A Novel Dichoptic Optokinetic Nystagmus Paradigm to Quantify Interocular Suppression in Monocular Amblyopia

Abstract

Purpose: To develop a novel dichoptic optokinetic nystagmus (OKN) paradigm and investigate its effectiveness in objectively quantifying the interocular suppression in subjects with monocular amblyopia.

Methods: Centripetal moving gratings with different contrast ratios and constant velocity were dichoptically presented to eight monocular anisometropic amblyopes and eight normal subjects. We analyzed the OKN records with an eye tracker (EyeLink; SR-Research, Ontario, Canada) to obtain the relationship between ocular-dominance of OKN and the interocular contrast ratio by fitting power curves, and examined the correlation between the effective contrast ratio for a balanced OKN and the visual acuity of the amblyopic eye in amblyopes.

Results: In normal subjects, the OKN pursuit times were roughly balanced for opposite directions when stimulated with centripetal gratings of same contrast; however, in amblyopes, the OKN pursuit times of the dominant eye exceeded that of the amblyopic eye. Increasing the contrast of one eye's grating led to an increase in its OKN dominance. The OKN directional ratio (y) could be well fitted by a power function of the interocular contrast ratio (x): y = ax^b. Moreover, in amblyopes, the effective contrast ratio (x_b) for a balanced OKN correlated significantly positively with the visual acuity of the amblyopic eye (Spearman's correlation coefficient, 0.9698).

Conclusions: The OKN induced by dichoptic gratings moving centripetally could be used as a reliable measure to objectively quantify the interocular suppression. This paradigm, avoiding the need for subjective report from patients, offers a promising alternative index for research on the mechanisms of amblyopia and in clinical practice.

Amblyopia is a condition characterized by reduced visual acuity that is uncorrectable by refractive means, in the absence of ocular pathology, and has a prevalence of 2%∼3%.¹ Researchers have reached a consensus that abnormal binocular interactions, especially interocular suppression, is involved in monocular amblyopia, rather than monocular deficits.^2–7 

Subjective methods, such as Worth 4-dot test and Bagolini glasses, are widely used to qualitatively evaluate interocular suppression in clinical practice. Recent scientific studies have quantified the degree of suppression using subjective responses to a series of psychophysical paradigms,^2,6,8,9 and found that it correlated significantly with the degree of amblyopia, which is to say that greater suppression was associated with greater losses in visual acuity and in stereopsis.¹⁰ All the methods mentioned above are subjective ways to assess interocular suppression, and there is as yet no objective method to do so. Antisuppression treatments have proven effective in adult strabismic amblyopes, with reduced suppression, improved monocular acuity, and reestablished stereopsis.^11,12 Based on these facts, the quantitative and objective measurement of interocular suppression is critical in not only clinical diagnosis and treatment of amblyopia, but also to comprehensively understand this disorder. Psychophysical paradigms seem unsuitable for subjects who lack compliance or the ability to discriminate subjectively. Therefore, an objective method to efficiently quantify interocular suppression in subjects with monocular amblyopia is required. 

Interocular suppression occurs during binocular rivalry, where the suppressed eye suffers a general reduction in sensitivity.^13,14 The eye rivalry hypothesis proposes that binocular rivalry results from the competition for dominance between the two eyes, or that interocular competition mediates binocular rivalry.¹⁵ It is accepted that the stimulus strength in binocular rivalry is closely related to the temporal pattern of the rivalry, where the stimulus strength can depend on several physical features, including contrast.¹⁶ An increase in the stimulus strength, such as an increase in contrast, increases the dominance of that stimulus. In view of these facts, binocular rivalry can be induced by opposing input stimuli, and interocular suppression can be further induced with unequal stimulus strengths (unequal contrast). 

Optokinetic nystagmus (OKN) is involuntary conjugate oscillations in response to visual-field motion, which consist of a slow phase (pursuit) in the same direction as that of the motion and a fast phase (saccade) in the opposite direction. Previous studies have demonstrated that OKN could be elicited by gratings and measured objectively.^17–19 Previous dichoptic experiments have shown that opposite directly moving grating stimuli resulted in alternating OKN and the perception of binocular motion rivalry.^20–24 Previous studies have also shown that the direction of OKN slow phase can be used as an objective indicator of the perceived direction of dominant motion during binocular rivalry, leaving voluntary reports unnecessary.²¹ One study showed that OKN dominance was biased toward relatively high-contrast and high-luminance stimuli.²⁵ Previous studies have also used OKN to measure the contrast sensitivity function,²⁵ which is a better predictor of functional vision.²⁶ 

According to the studies discussed above, the OKN directional shifts generated by dichoptic oppositely moving gratings with different contrast ratios may reflect the objective dominance relationship between the two eyes, and also the binocular rivalry and the extent of interocular suppression. Based on the hypothesis, OKN rivalry may be a promising paradigm with which to quantify interocular suppression. 

We investigated a novel paradigm in which centripetally moving gratings with different contrast ratios were used to induce OKN binocular rivalry, and quantify the extent of interocular suppression. In subjects with monocular amblyopia, if the strengths of the input signals for both eyes were adjusted experimentally to be almost the same, binocular rivalry should reach a balance point, and the suppression of the fellow eye to amblyopia eye would be relieved. 

Methods

Subjects

Eight amblyopia subjects (amblyopia group) and eight age- and sex-matched normal subjects (control group) were enrolled in the study. The characteristics of the amblyopia subjects are listed in Table 1. The mean age of the amblyopia group (two females, six males) was 21.63 ± 7.44 years. All the amblyopia subjects were recruited from the Amblyopia Clinic of Shanghai Eye and ENT Hospital, Fudan University, Shanghai, China. The study was performed in compliance with the tenets of the Declaration of Helsinki, with the approval of the Ethics Committee of Fudan EENT Hospital. All participants involved in the study gave their written informed consent before the experiments. 

Table 1

 

Characteristics of Amblyopia Subjects

 

We only included subjects with anisometropic amblyopia caused by unequal refractive error in their two eyes (>2.50 diopters [D]), in the absence of any ocular pathologic condition. We excluded subjects with monocular amblyopia attributable to strabismus or deprivation. Amblyopia 3 and 7 had late diagnoses and received no treatment, whereas poor outcomes in the remaining subjects were due to unsatisfactory compliance, despite early diagnoses and the early application of occlusion therapy. In our study, we specifically defined amblyopia as over 2-line difference in visual acuity of two eyes. 

Stimuli and Procedures

All stimuli in the subsequent experiments were generated and controlled in a computational environment (MATLAB with Psychophysics Toolbox extensions; MathWorks, Inc., Natick, MA, USA), and were displayed on a computer remote terminal (P1917, 4:3; Dell, Texas, USA) monitor. The resolution of the display was 1024 × 768 pixels and the refresh rate was 60 Hz. The eye movement of the left eye was recorded with an eye tracker (EyeLink 1000 Tower; SR Research), with a sampling rate of 1000 Hz. 

Two drifting horizontal sinusoidal gratings (30° vertically and 20° horizontally) were dichoptically presented with a mirror stereoscope. The gratings were drifted horizontally in centripetal directions to induce vigorous rivalry. The gratings presented to both eyes were moving at the same velocity (4° or 7°/second). The mean luminance of the gratings was 36 cd/m², measured with a luminance meter (CS-100A; Konica-Minolta, Tokyo, Japan). The contrast presented to one eye (the amblyopic eye in amblyopia subjects and the matched eye in control subjects) was fixed at 100%, whereas the contrast presented to the other eye (the fellow eye in amblyopia subjects and the matched eye in the control subjects) were set at five different levels (100%, 80%, 40%, 20%, or 10%). Each condition lasted 20 seconds followed by a 5-second break, and was repeated three times, followed by a 10-minute rest period. Five contrast conditions were performed randomly for each subject. 

During the experiments, the subjects wore a pair of refractive-corrected glasses and sat in a dimly lit room at a viewing distance of 57 cm from the monitor. A chin rest was used to help fix the subject's head position. Before each condition, the subject was instructed to perform an alignment task to ensure that his/her two eyes aligned properly. Although previous studies have demonstrated a close correlation between subjective reports and eye moving recordings, two subjects (subject 4 from the control group and subject 2 from the amblyopic group) were instructed to press keys as continuous responses to the directions of the movements they perceived during the whole experiment. 

Analysis

OKN Slow-Phase Eye Movements Analysis

Preprocessing: A Velocity-Time Plot of OKN Pursuits

In pilot experiments, the OKN record gradually stabilized within 2 seconds after initiation of the stimuli and recording. We used eye tracking software (SR Research) to obtain a consistent 10-second stable records with the least blinks under specific contrast conditions and drew velocity (horizontal)-time plots from the raw data. All the lost data (velocity = 1 × 10⁸) due to blinks or other errors and every saccade of the OKN in eye movement data were discarded.²⁷ After all the saccades and blinks were removed, a velocity-time plot of all the OKN slow phases during the chosen period appeared, in which the positive values represented OKN pursuits from right to left and the negative values represented those in the opposite direction. 

Definition of “OKN Directional Ratio”

In a preprocessed velocity-time plot of the OKN pursuits, the leftward OKN time (right eye dominant time) was defined as the total cumulative time of all the leftward OKN pursuits (all the positive velocity values), whereas the rightward OKN time (left eye dominant time) was defined as the total cumulative time of all the rightward OKN pursuits (all the negative velocity values). The OKN directional ratio was defined as the OKN time dominant by amblyopia eye (or the matched eye in the control subjects) divided by the OKN time dominant by fellow eye. 

Subjective Motion Perception: Cross-Correlation Function

To verify whether the subjective responses to binocular rivalry were consistent with OKN recordings, as in previous pilot studies,^21,22 we compared the subjective reports of perceived motion direction with the OKN results of two observers (one from the control group and one from the amblyopia group). We selected a 10-second stable eye movement record (blue plot) and the discrimination responses (black plot) during the corresponding period and superimposed the two plots (Table 2; Fig. 1). The negative and positive values represent the rightward and leftward directions of the perceived motion, respectively, in the subjective reports, which might match the OKN eye movement direction at that time. 

Table 2

 

Proportions of Correlated OKNs Under Five Different Contrast Conditions in Two Subjects (one from the control group, one from the amblyopia group)

 

Figure 1

 

Plots of 10-second stable eye movement records and the subjective response plots during the corresponding period in one control and one amblyopic subject. (A) Normal subject 4 with 40% contrast presented to the fellow eye. (B) Amblyopic subject 2 with 40% contrast presented to the fellow eye.

If the direction of perception reported was the same as the OKN direction, the OKN was regarded as a correlated one. The proportion of correlated OKNs was defined as the number of correlated OKNs divided by total number of OKNs. A larger proportion of correlated OKNs suggested a closer correlation between the subjective perceived motion and the objective OKN results. 

Fitting Curve Between OKN Directional Ratio and Interocular Contrast Ratio

The relationship between the OKN directional ratio (y) and the interocular contrast ratio (x) in every subject (5 points) fitted the curve of the power function y = f(x) = ax^b (a > 0, b < 0) by “cftool” in MATLAB, with a goodness-of-fit > 0.80. 

Definition of “Effective Contrast Ratio at Balance Point”

When the OKN directional ratio reached 1, the total cumulative time of leftward and rightward OKN each accounted for 50%, indicating that binocular rivalry had reached a balance point under equal centripetal stimuli. For each amblyopia subject, the effective contrast ratio (x_b) at balance point was defined as the predicted value for x when y = f(x) = 1, which was calculated based on the function of the fitting curve. The effective contrast ratio represented the contrast ratio at which the two eyes showed balanced performance when the signal presented to the fellow eye was artificially attenuated. 

Statistical Analysis

The correlation between the visual acuity of amblyopic eyes and the effective contrast ratio at balance point was measured with Spearman's correlation. 

Results

Direction of OKN Slow Phase and Perceived Direction

The relationship between the OKN slow phase recorded with EyeLink and the psychophysical results reported by the subjects was analyzed. The blue plot in Figure 1 represents the velocity-time plot and the black plot represents the perceived direction of motion. With the red line indicating 0, all the positive and negative values represent the leftward and rightward directions, respectively, on the two plots shown in Figures 1A and 1B. The OKN shifts in a consistent stable 10-second recording correlated strongly with the reported perceived direction of motion during the corresponding period in these subjects. The proportion of correlated OKNs was >85% in nearly all trials (Table 2). 

The blue plot represents the velocity-time plot and the black plot represents the reported perceived direction of motion (press-time plot). The red line indicates 0, and all the positive and negative values represent the leftward and rightward directions, respectively, in these two plots. The OKN shifts correlate strongly with the motion direction discrimination reports. 

Correlation of OKN Slow Phase in Normal Subjects With Interocular Contrast Ratio

The OKN directional ratio differed when the interocular contrast ratio changed, while the velocity remained fixed. The OKN directional ratio was nearly 1 in all the control subjects when the contrast presented to both eyes was 100%, suggesting a balanced binocular rivalry when the eyes were exposed to centripetal moving gratings with the same contrast. For a specific control subject, as the contrast presented to the fellow eye decreased (100%, 80%, 40%, 20%, and 10%), the OKN directional ratio increased monotonically, indicating that the fellow eye was gradually suppressed under the unbalanced grating stimuli. The variation trend generally fitted the curve of a power function, from which we determined the following equation (Fig. 2): y = f(x) = ax^b (a > 0, b < 0). Although the values of a and b varied among the control subjects, the OKN directional ratio (y) showed a remarkable negative correlation with the interocular contrast ratio (x) in all normal subjects. The average goodness of fit in the eight normal subjects was 0.91 ± 0.10 (mean ± SD). 

Figure 2

 

Fitting curves (OKN directional ratio versus interocular contrast ratio) for the eight normal subjects in the control group.

The data for different subjects are shown in separate panels. In each panel, the vertical axis represents the OKN directional ratio and the horizontal axis represents the interocular contrast ratio. The blue solid lines represent the fitting curves and black dots represent the OKN directional ratios with specific interocular contrast ratios. The identifier (ID) and power function f(x) = ax^b (a > 0, b < 0) of each subject are shown on the top of each panel. The OKN directional ratio was nearly 1 in all the control subjects when the contrast presented to both eyes was 100%. The OKN directional ratio (y) correlated negatively with the interocular contrast ratio (x) in all normal subjects. 

Correlation of OKN Slow Phase in Amblyopia Subjects With Interocular Contrast Ratio

In the amblyopic group, when both eyes were given gratings of 100% contrast, the OKN directional ratios were much less than 1, which suggests that the amblyopic eye was apparently suppressed by the fellow eye in subjects with monocular amblyopia. 

For a specific amblyopia subject, as the contrast presented to the fellow eye decreased (100%, 80%, 40%, 20%, and 10%), the OKN directional ratio increased correspondingly, as in the normal subjects, which suggests that the suppression of the fellow eye to the amblyopic eye was gradually relieved due to the increasingly balanced stimuli. 

The correlation between the OKN directional ratio (y) and the interocular contrast ratio (x) in the amblyopic subjects also fitted a power curve, as in the normal subjects. The average goodness of fit in the eight amblyopic subjects was 0.84 ± 0.25 (mean ± SD), when y = f(x) = ax^b = 1, x = a^−1/b (Fig. 3). As expected before, the predicted x value represented the effective contrast ratio at which the two eyes of a subject with monocular amblyopia were almost balanced. The effective contrast ratio (x_b) at balance point was the contrast ratio at which the fellow eye was considered to be artificially suppressed in order to achieve input signals of equal strength. 

Figure 3

 

Fitting curves (OKN directional ratio versus interocular contrast ratio) of the eight subjects in the amblyopic group.

The data for different amblyopia subjects are shown in separate panels. In each panel, the vertical axis represents the OKN directional ratio and the horizontal axis represents the interocular contrast ratio. The blue solid lines represent the fitting curves and the black dots represent the OKN directional ratios at specific interocular contrast ratios. The predicted x values when y = f(x) = 1 are shown with black broken lines. The ID and power function f(x) = ax^b (a > 0, b < 0) of each subject is shown on the top of the panel, together with their effective contrast ratio (x_b) at balance point and the visual acuity of the amblyopic eye. In all amblyopes, the OKN directional ratio was much less than 1 when both eyes were given gratings of 100% contrast. The OKN directional ratio (y) correlated negatively with the interocular contrast ratio (x) in all the amblyopic subjects. 

Effective Contrast Ratio and Visual Acuity, Interocular Suppression

The effective contrast ratio at balance point showed a notable positive correlation with the visual acuity of the amblyopic eye in subjects with monocular amblyopia (Spearman's correlation coefficient = 0.9698; Fig. 4). A lower effective contrast ratio at balance point indicated stronger suppression of the fellow eye relative to the amblyopic eye. According to the results described above, the poorer the visual acuity in the amblyopic eye, the more serious was the interocular suppression. 

Figure 4

 

Linear fitting of the effective contrast ratio at balance point and the visual acuity of the amblyopic eye in eight monocular amblyopic subjects—Spearman's correlation (r_s = 0.9698).

The vertical axis represents the effective contrast ratio at balance point and the horizontal axis represents the visual acuity of the amblyopic eye. The blue solid line represents the Spearman's correlation fitted line, and the eight black dots represent the effective contrast ratios at balance point in the eight monocular amblyopic subjects. 

Discussion

The traditional methods of measuring interocular suppression (Worth 4-dot test and Bagolini glasses) offer a subjective yet qualitative conclusion. Previous studies have quantitatively evaluated interocular suppression by analyzing the subjective responses to various psychophysical paradigms,^2,6,8,9 which may be difficult for those subjects who are not capable of discriminating them subjectively or those with unsatisfactory compliance. Therefore, an objective, unrestricted, and convenient measurement would be clinically useful. 

In this study, a novel OKN paradigm was designed to objectively quantify interocular suppression. We also established the correlation between the effective contrast ratio at balance point and the visual acuity of the amblyopic eye, which may offer an innovative approach to studying the mechanisms of amblyopia and its clinical management. 

As expected, the OKN generated by dichoptically presented centripetal gratings reflected the objective situation of binocular rivalry and interocular suppression. In our study, we enrolled subjects whose visual acuity differences between two eyes were over two lines. Hence, we had a wider range of amblyopia severity considering its usefulness for assessing the correlation between visual acuity and suppression. As expected, the poorer the visual acuity of amblyopic eye, the less was the effective contrast ratio at balance point, which suggests that the input to the fellow eye should be attenuated with a lower-contrast stimulus to strike a balance between the two eyes. However, the contrast manipulation might have no effect to amblyopia group participant 7, indicating that there might be individual differences in this OKN paradigm. 

It is widely accepted that stronger suppression is associated with deeper amblyopia.^28–30 Our results demonstrate the positive correlation of these two phenomena rather than a causal link. It is difficult to tell whether amblyopia leads to suppression or suppression brings about amblyopia, or whether suppression and amblyopia are two separate outcomes of a primary as-yet-unknown cause. The hypothesis that suppression appear to be a consequence of amblyopia might explain the success of the current and classical treatment strategy for amblyopia (patching), regardless of suppression, and the fact that many individuals with suppression have equal visual acuity in both eyes. Our results are compatible with recent findings that the degree of suppression correlates with the degree of amblyopia and the loss of stereopsis in amblyopic subjects of various types,¹⁰ which both suggest that amblyopia can be attributed to suppression. In another context, the idea that suppression is independent of amblyopia is supported by the previous negative correlation between suppression and amblyopia (contrary to our results), suggesting that when an eye is amblyopic, there is no longer a need for strong suppression of that eye by the contralateral eye.³¹ More research is required to figure out the casual connection of these two phenomena and the detailed mechanisms underlying them. 

Previous pilot studies concluded that OKN could be steadily elicited at the velocity value of a wide range (1.4°–181.3°/second in human,²⁵ 4°–12°/sec in macaque monkeys²²). In the present study, every subject chose the most comfortable velocity from either 1°, 4°, 7°, or 10°/second. In subjects, whether normal or amblyopic, the negative correlation between the OKN directional ratio and the contrast presented to the fellow eye did not differ significantly, despite varied conditions of velocity (1°, 4°, 7°, 10°/second). 

The alternation of dominance during binocular rivalry depends upon the stimulus strength variables, such as luminance,³² contrast,^33–35 and spatial frequency.³⁶ It remains unclear whether varying the spatial frequency or luminance affects our dichoptic OKN paradigm, because the spatial frequency and luminance were fixed in our experiments. 

A limitation of this study was that the sample size was too small. Secondly, the OKN paradigm now would be very difficult to deploy in a clinical setting. 

Moreover, we only enrolled adult anisometropic amblyopia subjects in our study; further attempts could be performed in other types of amblyopia subjects such as strabismus, deprivation amblyopia, or infant/young children in order to test whether the paradigm could be widely used. 

Acknowledgments

Supported by the National Nature Science Foundation of China grants (81500752, 81770957) and the Shanghai Committee of Science and Technology Project (134119a8900). 

Disclosure: W. Wen, None; S. Wu, None; S. Wang, None; L. Zou, None; Y. Liu, None; R. Liu, None; P. Zhang, None; S. He, None; H. Liu, None