
VEP amplitude differences for children with vari.
ous ocular disorders, including strabismus and an-
isometropia, and compared these with normals. The
interocular VEP amplitude ratios did not correlate
with the magnitude of the interocular differences in
Snellen acuity. Twenty-five of 26 children (96%)
with a two-fold (three lines) or more interocular dif-
ference in Snellen visual acuity tests bad abnormal
VEP interocular ratios. The proportion of abnor-
mals correctly identified dropped to 50% for an
acuity difference of two lines and to 30% for an
acuity difference of one line. Friendly et al. [105]
assessed the clinical usefulness of pattern-reversal
VEPs in the diagnosis of amblyopia in 27 children
with anisometropia and 4 children without ambly-
opia. Measurements of visual acuity with letter
charts were compared to normalized VEP ampli-
tudes to reversing checks subtending 15 minutes of
visual arc. Of the 31 children in their study, 25
(80%) were correctly identified with amblyopia by
VEP testing only. Odom et al. [106] reported the
usefulness of the VEP in monitoring acuity changes
during patching therapy in preverbal amblyopes.
Acuity was determined from an extrapolation to 0
pV of the transient VEP amplitude versus several
checksizes; however, neither normative data nor
validation of the VEP measures by comparison with
subjective acuities were reported. A limitation of
these VEP studies of amblyopia in children is that
the VEP measures of acuity were not validated by
comparison with optotype acuity [106] or a single
VEP amplitude criterion was used to predict the
amblyopic eye without yielding any information as
to the magnitude of the amblyopia [104, 105]. Stud-
ies using the sweep VEP technique enable rapid and
direct measures of visual acuity thresholds in a clin-
ical setting.
Validation studies using the sweep VEP reported
good correlations between VEP grating acuity and
optotype acuity in adults and children with strabis-
mus, anisometropia, or both and age-matched nor-
mals [69] and in children with various visual
disorders [107]. Sweep VEP grating and optotype
acuities were well correlated in amblyopia, in spite
of substantial absolute differences between the two
measures in patients with significant amblyopia
[69]. There was substantial agreement between VEP
and optotype acuity in observers with optotype acu-
ity better than 20/60; however, an increasing dis-
crepancy accompanied poorer acuity (Figure 7.11).
Tests using gratings overestimated acuity compared
with optotypes by a discrepancy factor of 2.5 in
MAR. Therefore, a measured grating acuity of 6
cycles/degree is expected to yield an optotype acu-
ity of approximately 20/250. The more exact con-
version factor [69] is determined using the
following equation:
This equation uses the loss of grating acuity (com-
pared to normal) to predict optotype acuity by con-
sidering the limitations imposed by retinal or
cortical magnification. Several other studies have
indicated that visual acuities measured with grating
targets are higher than those measured with opto-
types in amblyopic observers [108, 109] and pedi-
atric patients with significant ocular structural
anomalies [110]. A better prediction of optotype
acuity from the VEP grating acuity can be made by
taking this systematic overestimation into account.
Visual Evoked Potentials in Clinical Populations.
The VEP has been shown to be a sensitive indicator
Iv acuity losses in studies of infants with stra-
<. Measurements of monocular and binocular
f esotropic infants with the sweep VEP tech-
“que were significantly below the mean for age-
matched normal infants. [111]. The acuity reduc-
tions, although small, were significant. The infants
with esotropia and alternating fixation did not have
significant interocular differences. The VEP has
been used to monitor changes in the vision of the
amblyopic and fellow eye during occlusion therapy
[106, 112] and to record the postoperative acuity in
infants with congenital cataract. Cataracts were re-
moved from infants 7 hours to 41 days after birth
with final acuity outcomes of 20/30 in five patients
and 20/80 in three patients. No patients showed a
major discrepancy between VEP recording and eye
chart visual acuity measurements when the child
was old enough to perform both tasks [113].
PATIENTS WITH VISUAL IMPAIRMENT AND MULTIPLE
DISABILITIES. Several studies have reported that
the VEP measurcment of visual function can be
useful in the clinical management of nonverbal pa-
tients including patients with multiple disabilities
[114, 115], pediatric patients [107], and visually im-
paired children [116].
ALBINISM. The clinical findings in albinism in-
clude nystagmus, iris and fundus hypopigmenta-
tion, high refractive error. and strabismus. Not all
albinos display every one of these clinical features,
however. All forms of albinism involve foveal hy-
poplasia, reduced visual acuity, and a preponder-
ance of ipsilateral retinal fiber decussation that dis-
rupts retinotopic organization throughout the visual
pathway [117]. Nerve fibers from the temporal
retina erroneously decussate at the optic chiasm,
resulting in a decreased proportion of uncrossed
fibers and an abnormal visual pathway from the
lateral geniculate nucleus to the occipital cortex
[118]. The VEP correlate of misrouted optic path-
way projections is indispensable for the detection
and differential diagnosis of albinism [25]. With
full-field stimulation to the right eye, the largest
VEP amplitude is recorded over the left hemi-
sphere (contralateral asymmetry). This VEP test is
more difficult to interpret in children younger than
3 years old due to the immaturity of the pattern-
onset response, but the luminance flash condition
may still illustrate the asymmetry [119].
CorTicaL BLINDNESS. Cortical blindness is a loss
of vision secondary to damage to the geniculostriate
pathways and is characterized by reduced vision
and absence of optokinetic nystagmus with normal
ocular examination findings and intact pupillary
light responses [120]. Cortical blindness results
from hypoxic insults, meningitis, encephalitis,
metabolic disturbances, head trauma, or hydro-
cephalus. The recovery of vision is often protracted
and only partial; however, in some cases, recovery
may be complete and rapid. For this reason, cortical
visual impairment (CVI) is the more appropriate
term to describe this condition [121]. The most
common causes are generalized cerebral hypoxia at
the striate, parietal, and premotor regions and vas-
cular lesions of the striate cortex [122].
Several studies have evaluated flash VEP re-
sponses in CVI. The large variations in waveform
and amplitude of flash VEPs make them more diffi-
cult to interpret than pattern VEPs. The studies were
often limited to single case reports or small popu-
lation studies of patients with CVL Both normal
and abnormal flash VEP responses have been re-
ported, as well as VEP responses that improved
with time [122]. Frank and Torres [123] found no
difference in the amplitude of the flash VEPs
recorded from 30 children with CVI compared with
age-matched children with neurologic disorders but
no vision impairment. Conversely, Aldrich et al.
[124] reported abnormal flash and pattern VEPS in
15 of 19 patients with CVT; however, the VEP was
not correlated with the degree of visual loss. Other
studies have reported that the recovery of vision
was paralleled by normalization of the VEP re.
sponse. In one study, the VEP was used to deter-
mine visual prognosis following perinatal asphyxia
[125]. In this study, flash VEPs were recorded from
25 asphyxiated infants. Sixteen infants had normal
or only transient abnormalities of the VEP and with
follow-up all of these infants developed normal vi-
sion. The remaining nine infants who had abnormal
flash VEPs never developed normal vision. The
VEP findings in CVI remain controversial since
subjective measures of visual function cannot be de-
termined for comparison to the VEP measures and
the lesions that cause CVI are not often localized
exclusively to the striate cortex [122].
DELAYED VISUAL MATURATION. Infants with de-
layed visual maturation (DVM) appear to have CVI
with poor or no fixation up to 5-6 months of age.
At this time, visual responsiveness rapidly increases
to normal levels [126-128]. Pupillary response, oc-
ular examination, and neuroimaging studies are
normal even though the infant appears visually non-
responsive before 6 months. Although DVM may
be suspected, the diagnosis is confirmed only retro-
spectively after vision improvement. One study re-
ported normal ERGs but absent or immature flash
VEPs in infants suspected of having DVM. When
the infants showed vision improvement, the flash
VEPs were found to be normal; however, these VEP
responses were not compared with age-matched
norms [129]. A subsequent study compared the
VEP of 9 infants with DVM with age-matched
norms and reported normal pattern and flash VEPs
[127]. The authors concluded that intact pattern
VEPs indicate a good visual prognosis in visually
unresponsive infants. At present, DVM is incom-
pletely understood and the anatomic and molecular
developments that underlie the visual recovery have
not yet been identified [127].
Summary
There are several clinical applications of VEPs in
the pediatric population. The VEP is noninvasive
and can be used in a population that cannot com-
municate or cooperate for standard assessment of
visual function. Taylor and McCulloch [130] state
that the major application of VEPs in pediatric pa-
tients has been to quantify visual impairment by
using measures of visual acuity or contrast sensitiv-
ity or by quantifying flash or pattern VEP abnor-
malities. The flash VEP may help establish the
prognosis for visual recovery for specific pediatric
disorders. including perinatal asphyxia in full-term
neonates and acute-onset CVI. Tn some cases, flash
VEP may contribute to the differential diagnosis.
The VEP can help monitor patients who are at risk
for visual complications either from diseases (e.g,
hydrocephalus) or as a complication of therapeutic
intervention (e.g., neurosurgery). VEPs have be-
come an indispensable tool in pediatric ophthal-
mology and neurology and will probably play an
increasingly important role in the future sensitivity
of the VEP to subclinical damage [130].
A-SCAN AND B-SCAN ULTRASOUND
Background
A-scan and B-scan ultrasonography are diagnostic
procedures for the detection and differentiation of
ocular and orbital disorders [131, 132]. Thijssen
[133] provides an excellent review of the 50-year
history of the application of ultrasound in ophthal-
mology and optometry.
Definition
Ultrasound uses focused, short-wavelength, acou-
stic waves that are emitted via an echographic probe
as a wavefront advancing into the eye. The ocular
tissues reflect, refract, and scatter these sound
waves in a characteristic manner. The reflected
components, or echo, are received by the probe’s
transducer, amplified, filtered, and displayed on a
video screen or oscilloscope [134, 135].
Clinical Paradigm
The A- and B-scan ultrasound can be performed on
infants and young children. A topical ophthalmic
anesthetic is required and a drop of ophthalmic
methylcellulose is applied to the probe tip for B scan ultrasound. If the evaluation is conducted by a skilled examiner, sedation or anesthesia is rarely
needed, even in very young patients. Contraindica-
tions for A- or B-scan ultrasonography include re-
cent intraocular surgery, perforating injury, or
scleral lacerations. Complications of ultrasound in-
clude minor corneal abrasions or irritation resulting
from the echographic probe, anesthetic, or preserv-
ative in the methylcellulose solution [134, 135].
A-Scan Ultrasound
Biometry, or measurement of the eye and its struc-
tures, is achieved with the A-scan ultrasound. The
echographic probe is placed perpendicular and in
light contact with the anesthetized corneal apex.
The echoed sound waves transmitted along the op-
tical axis of the eye are plotted as a time-amplitude
recording and converted into an electrical distance
measure for display on an oscilloscope. Each spike
on the graph represents a specific ocular Lissue. The
measured time intervals between echo spikes can
be converted into a measurement (in millimeters)
of ocular structures along the optical axis, includ-
ing axial length. The axial scans in A-scan ultra-
sound are performed in both the horizontal and
vertical directions [134].
A-scan ultrasonography is most commonly used
to measure axial length of the eye for the determi-
nation of the proper dioptric power for intraocular
lens implantation. Other applications include quan-
tifying lesion size, internal structures, and intrinsic
vasculature [134]. Figure 7.12 shows the A-scans
for the right eye (upper panel) and left eye (lower
panel) in a 4-year-old male with the following re-
fractive error: right eye: —18.00 0.75 x 180; left
eye: +1.50 DS. These A-scans were conducted
through the child’s eyelid due to noncompliance;
therefore, the left-most spike normally correspond-
ing to the cornea and front and back lens surfaces
are not well differentiated. The right-most spike
corresponds to reflections from the retinal and
choroidal layers. The interocular difference in axial
length is approximately 6 mm. Since 1 mm of axial
length corresponds to a 3.00 D change in refractive
error, the anisometropia of 18.00 D is well pre-
dicted by the A-scan differential.
B-Scan Ultrasound
In B-scan ultrasound, the echogram is displayed as
a two-dimensional array in which the horizontal
axis represents tissue depth and the vertical axis
represents the scanned segment of the globe or orbit
(Figure 7.13). Several orientations of the echo-
graphic probe are used in B-scan ultrasound to
allow for transverse, axial, and longitudinal scans.
The continuous flow of images as the probe is
moved on the eye is evaluated by the examiner dur-
ing testing and is often videotaped.
B-scan ultrasound is recommended when infor-
mation is needed about the status of ocular struc-
tures posterior to corneal, lens, or vitreous opacities.
B-scan localizes lesions and yields information
about their configuration and gross reflectivity. A
complete B-scan examination is recommended with
opaque ocular media, vitreous hemorrhage, retinal
detachment, suspected ocular tumors, intraocular
foreign bodies, optic disc anomalies, proptosis, and
suspected extraocular muscle disease [135].
Clinical Applications of Ultrasound
in Pediatric Populations
A-scan ultrasound has been used to study the
growth of the eye from birth to puberty [136, 137]
and ocular biometry in premature infants [138,
139]. Ultrasound is most commonly used in the pe-
diatric population to make a decision about whether
surgical intervention for media opacities is appro-
priate. B-scan ultrasound is used to provide infor.
mation about the integrity of the vitreous, retina,
and posterior pole when a corneal opacily, cataract,
or retrolental mass obscures the clinician’s view of
the fundus. For example, ultrasound 1s recom-
mended before vitrectomy in ocular trauma to de-
tect retinal detachments or foreign bodies [140],
The differential diagnosis of a retrolental mass
includes retinoblastoma, persistent hyperplastic pri-
mary vitreous [141] retinopathy of prematurity,
dominant exudative vitreoretinopathy, congenital
cataracts, posterior uveitis, and retinal dysplasia,
Ultrasound and CT are useful in differentiating per-
sistent hyperplastic primary vitreous from retino-
blastoma [142], since the characteristic calcification
in the retinoblastoma lesion can be detected with
both procedures [143]. If the condition is bilateral
or there is a positive family history of retinoblas-
toma, an additional CT scan of the orbit and brain is
recommended.
The determination of refractive error in infantile
aphakia is aided by measures of axial length with
A-scan ultrasound [144]. An A-scan ultrasound is
recommended in developmental anterior segment
abnormalities if glaucoma is suspected.
Prenatal Diagnosis: Ultrasonic Imaging
of the Fetal Brain and Eye
An ultrasound image of structures shielded on all
sides by air or by bone more than a few millime-
ters thick cannot be made. For this reason, ultra-
sonographic studies of human anatomy are limited
to fetuses and infants of up to approximately 9
months of age, when the brain can be assessed
through a patent fontanel [145]. The principal clin-
ical concern in fetal cerebral imaging is the identi-
fication of major structural abnormalities such as
hydrocephalus and holoprosencephaly, the failure
of the forebrain to divide into hemispheres. These
anomalies are easily identified with ultrasound due
to the high contrast of spaces containing cere-
brospinal fluid [146]. Additional clinical concerns
in neonates are the identification of intracranial he-
morrhages, anoxic injury, and infection [145]. Cen-
tral nervous system lesions associated with
toxoplasmosis can be detected with ultrasound pre-
and postnatally [147].
Prenatal ultrasound has been used to monitor
the development of the fetal eye [148]. The norma-
tive data from a study by Achiron et al. that in-
cluded 450 fetuses from 12 to 37 weeks’ gestation
may be helpful in the prenatal diagnosis of sus-
pected congenital syndromes that are manifested in
ocular growth disturbances such as microphthal-
mos and anophthalmos. Prenatal ultrasonographic
diagnosis of retinal detachment has been demon-
strated in Walker-Warburg syndrome, a congenital
disorder associated with hydrocephalus and retinal
nonattachment [149].
NEUROIMAGING
Computed Tomography
Background
The introduction in 1972 of x-ray CT, also known
as computer-assisted tomography, opened up a new
era in imaging technology [150]. CT provides a
safer, more sensitive way to image soft tissue and
bone and is far superior to plain-film x-rays of the
orbits and skull. CT can provide images of fine
anatomic details of the orbit, oculomotor nerves and
muscles, and the central nervous system.
Definition
CT uses multiple x-ray projections and mathemati-
cally reconstructs data to create slice images [151].
A layer of any part of the body can be penetrated
by narrow x-ray beams that originate from a large
number of angles. These multiple transmissions are
ecorded. and an array of point-by-point relative ab-
sorption coefficients is calculated.
The absorption coefficient is the representation of the density or substance of the tissue. The resultant image is a gray-scale representation of the array of absorption coefficients. The earliest CT scanners required more than 5 minutes to produce images in an 80 x 80 matrix containing 29,000 bits of information. Scanning time now has been reduced to less than 1 second per image, with a 512 x 512 matrix and more than 1,000,000 data points [152].
The cerebral CT scan is a two-dimensional image of a brain slice that can be as thin as 1 mm
[150]. Discrimination of gray from white matter of the brain is achieved due to the high spatial and density resolution of CT scanners. CT can identify structural changes, including cerebral spinal fluid cavities, edema, demyelination, and extravasated
blood, and can detect the small amounts of cal-
cium that are present in tumoral, granulomatous,
and arteriosclerotic calcifications and in Sturge-
Weber syndrome.
Application in Pediatric Populations
Because of its excellent imaging of the orbit and
bony anatomy, CT is the preferred imaging tech-
nique for diagnosis of complex ocular trauma 1153]
and suspected nonaccidental injury [154]. Calcium
deposits within intracranial and intraorbital masses,
particularly retinoblastoma, are readily detected
with CT (Figure 7.14) [142, 143, 150, 155].
Risks to Pediatric Population
The brief scanning time has significantly reduced
the occurrence of CT image degradation secondary
to movement in uncooperative patients or children;
however, some patients may need to be sedated for
the procedure [156]. Intravenous iodine-based con-
trast enhancement media has been used to enhance
the differential diagnostic capability of the CT
scan, particularly in evaluation of the optic nerve
and its associated sheaths. The side effects of the
iodine contrast enhancement media and radiation
exposure should be considered with CT use in in-
fants and children.
Magnetic Resonance Imaging
Background
CT imaging and basic x-rays use propagating rays
to interact with biological tissue. Imaging with x-
rays is obtained from the scattered or absorbed rays.
MRI . however, uses nuclear magnetic resonance to
produce tomographic images [157]. MRI has been
available for use since the early 1980s and is now
accepted as the most sensitive and among the most
specific diagnostic imaging tools available for
neuro-ophthalmic applications [1501.
Definition
MRI encodes image information by using wave-
lengths of several meters and relying on spatial and
time-dependent magnetic fields to excite and mod-
ulate the resonance of the hydrogen nuclei in bio-
logical tissue [157, 158]. A magnetic resonance
instrument uses a strong magnetic field to partially
align the hydrogen nuclei, which otherwise would
have a naturally random distribution of unpaired
nuclear spins. In MRI, specific radiofrequency
pulses are used to interact with the hydrogen nu-
cleus in the tissue and transfer energy. The ab-
sorbed energy changes the nuclear spin and
temporarily distorts the alignment of the hydrogen
nuclei. When the radiofrequency pulse is turned
off, the nuclei gradually realign with the magnetic
field and emit varying radiofrequencies. T1 (longi-
tudinal) and T2 (transverse) relaxation times are
the basic parameters used to describe the nuclear –
realignment and resulting signal pattern. Generally,
solids have very short relaxation times, and the re-
laxation times for tissues increase with increasing
fluidity. For example, relaxation times for tumors
tend to be longer than those of the host tissue. The
differences in relaxation times of tissues enhance
the contrast of the magnetic resonance image (Fig-
ure 7.15). The TL or T2 characteristics of various
biological tissues can be emphasized by changing
the radio frequency pulse repetition time and the
signal sampling, or echo, time. T1-weighted im-
ages have a short repetition and echo time and
therefore result in a high signal-to-noise ratio and
relatively short imaging time. T1-weighted image
sequences have optimal spatial detail, whereas T2-
weighted sequences have optimal tissue contrast.
Fluid is generally dark on T1-weighted images
and bright on T2-weighted images. Therefore, T2-
weighted images are particularly sensitive to
2dema. MRI does not image bone well. Fat is seen
as a high-intensity (bright) signal on T1-weighted
images and may obscure the signal from an adja-
cent tissue lesion or tumor (e.g., in retrobulbar
pathology). Short T1 inversion recovery sequences
can be used to decrease the signal from fat. Con-
trast enhancement can be obtained with the use of
intravenous magnetic contrast agents. The primary
effect is a T1 shortening that increases the signal
on T1-weighted images. The intravenous contrast
agent used in MRI is safer and better tolerated than
the iodine contrast agents used in CT.
MRI protocols can be varied to optimize diag-
nostic yield. The most important parameters in-
clude imaging plane slice thickness (typically 3-5
mm) and pulse sequence selection. Imaging
planes include coronal, transaxial, and sagittal
sections. Image contrast and scan protocols can
be adjusted to emphasize characteristics of sug-
gested disease states.
Comparison of Magnetic Resonance Imaging
and Computed Tomography
There are several advantages of MRI over CT in or-
bital imaging [157]. MRI has superior imaging per-
formance and the ability to image soft tissue
noninvasively. No dose of x-rays is administered;
therefore, no orbital tissues are exposed to ionizing
radiation. MRI is diagnostic without the use of in-
travenous iodine-containing contrast agents. When
contrast enhancement is needed for diagnostic
yield, the gadolinium-containing contrast agent is
safe and well tolerated. Also, metallic dental hard-
ware does not degrade the MRI quality. Finally,
MRI achieves better imaging than CT of brain stem
structures and posterior fossa disease [157]. CT is
of limited value in evaluating the posterior fossa
due to the beam-hardening artifact of the surround-
ing bone [150].
There are, however, several disadvantages of
MRI compared with CT in orbital imaging. MRI
1) has poor specificity (e.g., tumors, infections, de-
myelinating foci, and edematous areas all produce
similar signals); (2) has poor detection of calcifi-
cation within a lesion or tumor, particularly in
retinoblastoma; (3) has poor imaging of bony
anatomy; (4) has poor detection of bony remodeling
or early invasion from an adjacent lesion:
(5) provides substandard images resulting from mo-
tion artifact; and (6) is time consuming due to the
repetition of scan sequences to compensate for mo-
tion artifact [157].
Positron Emission Tomography
Background and Definition
PET allows in vivo quantitative measurements of re-
gional physiologic and biochemical processes. In the
PET procedure, a tracer is labeled with a positron-
emitting radionuclide. The tracer is given intra-
venously to the subject and the three-dimensional
positron activity in the brain is assessed with a
positron emission tomograph [159].
PET studies use a variety of tracers to measure
local cerebral blood flow, cerebral blood volume,
oxygen consumption, glucose consumption, pro-
tein synthesis, neurotransmitter receptor proper-
ties, tissue proliferation, tissue pH and drug
distribution [159]. Research with PET continues
on the effects of visual stimulation and cognitive
tasks on cerebral glucose metabolism, pathophys-
iologic events occurring after a stroke, aging and
dementia, and grading the malignancy of brain tu-
mors. These studies have significantly increased
understanding of the metabolic and physiologic
processes of the human brain in normal and patho-
logic conditions [160].
In pediatric patients, PET has been particularly
informative in epilepsy [161, 162]. The PET of
local cerebral glucose utilization is highly sensitive
in detecting epileptogenic regions [163]. Expand-
ing PET technology provides a new approach that
holds great promise in the diagnosis and manage-
ment of brain disorders in children [163].
Clinical Application of Neuroimaging
in the Pediatric Population
Differential Diagnosis for Retinoblastoma
MRI does not detect the calcification in retinoblas-
toma as well as CT docs but may provide informa-
tion about whether spread along the optic nerve has
occurred since the artifact from surrounding bone is
reduced [164]. Choroidal melanomas arc better de-
tected with MRI than CT because the melanin can
cause a paramagnetic effect that enhances the bright
signal within the tumor [150]. Magnetic resonance
images display certain characteristics in other disor-
ders that aid in the differential diagnosis for
retinoblastoma, including (1) Coats’ disease, whose
clinical signs include retinal telangiectasis with a re-
sultant lipoproteinaceous subretinal exudate; (2) per-
sistent hyperplastic primary vitreous, particularly
when accompanied by hemorrhage; and (3) poste-
rior uveitis associated with toxocariasis [158].
Optic Nerve Hypoplasia
CT or MRI should be considered in any case of
optic nerve hypoplasia (unilateral or bilateral) as-
sociated with a history of neonatal hypoglycemia,
seizures, failure to thrive, delayed development, or
other neurologic signs. In one study [44] of chil-
dren with bilateral optic nerve hypoplasia, poor vi-
sion, and nystagmus, 39% also had abnormalities
shown with neuroimaging. Unilateral optic nerve
hypoplasia was associated with intracranial patho-
logic factors in fewer than 10% of the cases [44].
All infants with optic nerve hypoplasia should also
be referred to a pediatric endocrinologist for evalu-
ation. Bilateral optic nerve hypoplasia accompa-
nied by structural abnormalities along the midline
of the central nervous system is referred to as
septo-optic dysplasia or DeMorsier’s syndrome.
Endocrine dysfunction may accompany this disor-
der, which will result in short stature of the child
unless treated [165, 166].
Optic Nerve Atrophy
Optic nerve atrophy is a clinical sign, not a diagno-
sis. Children presenting at any age with nystagmus
and optic atrophy should have a CI or MRI scan to
rule out associated conditions. Optic atrophy may
be associated with hydrocephalus or anterior visual
pathway compression from a suprasellar tumor, in-
cluding craniopharyngiomas, gliomas, meningi-
omas, pituitary tumors, metastatic tumors, and
arachnoid cysts.
Hydrocephalus is an increased amount of cere-
brospinal fluid in the cerebral ventricles resulting
from impaired cerebrospinal fluid circulation, reab-
sorption, or hypersecretion. Hydrocephalus is a
common cause of optic atrophy [167], cortical vi-
sual impairment, or both in children. Rapid head
growth is the most notable clinical finding in hy-
drocephalus with an onset before 2 years of age.
After 2 years of age, various neuro-ophthalmologic
abnormalities may present in ocular motility, pupil-
lary responses, optic nerve appearance, vision func-
tion, and visual field. CT and MRI are both useful
in the diagnosis of hydrocephalus.
Craniopharyngioma is the most common non-
glial intracranial tumor of childhood [168]. Chil-
dren with craniopharyngioma often have only
nonspecific symptoms and psychogenic vision loss
is often suspected [169]. The vision loss is progres-
sive due to compression of the optic nerves, tracts,
or chiasm and resulting optic atrophy. By the time
the diagnosis of craniopharyngioma is made, the
optic atrophy may be profound. Calcification within
the tumor is easily seen on CT scanning or plain
skull films.
Optic nerve glioma occurs almost exclusively
before 20 years of age and generally occurs in
children. The tumor is often associated with neuro-
fibromatosis but may not be detected by ophthal-
moscopic examination [170]. MRI is important in
all children with neurofibromatosis. With optic
gliomas, the most common MRI appearance is
fusiform enlargement of the optic nerve (Figure
7.16). Posterior extension through the optic canal
with involvement of the chiasm or optic tract can be
seen on MRI (see Figure 7.16) [158]. In general,
posterior fossa abnormalities, suspected demyeli-
nating processes, and parachiasmal structures are
better imaged by MRI [158]. CT is of limited value
in evaluating the posterior fossa due to beam-
hardening artifact of the surrounding bone [150].
Brain Tumors
Primary brain tumors are the most common solid
neoplasms in children but differ considerably from
the adult form in incidence, location, morphology,
histology, and natural history [171]. The distribu-
tion of brain tumors differs in infants and older
children. Vomiting is the most common presenting
symptom in all pediatric age groups, but older chil-
dren are more likely to show localized neurologic
signs (e.g., cranial nerve palsy, hemiparesis, clum-
siness, ataxia), recurrent headaches, vomiting, and
visual complaints, including acute onset of es-
otropia [171]. The clinician should closely look for
papilledema or nystagmus in children with an acute
onset of comitant esotropia that does not appear to
be an accommodative esotropia [172]. Infants with
brain tumors tend to not show focal neurologic
deficits due to the immaturity of the brain. but they
may have hydrocephalus and papilledema due to
increased intracranial pressure [173]. A signifi-
cant portion of the clinical signs and symptoms in
children with brain tumors involves the visual
system due to tumor invasion of visual system
structures or the mass effects of the tumor (e.g.
associated hydrocephalus or secondary compres-
sion). MRT and CT and peuro-ophthalmologic
evaluation is an important component of follow-
up in these children [171].
Cortical Visual Impairment
Neuroimaging is valuable in revealing the structural
abnormalities of the central nervous system and is
critical to the diagnosis of CV1 in infants and chil-
dren; however, the findings with neuroimaging can-
not predict visual function or visual potential. In a
study of 30 infants with CVI, all but two infants had
abnormalities in the neuroimaging studies. None of
the abnormalities demonstrated with neuroimaging
correlated with the degree of visual recovery in
these infants (except for abnormalities of the optic
radiations) [174]. The case of a 20-month-old child
who had absent occipital and parietal lobes con-
firmed by neuroimaging but could use vision to
reach for small objects has been reported [175].
Areas that appear nonfunctioning on MRI and CT
may have some residual function. Positron emission
tomography (PET) and single photon emission
computerized tomography (SPECT) studies may be
more sensitive and better able to make inferences
about visual function, but these are areas of future
research [176, 177].
Summary
- Neuroimaging studies are indicated in the workup
‘of a possibly blind child when the diagnosis is still
uncertain after a careful history, clinical examina-
tion, and electrophysiologic testing have been com-
“pleted [42]. Brodsky et al. [171] recommend
neuroimaging studies in (1) infants with congenital
nystagmus and optic nerve hypoplasia (to look for
CNS anomalies), (2) infants or children with con-
genital nystagmus and optic atrophy (to rule out hy-
drocephalus or a congenital suprasellar tumor
[craniopharyngioma or chiasmal gliomal), and
(3) uncertain diagnoses of congenital nystagmus
when the possibility of spasmus nutans exists (to
rule out chiasmal gliomas or other suprasellar tu-
mors). Neuroimaging also provides useful informa-
tion for prognostic and genetic counseling.
Nadel [178] emphasizes that neuroimaging in
children is different from that in adults. Important
technical considerations include immobilization of
the child during imaging, appropriate dosing of ra-
diopharmaceuticals, and appropriate instrumenta-
tion. However, new advances in instrumentation,
such as multiple detector imaging, the possibility of
clinical PET imaging in children, and new radio-
pharmaceuticals, will further enhance the utility of
pediatric imaging.
PHOTOREFRACTION
Background
In the United States, vision disorders are the fourth
most common disability and the most prevalent dis-
abling condition in children [179]. For preschool
children, these vision disorders include (in decreas-
ing order of prevalence) significant hyperopia,
astigmatism, color vision defects, myopia, strabis-
mus, amblyopia, anisometropia, and ocular disease
[180, 181]. Although these vision disorders can be
readily detected with a vision examination, it has
been reported that only 14% of children younger
than 6 years of age have received a comprehensive
vision examination [179].
Photorefraction is a relatively new technique de-
signed to screen for vision disorders, particularly
amblyogenic factors, in infants and young children
who are unable to complete the subjective vision
tests required in traditional vision screening proto-
cols. Photorefraction uses a camera-based system
with a specially placed light source to provide pho-
tographic images that can indicate the presence of
significant refractive errors, strabismus, media
opacities, and ptosis.
Types
There are two principal photorefraction systems:
on-axis (co-axial) and off-axis. In the on-axis sys-
tem, the camera lens and flash are aligned with
the subject’s visual axis. Three photographs are
required to assess refractive error: a focused
image for calibration of the pupil and two addi-
tional photos with the camera defocused in equal
diopters in front of and behind the pupil plane
(Figure 7.17). The sign and magnitude of the
spherical and cylindrical errors can be assessed
from comparison of the size of the blur circles in
the two defocused images. The major axis of the
blur ellipse gives the axis of astigmatism for a
negative cylinder correction [182].
On-axis photorefraction has several limitations.
The apparent size of the blur circle is influenced by
fundus pigmentation, pupil size, and the contrast
between the subjects face and the blur circle. Scor-
ing the defocused images is difficult and the sub-
ject’s direction of gaze, strabismus, and ocular
media opacities cannot be detected. Although ac-
tive fixation on the photographer is encouraged
while taking the photos, accommodation to the
plane of the camera is not assured; therefore. the
presence of significant hyperopia may be masked.
The range of refractive errors the camera can de-
tect is from 4.0 D of hyperopia to 4.0 D of myopia
[183]. Using an on-axis photorefractor. Hamer et
al. [184] reported a sensitivity of 85% and a speci-
ficity of 53% in screening for significant refractive
error in an infant population.
Off-axis photorefraction systems provide more
easily interpreted, focused photographs than on-axis
systems. In early versions of the off-axis system, &
35-mm camera was equipped with a telephoto lens
and the strobe flash was placed directly below the
optical axis of the camera lens [185]. These systems
only refracted in one orientation and were only sen-
sitive to refractive errors larger than 3 D, depend-
ing on the camera aperture and pupil diameter
[180]. The next version of off-axis photorefractors
substituted a catadioptric (or mirror) telephoto lens
and detected refractive errors as small as 0.75 D
[186, 187] with a sensitivity range up to 11.0 D for
myopia and 7.5 D for hyperopia. Selection of the
flash eccentricity determines the range of refractive
errors to which the camera is sensitive.
Off-axis photorefraction systems provide a fo-
cused image of the pupil and corneal and retinal re-
flex and therefore overcome mary of the limitations
of the on-axis systems. A focused image of the
corneal reflex (Hirschberg) can be measured to as-
sess fixation and eye alignment. In esotropia the
corneal reflex in the deviating eye is displaced tem-
porally and in exotropia the comeal reflex in the de-
viated eye 18 displaced nasally (Figure 7.18). The
presence of a Bruckner reflex, a unilateral brighten-
ing of the red reflex, may indicate either strabismus
or significant anisometropia. The clarity of the red
fundus reflex can be evaluated to indicate the pres-
ence of media opacities. A corneal opacity appears
as a bluish haze obscuring the view of the iris and
pupil (Plate 7.2A). A cataract or Jens opacity ap-
pears black or blue within the confines of the pupil-
lary margin (Plate 72B). Ametropia is indicated by
the presence of a light or yellow crescent within the
red reflex. The size of the bright crescent indicates
the magnitude of the refractive error and the loca-
tion of the crescent indicates the direction of the
ametropia. The bright crescents of hyperopia are lo-
cated on the opposite side of the camera flash while,
myopic crescents are located on the same side 2
the camera flash. In Plate 7.3, the right eye is my-
opic along the vertical meridian of the eye and the
bright photorefraction crescent is located at the bot-
tom of the pupil margin. The left eye is hyperopic
along the vertical meridian and the corresponding
bright crescent is located at the upper pupillary
margin. Photographs of the child’s eyes are taken
with the flash oriented both vertically and horizon-
tally to “refract” the 90 degree and 180 degree
meridians, respectively. Astigmatism is detected by
comparison of the orthogonal crescent with the
camera flash oriented vertically and horizontally. In
this infant, the hyperopia crescent along the verti-
cal meridian is larger than along the horizontal
meridian in each eye, indicating against-the-rule
astigmatism (Plates 7.44 and 7.4B):
Current off-axis systems utilize cameras with in-
stant film developing yielding color [188] or black
and white images [189, 190]. The advantage of in-
stant camera-based systems is the immediate feed-
back and compactness and portability of the system.
Clinical Applications
The sensitivity and specificity of off-axis photore-
fraction as a screening tool has been evaluated by
comparison fo a clinical vision examination in a few
studies [190, 191-195]. These studies included
subjects from 3 months to 23 years. The sensitivity
4nd specificity values ranged from 0.41-0.95 10
0.39-0.96, respectively. The lack of standardization
of pass versus fail criteria complicates comparison
across photorefraction studies. Those studies re
porting the highest sensitivity and specificities ¢X-
cluded children for whom analyzable photographs
were not obtained [190, 193, 194). If failure to ob-
tain a readable photograph was counted as an auto-
matic referral, the specificity value in particular
would decrease.
Although there is much interest in photorefrac-
tion as a screening tool for infants and children, a
recent photorefraction study reported that only
38% of the photographs were consistently catego-
rized as pass or fail when scored by a group of
health professionals [196]. Further research in pho-
torefraction is needed before routine administration
of photoscreening in those groups at greatest risk
for vision disorders.