Visual Evoked Potentials in Infants and Children

Visual evoked potentials (VEPs) are massed electrical signals generated by occipital cortical areas 17, 18, and 19 in response to visual stimulation. VEPs differ from the electroencephalogram (EEG) in that the EEG represents ongoing activity of wide areas of the cortex, whereas the VEP is a specific occipital lobe response triggered by a visual stimulus, primarily dependent on the functional integrity of vision in the central visual field. Thus, VEPs can be used to assess the integrity or maturational state of the visual pathway in infants and preverbal children.

The basic methodology for recording VEPs is straightforward and is described in the Visual Evoked Potentials Standard developed by the International Society for Clinical Electrophysiology in Vision (ISCEV). An active electrode is placed on the scalp over the visual cortex (Oz), along with a reference electrode at Fz and a ground electrode at Cz or on the ear lobe. Signals are led from the electrodes through a preamplifier, which is located near the infant to boost signal amplitude prior to further contamination by outside noise sources. The preamplifier also typically acts as a bandpass filter. From the preamplifier the signal is led into a computer capable of averaging VEPs. The main purpose of filtering and signal averaging is to improve the signal-to-noise ratio. The VEP is contaminated by EEG, as well as by outside noise sources. However, since the VEP in response to a given pattern is fairly constant in amplitude and latency (or phase) while the EEG occurs randomly with respect to the visual stimulus, stimulus averaging will decrease the unwanted contamination by EEG and other noise sources in proportion to the square root of the number of VEPs averaged.

Most commercial equipment for recording the VEP is not designed specifically for use with infants and young children and therefore requires several modifications for use in this patient population. Since it is important that the infant be visually alert and attentive to patterned stimuli, it is helpful to have some small toys that can be dangled in front of the video display when collecting pattern VEPs. The majority of the VEP response is generated by the cortical projection of the macular area (central 6 to 8 degrees of the visual field ); therefore, the toys must be very small or open in the center so as not to block the stimulus from reaching the macula. An observer must watch the infant and signal to the averaging equipment when the infant is looking at the pattern and when the infant is looking elsewhere. Both hand-held buttons and foot pedals can be constructed for this purpose. Sudden head and body movements, including yawns, crying, or vigorous sucking, can produce large-amplitude broad-band electrical artifacts. Accordingly, the averaging or analysis equipment must contain hardware or software to reject these artifacts because averaging alone is not sufficient to eliminate their contribution to the recorded response. Alternatively, trials containing artifacts can be deleted manually from the signal average.

Minimum standards for reporting VEPs have been published. At least two averages should be obtained, to demonstrate reproducibility of waveforms. Traces of VEP recordings should have a clear indication of polarity, time in milliseconds, and amplitude in microvolts. The ISCEV recommends that VEP traces be presented as positive upwards, although in many clinical neurophysiology laboratories the traces are presented with a downward deflection representing positivity at the active electrode. Normative values and normal tolerance limits for amplitude and latency should be reported along with the results. Since VEP latency, amplitude, and waveform change with age, comparison to same-age normative values is useful in clinical practice.

VEPs are elicited by a temporal change in visual stimulation. Three types of visual stimuli commonly are used to elicit VEPs: luminance (light) flashes, pattern onset/offset, and pattern contrast reversal. Each of these stimuli may be presented as single events or in a repetitive manner. When the light flashes, pattern onset/offset, or contrast reversal occurs infrequently (i.e., at 1 Hz or less), the entire VEP waveform can be seen; this is called a transient response. Peak-to-peak amplitudes and latencies of several major peaks can be measured ( Fig. 23-1 ). When the light flashes or pattern reversals are repeated frequently at regular intervals (at rates of 6 Hz or higher), a simpler periodic waveform is seen; this is called a steady-state response. Response amplitude and phase are measured.

Transient luminance (flash) VEPs

Transient luminance VEPs are obtained in response to a strobe light or flashing light-emitting diode (LED) display or goggles. The flash should subtend at least 20 degrees and should have a stimulus strength of 2.7 to 3.3 cd/m . The flash rate should be 1.0 Hz ± 10 percent. The transient luminance flash VEP is a complex waveform with multiple negative and positive changes in voltage. Various approaches have been taken to naming the peaks and troughs in this response, which has led to some difficulty in interpreting differences among studies. Recently, recommendations for standardized reporting of VEP data have been made by the ISCEV. As shown in Figure 23-1 , peaks are designated as positive and negative in numerical sequence (P1, P2, P3 and N1, N2, N3). The most commonly reported amplitude is the N2-P2 peak-to-peak amplitude.

Using both source localization techniques in humans and intracortical recording in primates, several recent studies suggest that the transient luminance flash VEP primarily reflects the activity of striate and extrastriate cortex. In addition, some wavelets in the VEP appear to be subcortical in origin. These subcortical wavelets are not major components of the flash VEPs obtained from patients with healthy striate and extrastriate cortical areas. However, flash VEPs have been recorded in newborns who lack functional striate and extrastriate cortex. These VEPs may have reflected subcortical function that was more apparent in the VEP when the cortical components of the response were missing. Thus, the presence of a flash VEP response cannot be taken as unequivocal evidence of cortical function.

Pattern onset/offset VEPs

Pattern onset/offset VEPs are obtained in response to a pattern abruptly exchanged with a diffuse background. The pattern should subtend at least 15 degrees and the pattern and diffuse fields should be well matched in mean luminance. Pattern onset duration should be 200 msec separated by 400 msec of diffuse background. In most cases, a minimum of two pattern-element sizes should be included, 1 degree and 15 minutes, but normal newborns may not respond to the 15-minute pattern-element size. Pattern VEPs primarily reflect activity of the striate and extrastriate cortex. The pattern onset/offset VEP contains three peaks in adults: a positive peak followed by a negative peak and a second positive peak. As shown in Figure 23-1 , ISCEV proposes that these components be termed C1, C2, and C3. Amplitude is measured from the preceding negative peak. Pattern onset/offset VEPs are useful for evaluating infants and children with nystagmus, as they are less sensitive than pattern-reversal VEPs to confounding by involuntary eye movements and poor fixation stability.

Pattern-reversal VEPs

Pattern-reversal VEPs are elicited by abrupt contrast reversal. The pattern should subtend at least 15 degrees and, typically, a minimum of two pattern-element sizes are included, 1 degree and 15 minutes. Pattern reversal at 1 Hz (2 reversals per sec) yields transient VEPs, whereas higher reversal rates (e.g., 6 to 10 Hz) yield steady-state VEPs. In adults, the transient pattern-reversal VEP typically contains a small negative peak followed by a large positive peak and a second negative peak. As shown in Figure 23-1 , ISCEV proposes that these components be termed N75, P100, and N135 to indicate their polarity and their approximate latency (in milliseconds) in normal adults. The most commonly reported amplitude is the N75-P100 peak-to-peak amplitude. The steady-state pattern-reversal VEP has a relatively simple, almost sinusoidal waveform; amplitude and phase of the response typically are reported. Since the mean luminance of the pattern remains constant throughout the contrast reversals, pattern-reversal VEPs reflect pattern sensitivity rather than light sensitivity.

Using intracerebral recording in awake humans, Ducati and co-workers found that P100 appears to be generated by the pyramidal cells in layer IV of area 17. However, imaging studies in humans point to the source of the early phase of the P100 peak as being in dorsal extrastriate cortex of the middle occipital gyrus, whereas the late phase of P100 appears to be generated by the ventral extrastriate cortex of the fusiform gyrus. There have been no reports of pattern-reversal VEPs being recorded in newborns who lack functional striate and extrastriate cortex. Thus, the presence of a pattern-reversal VEP response may be a good indicator of the integrity of cortical function.

Normal maturation of VEPs

Visual responses have been documented in preterm infants as young as 24 weeks gestational age (GA), but these responses are rudimentary. Infants tested at 22 to 23 weeks GA had very poor or absent VEPs. Considerable visual development occurs during the third trimester of gestation and the first post-term year. While, on funduscopic examination, the fovea appears mature soon after term birth, detailed anatomic studies have shown that neither the migration of cone photoreceptors toward the foveal pit nor the movement of ganglion cells away from the foveal pit is complete during the first months of life. Moreover, the fine anatomic structure of the foveal cone photoreceptors, which subserve fine detail vision, is not mature until at least 4 years of age. Myelination of the optic nerve and tract is incomplete at term birth and continues to increase for 2 years postnatally. Although the number of cells in the primary visual cortex appears to be complete at birth, considerable increases in cell size, synaptic structure, and dendritic density take place during the first 6 to 8 months of life. One approach to monitoring the progress of anatomic maturation has been to evaluate developmental changes in the VEPs of healthy alert infants. These data also provide a baseline for assessment of the degree of visual impairment in pediatric patients.

Flash VEP amplitudes may be influenced by arousal state, but latency is less affected. The mean latency of N3 across various arousal states (awake, drowsy, active sleep, and quiet sleep) was within 15 msec in preterm infants at 30 to 37 weeks GA, but sleep significantly reduced N3 amplitude. Similarly, the mean latencies of P1 for the alert state and the sleep state were within 15 msec at 40 weeks GA. Pattern VEPs are affected by arousal state; good-quality pattern VEP recordings require an alert, attentive infant whose eyes are focused on the pattern.

Amplitude and Latency of Transient Luminance (Flash) VEPs

In preterm infants less than 30 weeks GA, a single long-latency (about 300 msec) negative peak is the most prominent component of the flash VEP ( Fig. 23-2 ). This peak has been identified by some authors as N1, since it is the earliest negative peak observed in young infants. However, other authors have suggested that this peak corresponds to the N3 component of the mature waveform and that the N1 and N2 peaks are “missing.” Some difficulty in establishing correspondence between peaks in infant and adult waveforms occurs because the early negative peak in young infants is often bifid, with two subpeaks that may or may not be symmetric in amplitude.

In Figure 23-2 , flash VEP latency from various studies is plotted to show the maturation of the early negative peak, which appears to correspond to the adult’s N3 (adult latency, 150 msec). Tsuneishi and colleagues assessed changes in the early negative peak longitudinally and found that latency decreased at a rate of 4.6 msec per week between 30 and 40 weeks GA. Pike and colleagues report a similar rate of 5.5 msec per week between 28 and 42 weeks GA. There is some evidence that the change in latency does not occur smoothly but, instead, occurs in “spurts” of at least 6 msec, possibly reflecting the myelination process in the visual pathway.

At term (40 weeks GA), earlier negative components are present in the waveform which shorten in latency with age; these appear to correspond to N2 (adult latency, 75 msec) and N1 (adult latency, 40 msec). As seen in Figure 23-2 , Pike and associates report that the earliest negative components can be obtained from infants as young as 34 weeks GA, while all other studies indicate that these components arise at term (40 weeks GA). Yet, even in that study, only at term age did all infants simultaneously have both an early N0 and N1. One possible explanation for this finding is that Pike and colleagues included only infants who later had normal neurologic examinations over the first 2 years of life.

The youngest infants reported to show an early positive peak with a latency of approximately 200 msec were between 30 and 35 weeks GA. This positive peak likely corresponds to the P2 peak in the adult waveform (latency, 100 msec). P2 is consistently present in all normal neonates from 37 weeks GA on, and is present in 90 percent of neurologically normal infants by 35 to 36 weeks GA. When present, the latency of the positive component decreases from about 220 msec at 34 weeks GA to 150 to 190 msec at term and to 100 msec by 8 to12 weeks post-term age (48 to 52 weeks GA) ( Fig. 23-3 ). The P2 peak is clearly identifiable in the VEPs of healthy infants by 6 weeks post-term (46 weeks GA), and its amplitude exceeds that of N3 by 8 weeks post-term (48 weeks GA).

A rapid increase in the amplitudes of most of the flash VEP peaks occurs during early childhood, with the largest amplitudes present at about 6 years of age. Amplitudes of the various peaks reach adult levels by about 16 years of age. The maturational changes in amplitude, however, are age trends that are present despite very large individual differences in amplitude within any given age group. This variability limits the utility of flash VEP amplitude in detecting developmental visual abnormalities. Clinical evaluation has focused largely on the latency of the negative component and appearance of the first positive peak (which corresponds to P2 in adults). Prolonged latency of the initial negative peak (more than 370 msec) and the absence of the positive peak at 40 weeks GA or older have been linked to poor outcomes in preterm infants.

Steady-state flash VEPs have been used to investigate the maturation of temporal resolution during infancy. The data suggest three phases in maturation. During the first month of life (40 to 44 weeks GA) temporal resolution shows little maturation, improving from 15 Hz to 19 Hz. During the next 5 months, temporal resolution matures rapidly to 45 Hz, and then undergoes further maturation, slowly approaching the adult level (55 Hz) by 9 months of age. In addition, the optimal stimulus frequency—that is, the frequency that elicited the largest amplitude response—increases with age during the first 2 years of life.

Amplitude and Latency of Pattern VEPs

The presence, amplitude, and latency of pattern VEPs change with maturation and age. With large pattern-element sizes, transient pattern-reversal VEPs can be recorded from preterm infants as early as 30 weeks GA. At this age the pattern-reversal VEP to a large-element checkerboard (each check subtending about 2 degrees of visual angle) has a simple waveform consisting of a single positive peak with a latency of approximately 330 msec. As shown in Figure 23-4 , latency for large-element patterns decreases to about 250 msec by 40 weeks GA (term), 150 msec by 10 weeks post-term (50 weeks GA), and 110 msec by 25 weeks post-term (65 weeks GA). For small pattern-element sizes (less than 15 minutes), transient pattern-reversal VEPs may not be recordable up to 50 to 54 weeks GA.

Beyond 4 weeks post-term (44 weeks GA), the VEP waveform changes from a simple one to a more complex waveform with multiple peaks, and peak latency grows progressively shorter ( Fig. 23-5 ). Latency is dependent on pattern-element size, with latency to large-element patterns decreasing rapidly over the first months of life and latency to small-element patterns decreasing more gradually (see Figs. 23-4 and 23-5 ). Similar changes in latency have been reported for transient pattern onset/offset VEPs.

Steady-state pattern VEPs can be recorded from preterm infants as early as 35 weeks GA when large pattern-element sizes and relatively slow pattern alternation rates are used. As infants mature, responses can be recorded to progressively smaller pattern-element sizes and progressively faster pattern-alternation rates.

Visual Acuity Development

The pattern VEP has been used to estimate visual acuity in infants for the past 35 years. Just as a standard eye chart for measuring visual acuity contains letters ranging from very large to quite small, a pattern VEP acuity test includes a group of checkerboard or striped grating patterns with elements that range from coarse to fine. In both cases, the goal is to determine the finest pattern that the patient’s visual system can resolve. When the VEP is used, visual acuity is estimated most often by examining the relationship between amplitude and pattern-element size. Ideally, many pattern-element sizes would be evaluated to pinpoint the exact size at which a response can no longer be recorded. With the limited attention span of infants and young children, typically a limited set of four to six patterns is presented. Particularly in clinical settings where there is often little prior information on which to base a choice of the optimal pattern-element sizes, logarithmic spacing between pattern-element sizes in the test series is used (so that, for example, each pattern size is one-half that of the previous pattern in the series). Logarithmic spacing of a wide range of pattern-element sizes despite a small number of steps maximizes the possibility that acuity can be estimated from the data obtained. Linear regression of amplitude on pattern-element size is used to extrapolate the pattern-element size that corresponds to 0.0-μV amplitude in order to provide the visual acuity estimate.

Many pediatric laboratories and clinics have adopted the sweep VEP for acuity testing in recent years. Sweep VEP protocols present pattern-element sizes to the infant in rapid succession during a 10-second sweep. Using Fourier analytic techniques to extract the VEP responses to each of the brief stimuli (specifically, the amplitude and phase of the harmonics of the stimulation rate), sufficient information can be obtained from the VEP records to estimate visual acuity from only a few brief test trials. This technique has three significant advantages. First, test time is reduced. Second, the infant’s behavioral state changes little during the brief recording session. Third, since many more pattern-element sizes are included in the test protocol, linear spacing in the series can be used and a more accurate estimate of acuity can be obtained ( Fig. 23-6 ).

Two classic studies of the maturation of visual acuity during infancy using transient pattern VEPs were conducted during the mid 1970s by Sokol and by Marg and associates. From an initial visual acuity of 1.30 logMAR (20/400 Snellen equivalent) at 4 weeks of age, visual acuity rapidly matured to nearly adult levels (0.2 to 0.0 logMAR; 20/30 to 20/20 Snellen equivalent) by 6 to 7 months of age. MAR represents the minimum angle of resolution. Although some controversy remains, it appears that visual acuity matures according to age corrected for preterm birth (GA) rather than age from birth ( Fig. 23-7 ).