Measurement of Position Acuity in Strabismus and Amblyopia: Specificity of the Vernier VEP Paradigm

Abstract

purpose. An objective measure of positional acuity is desirable in the nonverbal clinical population. This study was conducted to investigate the specificity of the vernier VEP as a measure of positional acuity, evaluating the potential confound of asymmetric motion responses that may be present in some groups of patients. These motion responses could masquerade as position-specific responses, since they occur at the same response frequency as the vernier-related response.

methods. Twelve observers with early-onset esotropia (EOE), 30 children with untreated amblyopia, and 15 control children underwent swept vernier VEP acuity testing accompanied by a swept motion control stimulus. The control condition was used to detect the presence of artifactual responses not related to position sensitivity. The patients with EOE were selected for high levels of motion asymmetry as documented with oscillating gratings presented monocularly. As a measure of motion confound (penetration), the proportion of first-harmonic responses recorded in the control condition was determined.

results. The penetration rate in the vernier condition in each study group (EOE: 0.93%; amblyopes: 4.26%; normal subjects: 2.40%) and the entire group (2.85%) was acceptably low. The level of penetration was not significantly influenced by the presence of amblyopia.

conclusions. The vernier VEP paradigm, when applied in the manner described, can be interpreted as a measure of position sensitivity. The presence of motion asymmetry or untreated amblyopia does not affect the validity of vernier measurements made.

It is well-known, on the one hand, that grating acuity systematically underestimates optotype acuity losses in amblyopia. ^{1 2 3 4 5 6 7} On the other hand, losses on vernier acuity more closely match the loss of optotype acuity than do losses on grating acuity. ^{8 9} These results have prompted several attempts ^{10 11 12 13 14 15 16 17 18 19 20 21 22} to develop psychophysical vernier acuity tests for pre- or nonverbal subjects. 

Visual evoked potentials (VEPs) provide another means of assessing visual function in nonverbal subjects. VEPs specific to vernier offsets have been measured in several studies. ^{23 24 25 26 27 28 29} In each of these studies vernier offsets were introduced and withdrawn from the stimulus, resulting in a combination of vernier offset and motion-related responses. In each case, control conditions were used to verify which aspects of the response were specific to vernier offsets. Studies using the transient VEP ^{23 24 25} found that vernier offset stimuli produced an evoked response to very small offsets that did not produce a response to motion-control conditions. 

The steady state vernier VEP ^{26 27 28 29} involves the periodic introduction and withdrawal of vernier offsets in a high-contrast square-wave grating target (Fig. 1A) . The response contains two components: one related to the magnitude of the misalignment (positional information) that occurs at the odd harmonics of the stimulus frequency (first, third, fifth, and so forth) and the other related to the motion of the stimulus that occurs at the even harmonics. These components can be readily differentiated by using spectral analysis. ²⁶ The control condition utilizes the same amount of motion, but the moving elements are placed symmetrically with respect to the static reference elements (Fig. 1B) . This results in a symmetric misalignment–misalignment stimulus, with no change from collinear to noncollinear. The response of normal observers to the control condition comprises only even harmonics. In these observers, the odd harmonics to the alignment–misalignment stimulus are thus specific to the relative position of the static and moving elements and they reflect the output of a position-sensitive mechanism. ²⁶  

A difficulty in applying the steady state vernier VEP in clinical populations is the possible presence of motion responses that are not symmetric. A clear example of asymmetric motion VEP responses occurs in patients with early-onset esotropia (EOE) in which the monocular response to rapidly oscillating gratings is dominated by odd-harmonic components that reflect the asymmetry of motion processing. ^{30 31 32 33 34 35} A similar asymmetry is found in the monocular response of normal infants. ^{30 32 33 36} Odd harmonic responses due to asymmetric motion processing or other sources not related to position cues could be confused with odd harmonics arising from the processing of the vernier offsets. 

Asymmetry to horizontally induced motion has been extensively investigated using steady state VEP, ^{30 31 32 33 34 35 36 37 38} transient VEP, ^{39 40} oculomotor parameters (optokinetic nystagmus, pursuit eye movements), ^{39 40 41 42 43 44 45 46 47} and psychophysical and/or perceptual studies. ^{48 49 50 51 52}  

There have been fewer reports of vertical motion asymmetry, however, ^{43 45 49 52 53} and to our knowledge there is only one report of vertical motion asymmetry in specific relation to the steady state VEP. ³⁷ In these studies in which horizontal and vertical motion asymmetry were each assessed, there is the consistent finding that, when asymmetric responses to vertical motion are present, they are consistently less robust than asymmetric responses to horizontal motion. 

This evidence suggests that a horizontally oriented carrier grating with vertically introduced vernier offsets offers a better choice of stimulus design. In this study, we examine the levels of motion contamination and confound of the steady state vernier VEP by recordings of patients with early-onset esotropia who first had a high level of motion asymmetry documented with oscillating gratings. ^{32 34 35} In this group, the presence of a motion asymmetry was investigated in response to both horizontally and vertically oriented gratings. After this, the level of motion confound of the steady state vernier VEP was also tested for both vertically and horizontally introduced vernier offsets. 

The level of motion confound of the steady state vernier VEP was also tested in a group of children with untreated amblyopia (vertically introduced offsets only), since this group is a likely target for application of vernier acuity measurements in the clinical setting. A group of normal children served as control subjects. 

Methods

Observers

Early-Onset Esotropia.

Twelve participants with a history of EOE (mean onset, at 5 months) by history and/or medical records, were recruited for the study. The mean age at testing was 18.04 ± 12.94 (SD) years (range, 5–42). Each had a documented VEP motion asymmetry, as determined by a preliminary experimental protocol of oscillating gratings. ^{30 31 32 33 34 35}  

Participants in this group were tested with their habitual spectacle correction (if one was worn), which had been prescribed within the preceding year. In this as well as other study groups, clinical acuity was measured in logMAR (minimum angle of resolution) units, using the crowded Lea symbols, Early Treatment Diabetic Retinopathy Study (ETDRS)–style distance chart (Catalog no. 2503; Precision Vision, Bloomington, IL). Nine of the youngest children (six amblyopes and three normal) were not testable with the Lea chart, and the presence of amblyopia was assessed by fixation preference (FP), as described in the next section. In the EOE group, mean right eye acuity was 0.08 ± 0.14 logMAR (SD; range, −0.10–0.30). Mean left eye acuity was 0.00 ± 0.14 logMAR (−0.20–0.30). The group mean interocular acuity difference (IAD) was 0.09 ± 0.15 logMAR (0.00–0.50). Only three participants from this group had amblyopia, with IADs of 0.20, 0.20, and 0.50 logMAR. 

Amblyopia.

Thirty children with untreated amblyopia aged between 0.5 and 8 years were recruited consecutively. The group comprised 13 male and 17 female patients (10 with anisometropia, 10 with strabismus and, 10 with mixed disease) with a mean age of 4.76 ± 1.63 years (SD; range, 1.02–7.10). Children were considered to have amblyopia if the IAD was ≥0.2 logMAR. Those unable to perform this test reliably were assessed for amblyopia by the state of FP based on previously described techniques. ^{54 55 56} Briefly, in this study, grade 1 FP referred to patients consistently demonstrating equal FP between eyes (in three or all four of four trials). Grade 2 referred to patients able to hold fixation with the nonpreferred eye through a blink. Grade 3 patients were able to hold fixation with the nonpreferred eye only up to a blink and, grade 4 patients demonstrated an immediate switch back to the preferred eye when it was uncovered. Only Grade 3 and 4 patients were recruited as having amblyopia. (This equated to the grades used by Sener et al. ⁵⁴ grades 0 to 2 and grades 1 and 2 of Zipf et al. ⁵⁵ ) Participants with amblyopia were prescribed their full optical correction after cycloplegic retinoscopy, adjusting only for working distance. In participants able to perform logMAR chart acuity reliably, VEP testing began on the first occasion after spectacles were prescribed when the fellow-eye acuity became equal to or better than its previously recorded threshold in the unaided state. In those unable to perform chart acuity, VEP testing was performed when spectacle-wearing time was greater than 50% of waking hours. 

The group mean acuity of the amblyopic eye was 0.59 ± 0.38 logMAR (range, 0.20–1.77), and the group mean IAD was 0.46 ± 0.36 logMAR (0.20–1.52). Six patients were assessed by FP, and their grades were all four of four (immediate switch back to preferred eye). 

Normal Subjects.

Fifteen normally developing children without strabismus or amblyopia (seven male and eight female) with a mean age of 5.37 ± 2.28 years (SD; range, 0.60–8.58) were recruited. They were required to have a normal birth history without being more than 2 weeks premature or having low birth weight, to possess no manifest strabismus, and to have IAD ≤ 0.1 logMAR and/or FP of no worse than 1 (i.e., to prefer each eye equally). Anterior and undilated posterior segment examination of normal subjects, performed by one or both of two fully qualified pediatric ophthalmologists (SC or AC), also had to yield clinically normal findings. Potential observers were subjected to manifest retinoscopy with fogging control. Normal recruits found to have spherical and/or cylindrical refraction outside acceptable limits for their age ⁵⁷ were to be subjected to cycloplegic refraction; but none of the normal children met this requirement, and all were therefore tested in the unaided state. 

Group mean acuity was 0.03 logMAR in right and 0.05 in left eyes, with the same ranges (−0.10–0.20) and similar standard deviations (right = 0.10; left = 0.11). Group mean IAD was 0.01 ± 0.03 (0.00–0.10). Three children were assessed according to FP, as they were too young to be tested with the Lea chart. 

This work was performed with IRB approval. Informed written consent was obtained from the participant or guardian, and the protocol conformed to the tenets of the Declaration of Helsinki. 

VEP Recording and Analysis

The EEG was amplified by 50,000 for adults and 20,000 for children under 2 years of age over an amplifier pass-band of 1 to 100 Hz (−6 dB), using Grass amplifiers (model P511; Grass Telefactor, West Warwick, RI). The sampling rate was 600 Hz (16 bits), analog to digital conversion. The electrode montage consisted of Oz, O1, and O2, each referenced to Cz with additional bipolar derivations (O1–Oz, O2–Oz) and one Laplacian derivation Oz − [(O1 + O2)/2] being calculated off-line. Spectrum analysis was used to extract the amplitude and phase of the evoked response at the first six harmonics of the stimulus frequency. Specially written software (PowerDIVA, software, ver. 2.0; developed by one of the authors [AMN], Smith Kettlewell Institute, San Francisco, CA) was used for recording and analysis of evoked potential data. Details of the spectrum analysis ⁵⁸ have been presented elsewhere, and information related to the software’s algorithm for computing threshold data is presented later in this article. 

Oscillatory Motion VEP Test for Response Asymmetry

The symmetry of the monocular motion VEP was assessed in the EOE group by presenting the participants with high-contrast horizontal and vertical sine-wave gratings, each oscillating at 6 Hz and also at 10 Hz, hereafter termed motion VEP. The spatial frequency of the gratings was 2 cyc/deg and the oscillation was through 90° of spatial phase. The gratings were presented at 80% Michelson contrast at a space-averaged luminance of 150 candelas per square meter (cd/m²). The field size was 8° × 8° throughout. Stimuli were generated at a resolution of 1600 × 1200 on a high-bandwidth video monitor (model MR2000-HB; Richardson Electronics, LaFox, IL). The frame rate was 60 Hz, and the monitor contrast was linearized through a gamma table correction. 

The symmetry of the motion VEP was quantified with an asymmetry index. ^{30 31 33 34} Patients with asymmetric motion responses such as EOE display significant first-harmonic responses to oscillatory motion that are smaller or nonexistent in normal patients. Figure 2shows the spectral analysis of the VEP response to 10-Hz horizontal oscillatory motion in one of the EOE observers. The temporal frequency spectrum is profiled on the x-axis and response amplitude on the y-axis. The first (1F1, 10 Hz) and second (2F1, 20 Hz) harmonic responses are in black, and the responses of neighboring frequencies are shaded. It is clear that the 1F1 (4.8 μV) dominates the response and is greater than the 2F1 (1.3 μV), making the asymmetry index in this case equal to 0.79 [1F1/(1F1 + 2F1)]. In this study, only EOE patients with an asymmetry index ≥0.35 to horizontal motion were included as motion asymmetric. ^{31 59}  

Swept-Parameter Measurement of VEP Vernier Acuity

Vernier acuity thresholds were measured using vertical (EOE group only) and horizontal (all observers) 2-cyc/deg square-wave gratings into which vernier offsets were introduced. The display comprised equal height regions of moving and static pattern, with the distance between offsets being 0.50°. In all groups tested, vernier offsets were introduced vertically within a horizontally oriented square-wave grating (Fig. 1A) . Figure 1Adepicts a schematic of the vernier acuity stimulus. The offsets were presented in alignment–misalignment mode with the offset parameter swept from small to large (invisible to visible). In the vernier control condition (Fig. 1B)the offsets were presented symmetrically about the static reference panels to produce a misalignment–misalignment mode. 

All vernier acuity and vernier control conditions were swept-parameter conditions. The offset was swept in logarithmic steps over a 10-second recording period divided into 1-second epochs termed bins. The sweep range was 0.25 and 4 arcmin in observers aged 8 years or more (0.5–8 arc min under 8 years). ²⁷ This sweep range was altered toward the larger offset direction in patients with amblyopia who failed to produce a vernier acuity threshold for the starting sweep range. The sweep range for the control stimulus always matched that of the vernier acuity condition. Between 3 and 6 trials were averaged for each stimulus condition. 

Threshold Determination in Swept-Parameter Conditions

For each swept-stimulus condition, response thresholds are estimated by regression of amplitudes from the trial-average bins where the response increased monotonically with stimulus visibility. The range of bins eligible for regression depended on the statistical significance and phase consistency of the response, according to an algorithm adapted from Norcia et al. ⁶⁰ The regression range was limited to those bins in which the following criteria were met: (1) The response probability in each bin was at most 0.16; (2) the difference in response phase for each pair of consecutive bins was between 80° and −100°, where phase increased with response latency; (3) at least one pair of consecutive bins had a response P ≤ 0.077; and (4) to exclude spike artifacts, the amplitude of the bin immediately before and the bin immediately after any given bin in the range could not both be <0.3 times the amplitude of that given bin. Once the regression range was established, the threshold stimulus value was determined by extrapolating the regression line to the zero response amplitude. When applied to spectral data from background EEG, these criteria yielded a 5% false-positive rate (data not shown). 

Motion Artifact Categorization

Our criterion for detecting the intrusion of a motion artifact (penetration) into the measurement of the vernier VEP was the rate at which the vernier acuity scoring algorithm was triggered at the first harmonic (1F1) in the vernier control stimulus. The level of penetration is reported herein as both the total (relevant + irrelevant) number of penetrations and the number of relevant penetrations (when the 1F1 threshold of the control was less that that of the vernier condition). The notion of relevance takes into account clinically relevant versus irrelevant penetrations. A threshold at the first harmonic due to motion asymmetry is problematic (clinically relevant) if that threshold is better (lower) than the true vernier acuity and one would erroneously attribute too good an acuity to the patient. If the vernier onset–offset stimulus produces a lower threshold than the control condition (irrelevant penetration), it can only have resulted from mechanisms sensitive to the offset, per se, since the motion is the same in test and control conditions and the threshold measure is likely to be valid. 

Procedure

Early Onset Esotropia.

Monocular motion VEPs were measured in each eye under four viewing conditions (horizontal and vertical gratings, 6 and 10 Hz temporal frequency). The order of presentation of conditions was randomized within an eye and the order of eye tested first was randomized across observers. Observers from this group were then presented with an additional eight swept-parameter conditions (vernier acuity stimulus: alignment and misalignment; and vernier control stimulus: misalignment–misalignment control, in both horizontal and vertical orientation, right and left eyes). The order of presentation was randomized within eye, and the order of testing of eyes was randomized across observers. 

Children with Amblyopia and Normal Children.

VEP data were recorded in each eye for vernier acuity and vernier control stimuli in horizontal orientation with vertically introduced vernier offsets. These younger participants were part of a longer experimental protocol, and for this reason only a horizontally oriented stimulus with vertical offsets was used. The order of presentation was randomized within eye, and the order of testing of eyes was randomized across observers. 

Results

Motion Asymmetry in the EOE Group

The EOE group demonstrated horizontal asymmetry indices that were higher than previously reported in normal observers (Fig. 3 , Table 1 ). ^{31 33 34} Figure 3charts the mean asymmetry indices for the EOEs, grouped by temporal frequency, eye, and recording channel. It is clear that in every category, the indices for horizontal oscillatory motion (mean, 0.41 ± 0.14) were greater than for vertical (mean, 0.30 ± 0.10). Also across all channels and eyes, asymmetry indices were greater for 10-Hz (mean, 0.47 ± 0.14) than for 6-Hz motion (mean, 0.25 ± 0.14). 

The mean asymmetry index across channels was determined for each patient with EOE, in each of the eight different conditions tested. These eight conditions represented the possible combinations of three two-level factors: temporal frequency (6 or 10 Hz), grating orientation (horizontal or vertical), and eye (right or left). Because three observers were amblyopic, however, data from them were excluded from a repeated-measures ANOVA (Statistica GLM module; Statsoft Inc., Tulsa, OK) testing for interactions between experimental conditions. Two main effects were revealed: temporal frequency (F_(1,8) = 16.075, P = 0.004) and grating orientation (F_(1,8) = 7.315, P = 0.027). The effect of eye was not significant (F_(1,8) = 0.963, P = 0.355), nor were there any interactions between any of the factors. This group of patients thus manifests a high level of motion asymmetry that was significantly greater for horizontal than vertical oscillatory motion and was also greater at higher temporal frequencies (10 Hz vs. 6 Hz). 

Vernier Penetration in the EOE Group

Having established abnormal levels of motion asymmetry in the EOE group, we then proceeded to determine whether these asymmetries were likely to interfere with the measurement of vernier acuity. 

Figure 4shows first harmonic response functions from an EOE observer in the vernier condition and in the control condition and average noise. Increasing vernier offset is plotted in log terms on the x-axis and response amplitude in linear terms on the y-axis. This observer did not have a significant first-harmonic response in the control condition and thus does not demonstrate penetration. 

Figure 5shows an example of irrelevant penetration. In this figure the line segments between points of each function have been removed for clarity, and the regression lines determining the thresholds are included. The 1F1 for the vernier stimulus and its control both triggered thresholds. The vernier acuity threshold (0.30 arcmin) was, in this case, better than the control threshold (0.57), making this penetration irrelevant. 

An example of a relevant penetration is shown in Figure 6 , which has the same format as Figure 5 . In this case, the control threshold (0.57) was better than the vernier acuity threshold (0.62). Note that the peak signal-to-noise ratio in the nonpenetrating case (Fig. 4 , compare amplitude of solid lines to that of noise levels shown by dotted lines) was better than the cases in which penetration occurred (Figs. 5 6) , especially for the 1F1 of the control condition—a frequent observation. 

We report herein, the total penetration rate (relevant + irrelevant) and the relevant penetration rate. In this study, the scoring algorithm was applied 288 times to the EOE control data (six channels, two orientations, 24 eyes of 12 observers) yielding 205 thresholds. The total penetration rate (all eyes both orientations) was 17 of 205 applications for an 8.29% penetration rate but the relevant penetration rate was 0.49% (1/205). For vertically introduced offsets (Table 2) , the total penetration rate was 7.41% (8/108), but the relevant penetration rate was 0.93% (1/108) and therefore much less than the false-positive rate of the scoring algorithm itself. For horizontally introduced offsets, the total penetration rate was 9.28% (9/97), but the relevant penetration rate was 0% (0/97; not shown). 

Acuity Results: EOE Group

The logMAR acuity did not differ significantly between eyes of the EOE or normal group or between the right eye of EOE and normal subjects. As in the case of logMAR acuity, there was no significant difference in VEP vernier acuity between the EOE and normal subjects for vertical vernier offsets. There was no significant (P = 0.26) difference between the VEP vernier acuity thresholds generated by the two vernier orientations. 

Vernier Penetration in the Amblyopic Group

In this group, vernier offsets were introduced only in the vertical orientation. Four eyes in this group were untestable because of reduced cooperation. The scoring algorithm was applied 336 times to the amblyopic group data (six channels, one orientation; 56 eyes of 30 observers) yielding 188 thresholds. The total penetration rate was 17 of 188 applications for a 9.04% penetration rate (Table 2) . The relevant penetration rate, however, was 4.26% (8/188). 

Acuity Results: Amblyopic and Normal Group

There was a significant (P = 2.2 × 10⁻⁶) difference in logMAR acuity between amblyopic and fellow eyes of the amblyopic group, but no difference between fellow eyes of the amblyopic group and the right eyes of normal subjects. Findings were similar for the VEP vernier acuity thresholds, in that a significant (P = 0.0001) difference was found only between amblyopic and fellow eyes of the amblyopic group (Table 3) . 

Vernier Penetration in the Normal Control Group

As in the amblyopic group, vernier offsets were introduced only in the vertical orientation. The scoring algorithm was applied 178 times to the normal group data (six channels, one orientation; 30 eyes of 15 observers) yielding a total of 125 thresholds. The total penetration rate was 14 of 125 applications for an 11.20% penetration rate (Table 2) . The relevant penetration rate, however, was lower than the amblyopic group at 2.40% (3/125). 

Vernier Penetration in the Entire Study Group

Considering vertically introduced vernier offsets only, which was tested across all study groups, the total penetration rate was 9.26% (39/421), whereas the relevant penetration rate was 2.85% (12/421). Concerning relevant penetrations, the mean difference in thresholds between the vernier and control conditions was 0.52 arcmin (median, 0.36 ± 0.52). The extremely low number of relevant penetrations between horizontal and vertical vernier offset conditions tested in the EOE group precludes meaningful comparison. 

It may appear from cursory inspection of the total penetration rates (Table 2)that left eyes seem more likely to penetrate, at least in the normal and EOE groups. Overall, there were 21 penetrations in left eyes and 18 penetrations in right eyes. As expected therefore, χ² analysis ⁶¹ ruled out eye as a factor in total penetration rate (P = 0.37). Furthermore, the conditions and eyes were randomized, making order effects much less likely. 

Discussion

The present study shows that monocular, position-specific VEP responses can be measured with low levels of contamination by motion artifacts in patients with strabismus and/or amblyopia. Acuity estimates based on regression of the first harmonic recorded from vernier onset–offset stimuli are better than those recorded from a symmetric control condition in 97% of cases. Thus, in only 3% percent of cases did we find evidence for a motion response that would lead to an erroneous determination of the vernier threshold. These erroneous determinations constitute false positives for the vernier acuity test procedure (e.g., concluding position-specificity where there is none). This false-positive rate is similar to that of the scoring algorithm itself when applied multiple times across recording channels and stimulus conditions. 

Previous work ^{26 27} using a vernier stimulus such as the one used herein failed to detect evidence of motion penetration into the first harmonic response of the alignment–misalignment stimulus, but these studies were on a smaller scale and tested only normal children and adults. More important, these two studies used binocular recordings which may have obscured motion asymmetries that would be expected to cancel under binocular viewing conditions. We show here that the steady state vernier VEP is similarly immune to motion artifacts both in untreated patients with amblyopia (age range tested here), as well as in patients with EOE who had documented high levels of motion asymmetry. When penetration did occur, the signal-to-noise ratio was usually lower, especially in the control condition. 

Horizontal Versus Vertical Motion VEP Asymmetries for Oscillating Gratings

As part of our effort to determine the extent to which the vernier VEP paradigm can be contaminated, asymmetries in motion processing were studied in a group of patients with high levels of horizontal motion asymmetry, as determined with oscillating gratings. ^{32 34 35} Previous behavioral studies in infants ⁵² and in adults with strabismus ^{43 45 49} or amblyopia ^{45 49} and previous VEP work from our laboratory ³⁷ has found lower levels of motion asymmetry for vertical as opposed to horizontal motion. We also find the magnitude of VEP vertical motion asymmetry to be significantly less than that for horizontal. We may have observed low penetration rates for the vertical motion condition because of this, but if motion orientation on its own is the sole controlling factor, we would expect to have seen more penetration with horizontal motion. We did not. 

It is more likely that either the temporal frequency or spatial configuration of the vernier stimulus do not engage the motion asymmetry mechanism. The VEP motion asymmetry is usually measured with higher temporal frequencies (6 or 10 Hz) and with full-field grating oscillation rather than with smaller, interspersed sections of static and moving gratings. Regarding temporal frequency as a factor, we found a lower level of asymmetry for 6-Hz than for 10-Hz gratings, and this trend may continue as temporal frequency is lowered. At present, it is not known what the lower temporal frequency is for the mechanisms underlying the VEP motion asymmetry. Psychophysical direction selectivity at threshold ceases below approximately 0.5° per second ⁶² and one would thus expect there to be a lower temporal frequency limit for the VEP motion asymmetry. Normal development of a symmetric motion VEP proceeds from lower to higher temporal frequency in infants, ⁶³ and recovery of a symmetric motion response after surgery for EOE may be more complete for lower than higher temporal frequencies. ³³ Finally, the use of a pattern of interleaved static and moving regions may interfere with integration of motion responses by large receptive field mechanisms. 

Motion Artifacts in Behavioral Vernier Acuity Tasks for Preverbal Observers

In the pre- or nonverbal population, attempts to determine vernier acuity by the forced-choice, preferential-looking method ^{10 11 12 13 14 15 16 17 18 19 20 21 22} have had to rely on difference in saliency between a stimulus with a vernier offset and a collinear one. The discrimination between two salient stimuli in this case may lead to a less robust preference than is the case in grating acuity measurements in which only one salient stimulus is presented. It is commonly reported ^{10 13 15 21 22} that many infants were either unable to complete the testing protocol or had psychometric functions that did not reach high levels of percent correct. It is well known that infants prefer to look at a moving or flickering stimulus, ^{20 64} rather than a stationary one. Many studies of infant vernier acuity ^{11 14 15 16 17 18 20 21 22} incorporated some element of motion in their vernier stimulus that otherwise maintained the magnitude of the offsets presented. Although attempts to counteract this introduced motion were made in some, ^{15 16} none used adequate control for the introduced motion. The fact that motion can influence behaviorally measured vernier thresholds has been clearly documented. ^{17 18} The stimuli used in these studies ^{11 14 15 16 17 18 20 21 22} therefore could very well have been confounded by motion and as such may not reflect solely the measurement of a position-sensitive mechanism. 

Potential limitations of the VEP Vernier Paradigm

The results for high-frequency oscillation described earlier and in Figure 3suggest that caution should be used at higher temporal frequencies. The VEP motion asymmetry was in fact first observed in a deeply amblyopic patient tested with the vernier onset stimulus at 10 Hz (Norcia AM, Wesemann W, unpublished observations, 1985). 

The group of patients with amblyopia tested in this study had no previous treatment and as such, our findings may not be applicable at other stages of treatment. Changes in motion asymmetry during treatment have been found after surgical realignment of strabismus ^{30 33} and alternate occlusion. ³¹ In both cases, treatment resulted in decreases in motion asymmetry, however, and thus would not be expected to increase motion penetration in the vernier task. 

If the vernier VEP technique we have described is to be applied in a novel situation, such as patients with a different diagnoses, it would be advisable to include the motion-control condition as a precaution. 

 Supported by Grant OR2001-99a from The Guide Dogs for the Blind Association, United Kingdom, and The Smith Kettlewell Eye Research Institute, San Francisco, California.

 Submitted for publication June 22, 2005; revised July 26, 2005; accepted September 26, 2005.

 Disclosure: S.I. Chen, None; A.M. Norcia, None; M.W. Pettet, None; A. Chandna, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Schematic of swept vernier acuity and vernier control stimuli. The vertically introduced offsets are modulated between alignment and misalignment (A, vernier acuity) and misalignment and misalignment (B, vernier control) at 3.76 Hz. The magnitude of the offset is increased in 10 logarithmic steps toward the visible direction. The range of offset and amount of symmetric motion are matched between (A) and (B).

Figure 2.

 

Motion asymmetry in a patient shown by spectrum analysis of the response to 10-Hz horizontal oscillatory motion. Temporal frequency and response amplitude are plotted on the x-axis and y-axis, respectively. Compared with other frequencies and the second harmonic (2F1 = 1.3 μV), the first harmonic (1F1 = 4.8 μV) dominates, making this patient’s response highly asymmetric (asymmetry index = 0.79).

Figure 3.

 

Results of group mean (±SEM) asymmetry indices for EOEs by temporal frequency, eye, and recording channel. Responses to horizontal induced motion was consistently more asymmetric than those to vertical induced motion, and 10-Hz motion induced greater response asymmetry than 6 Hz in this group.

Table 1.

 

Early-Onset Esotropes: Mean Asymmetry Index for Horizontal and Vertical Induced Motion

	6 Hz					10 Hz
		RE	LE	Both	RE	LE	Both
Horizontal	0.26 (0.14)	0.31 (0.20)	0.29 (0.17)	0.50 (0.22)	0.53 (0.19)	0.52 (0.17)
Vertical	0.19 (0.13)	0.22 (0.14)	0.21 (0.13)	0.41 (0.16)	0.37 (0.19)	0.40 (0.16)

 Data are the mean ± SD group asymmetry index in response to induced motion in either eye or both eyes.

Figure 4.

 

Patient data showing the specificity of the first-harmonic (1F1) response to the vernier stimulus (no penetration). The 1F1 response for the control condition was at noise level, whereas the vernier acuity response rose monotonically with increasing vernier offset.

Figure 5.

 

Patient data showing irrelevant penetration. The vernier acuity response triggered a threshold (0.30 arcmin) by regression to zero amplitude. The vernier control response also triggered a threshold (0.57 arcmin) by regression that was worse, making this an irrelevant penetration.

Figure 6.

 

Patient data showing relevant penetration. The vernier acuity response triggered a threshold (0.62 arcmin). The vernier control response triggered a threshold (0.57 arcmin) which was better (lower) than the vernier acuity threshold.

Table 2.

 

Total and Relevant Penetration Rates for Vertical Vernier Offsets

Study Group	Eye	Penetration Index			Vernier Thresholds n (100%)
				Total Penetrations n (%)		Relevant Penetrations n (%)
EOE	RE	2 (3.85)	0 (0.00)	52
	LE	6 (10.71)	1 (1.79)	56
Amblyopes	Amblyopic	6 (7.23)	4 (4.82)	83
	Fellow	11 (10.48)	4 (3.81)	105
Normal Subjects	RE	5 (7.46)	1 (1.49)	67
	LE	9 (15.52)	2 (3.45)	58
All Groups	Both	39 (9.26)	12 (2.85)	421

Table 3.

 

Threshold Yield and Mean VEP Vernier Acuity Thresholds for Vertical Vernier Offsets

Study Group	Eye	Threshold Count			Mean Threshold Value arcmin (SD)
				Possible (n)		Actual (n)
EOE	RE	72	52	0.71 (0.57)
	LE	72	56	0.56 (0.27)
	Both	144	108
Amblyopes	Amblyopic	168*	83	1.65 (1.45)
	Fellow	168*	105	0.91 (0.67)
	Both	336*	188
Normal Subjects	RE	89	67	0.86 (0.85)
	LE	89	58	0.86 (0.70)
	Both	178	125

 Possible (n) is the number of thresholds that could be derived, given the number of observers, eyes, and recording channels. Actual (n) is the number of instances in which a threshold was actually returned.

*  Reduced number, since recording was not possible in four eyes in this group (30 observers).