Evaluating Eye Tracking During Dichoptic Video Viewing With Varied Fellow Eye Contrasts in Amblyopia

Abstract

Purpose: This study uses eye tracking to investigate how varying fellow eye (FE) contrast during dichoptic video viewing influences eye movement patterns, and their associations with interocular suppression, visual acuity, and stereoacuity deficit in amblyopia.

Methods: Eye movements of 27 amblyopic and 8 healthy control participants were recorded during dichoptic viewing of stationary dots and videos with FE contrasts (100%, 50%, 25%, and 10%). Analysis included durations the amblyopic and FE spent in different stimulus regions, fixation switches, and eye deviation, and correlating these with suppression, visual acuity, and stereoacuity.

Results: Participants with pronounced suppression, visual acuity, and stereoacuity deficits demonstrated reduced amblyopic eye fixation in the amblyopic eye (AE) region at 100% FE contrast. Lowering FE contrast increased amblyopic eye duration in stimuli presented within the AE region, notably in anisometropic and treated strabismic participants, and strabismic participants exhibiting fixation switches during viewing of dichoptic stationary dots. Even at lower FE contrasts, participants with greater stereoacuity and visual acuity deficits continued to exhibit diminished AE fixation in the AE region. Increased eye deviation was seen in strabismic participants with lowering of FE contrasts.

Conclusions: Dichoptic contrast modulation holds promise for reducing suppression with responses varying by amblyopia type and visual function deficits. Larger strabismic angles may hinder binocular benefits of dichoptic treatments. Fixation switches may serve as an indicator of favorable outcomes. Eye tracking is crucial for understanding these dynamics, providing essential insights into visual attention dynamics of the FE and AE, and may serve as a valuable tool in optimization of amblyopia treatments.

Amblyopia is a neurodevelopmental disorder that arises from discordant visual input from the two eyes during critical periods of visual development from causes such as strabismus, anisometropia, deprivation, or other mixed mechanisms. Amblyopia is a multifaceted condition affecting visual sensory functions in both the amblyopic eye (AE) and fellow eye (FE), along with reduced depth perception and interocular suppression.^1–3 Treatment typically involves patching, but residual visual function deficits or regression after treatment remain common.^4–10 There is increasing interest in newer dichoptic therapies which target interocular suppression. One such therapy involves reducing the contrast in the FE while maintaining 100% contrast in the AE. This is combined with dichoptic masking, where complementary parts of images are presented to each eye separately.^11–15 Although the premise of the dichoptic treatments is a greater reduction in suppression with better stereopsis than patching, the studies have produced mixed results.^16–18 New-onset or worsening of eye deviation is seen in up to 8% to 10% of patients treated with patching, atropine, and dichoptic treatments.^19–21 We have quantified fixation eye movements while amblyopic participants looked at dichoptic stationary dots, where the contrast of the dot presented to the FE was varied, whereas the contrast of the dot presented to the AE was maintained at 100%.^22,23 We found that the eye deviation can change while viewing dichoptic stationary dots.²² Eye tracking has shown that a dichoptic stationary dot stimulus can result in fixation switch in strabismic participants, whereas the AE fixates and attends to the target stimulus with reduced FE contrast. Interestingly, different individuals showed this fixation switch at different FE contrasts.^22,23 We found that participants who exhibited the fixation switch at 100% FE contrast had less suppression than those who switched at lower FE contrasts, with the highest suppression in participants without a switch even at 10% FE contrast.²³ 

This study aims to utilize eye-tracking technology to measure the fixation duration of the AE and FE in participants with existing amblyopia or previously treated amblyopia. The durations were computed across different stimulus regions arising due to the complementary parts of images being presented to each eye during viewing of dichoptic videos. Participants also viewed dichoptic stationary dots that were placed independently but coincidentally to each eye. The goal is to determine how viewing durations change at various FE contrasts across healthy controls, participants with anisometropia, and participants with strabismic amblyopia, and how they relate to suppression, stereo-acuity, and AE visual acuity deficit. We hypothesize that those with higher suppression and stereopsis deficits will fixate more in the regions seen by the FE only and the AE will be less likely to fixate in the regions visible to the AE only while viewing the dichoptic videos, particularly at high FE contrasts. We also hypothesize that strabismic participants who exhibit a fixation switch during viewing of dichoptic stationary dots will have less suppression and the AE will be more likely to fixate in the regions visible to the AE only during viewing of dichoptic videos compared with those that do not exhibit a fixation switch. We also aim to analyze eye deviation during the viewing of dichoptic stationary dots versus videos at varied FE contrasts. We hypothesize that the differing input to each eye in dichoptic viewing could potentially affect eye deviation, with the most significant effects expected at low FE contrasts. 

Methods

The experiment protocols complied with the tenets of the Declaration of Helsinki. They were approved by the Cleveland Clinic Institutional Review Board. Informed consent was obtained from the study participants and parents or legal guardians on behalf of minors/children. We recruited both children with amblyopia (n = 18) and children who had been successfully treated for amblyopia (n = 9) into the amblyopia cohort and eight healthy controls. All the participants had a comprehensive eye examination, and the clinical parameters were extracted at the time of the recordings (Table 1). Participants wore optical corrections per their cycloplegic refraction, with hyperopic corrections reduced symmetrically for the two eyes as clinically necessary. Participants were divided into four groups: the control group = 8, the anisometropia group = 8, the strabismus < 5Δ group (prism diopters) at distance measured by simultaneous prism cover test in current refractive correction = 9, and the strabismus > 5Δ group at distance = 10 (see Table 1). The severity of amblyopia was determined per the Pediatric Eye Disease Investigator Group studies²¹ as treated = 4, mild = 5, moderate = 12, and severe = 6. Subjects with amblyopia and those with treated amblyopia are collectively described as amblyopic participants. 

Table 1.

 

Demographic and Visual Function Data of Amblyopic Subjects

 

Table 2.

 

Pearson Correlation Results Between Visual Function and Time Spent by Fellow Eye and Amblyopic Eye in Different Regions of Dichoptic Video Stimuli

 

Visual Function Measurements and Analysis

Interocular Suppression

The Dichoptic Motion Coherence Test was used to assess suppression using an film patterned retarder (FPR) liquid crystal display (LCD) and polarized glasses.^24,25 Signal dots at 100% contrast were shown to the AE, whereas noise dots at varying contrast (100%, 90%, 75%, 62.5%, and 56.7%) were shown to the FE. The dichoptic motion coherence threshold at each contrast was determined by the proportion of signal to noise dots needed for accurate motion direction identification. Suppression was computed by fitting a third-order polynomial to the log of the number of signal dots required at each noise contrast level and calculating the area under the curve (AUC). 

Visual Acuity

Monocular visual acuity was assessed using the Early Treatment Diabetic Retinopathy Study (ETDRS) optotypes with crowding bars on a 32-inch LCD monitor (1920 × 1080 resolution, 120 hertz [Hz], and 111 cd/m² brightness) at 3.1 meters in a dark room.^22,23 Optotype sizes in arcmin were adjusted using an adaptive staircase procedure with six reversals. The log MAR value, representing the arithmetic mean of the reversals, was computed. 

Stereopsis

Stereopsis was measured in log arcsecs using the Titmus Stereoacuity Test, as described previously.^8,23 Participants with absent stereopsis were assigned a value of 3.85 log arcsecs. 

Eye Movement Recordings

We used a high-resolution EyeLink 1000 plus eye tracker to measure binocular eye positions.^23,26,27 Each subject's head was supported on a chinrest 84 cm from the monitor used for visual acuity testing. Interleaved polarization delivered different images to each eye. Monocular calibration and validation were done using a 5-point constellation of dots. For stimuli presentation and analysis, the AE was designated as the right eye for controls. For amblyopic participants, it was the eye with existing amblyopia, and for treated amblyopia, it was the same eye as before treatment. 

Recordings were obtained while viewing two different stimuli in a dark room as follows: 

• 1) Dichoptic stationary dot (0.5 degrees visual angle) that was presented independently but coincidently to each eye presented at the center of the screen.
• 2) Dichoptic videos (Fig. 1) with a mask of irregular blobs created using a specialized MATLAB program. The mask was applied to the AE (or right eye for controls) and the inverse to the FE (or left eye for controls). The blobs had Gaussian edges, causing overlapping areas to be viewed by both eyes at different contrast (transition region). It was necessary to piece the stimuli presented to the FE region and the AE region together to appreciate the videos as shown in Figure 1 and as reported in Li et al. 2015, Birch et al. 2019, and adopted for use by Xiao et al. 2023.^20,28,29

Figure 1.

 

(A) Shows a sample videoframe of the dichoptic stimulus presented to both amblyopic eye and fellow eye. The amblyopic eye stimulus contains the amblyopic eye (AE) region of the dichoptic stimulus that is visible only to the amblyopic eye. The fellow eye stimulus contains the fellow eye (FE) region of the dichoptic stimulus that is visible only to the fellow eye. The transition region indicates the regions of transition from an AE region to an FE region and is visible to both eyes albeit at different contrast levels. (B) illustrates how the image will appear to participants who are able to perceive the stimulus presented to the fellow eye and amblyopic eye simultaneously during dichoptic viewing (in other words, the participant must be able to combine the stimuli presented to the fellow eye [FE region] and amblyopic eye [AE region] together to appreciate the entire image). It also has eye positions overlaid for controls (green), anisometropic (yellow), < 5Δ (blue), and > 5Δ (red) strabismic participants on the stimulus as color filled circles to demonstrate amblyopic eye and fellow eye (black edges) fixations within that videoframe. Examples of fixations in synchronous time, where both eyes are in the same region, asynchronous time, where each eye is within the video frame but are in different regions, and asynchronous time out, where only one eye is within the video frame while the other eye is outside the video frame are marked with their respective signs.

The contrast of the target presented to the AE was kept at 100%, whereas the FE contrast varied from 100%, 50%, 25%, and 10% for stationary dots and movies. Four trials lasting 45 seconds each were done for dichoptic stationary dots and 4 trials lasting 60 seconds each were done for dichoptic movies at each FE contrast with less than a 1-minute break in between trials as needed. 

Eye movement analysis was performed using MATLAB. 

Dichoptic Stationary Dots

We used Engbert and Kliegl algorithms^30–32 to identify fast fixation eye movements in controls and amblyopic participants without nystagmus, and fast phases in those with nystagmus. A moving average filter (1.5-second window) was applied to remove fast eye movements. The filtered data were used to compute composite eye positions and calculate eye deviation. The 25th, 50^th, and 75th percentiles of eye deviation were computed for each subject across trials. We evaluated eye position traces of strabismic participants for fixation switches, where the AE fixates on the target previously fixated by the FE. 

Dichoptic Videos

We used the REMoDNaV algorithm,³³ to detect fixations (cream-colored sections), pursuits (brown-colored sections), and saccades (light-\ green sections) in eye movement data (Fig. 2) while participants watched dichoptic videos. The videos were each divided into four regions. The FE region (regions visible only to the FE), the AE region (regions visible only to the AE), the transition region (regions between the FE and the AE that are visible to both eyes), and outside region (regions outside the video frame), as depicted in Figure 1. Eye positions and movements were tracked within these regions throughout the 60-second clips. The regions of interest on the screen changed every 10 seconds as the blobs changed (6 masks for 10 seconds each for a period of 60 seconds). However, for analysis, the total data was grouped together, meaning the durations for each eye spent in various regions were summed across the entire 60-second video clip, not just for each 10-second segment. We combined fixation and pursuit durations to calculate the total time each participant's FE and AE spent in each region for each dichoptic video trial. 

Figure 2.

 

Epoch of eye positions acquired over a 10 second period while watching dichoptic video clip from one subject. The plot shows filtered gaze coordinates (black) measured in degrees and eye movement events segmentation with periods of fixation (cream), pursuit (brown), and saccades (light green), as detected using REMoDNaV algorithm.

We analyzed the temporal coordination of the FEs and AEs within each trial, classifying the time spent as follows (see Fig. 1): 

• 1. Synchronous durations: Both eyes were in the same region simultaneously (e.g. both in the FE region).
• 2. Asynchronous durations: The eyes were in different regions (e.g. one in the FE region, and the other in the transition region) at a given time.
• 3. Asynchronous out durations: One eye was within the video frame (the FE region, the AE region, or the transition region), whereas the other eye was outside the video frame at a given time.

We evaluated the %durations for each of the above 3-time classifications across four groups (the controls, anisometropic, strabismus < 5∆, and strabismus > 5∆) for each video clip. We also computed the %durations of each eye in the FE, AE, transition, and outside regions at varied FE contrasts. Because these durations did not pass the Levene's test for homogeneity of variance, we used the Brown-Forsythe test with Bonferroni correction for post hoc comparisons. Additionally, we assessed the influence of suppression, visual acuity, and stereo-acuity deficits on these durations using Pearson correlation analysis. 

We compared suppression in strabismic participants with and without fixation switch during viewing of dichoptic stationary dot stimuli using the Mann-Whitney U test. We calculated the %durations of the FEs and AEs in each of the four regions for dichoptic videos at varied FE contrasts. Participants were grouped into controls, anisometropic, strabismic with no manifest deviation (treated strabismus), and strabismic subjects with and without fixation switches during viewing of dichoptic stationary dots. For analysis, we combined treated strabismus and with fixation switch participants due to their lower interocular suppression compared with without fixation switch participants. Because the durations did not pass the Levene's test for homogeneity of variance, we used the Brown-Forsythe test with Bonferroni correction for post hoc comparisons. 

We computed eye deviation and its percentiles during dichoptic video viewing after removing fast eye movements, similar to the method used for dichoptic stationary dot stimuli for each subject. We then analyzed eye deviation percentiles for both stationary dot and video stimuli as a within-subject factor, comparing controls, anisometropic, strabismic < 5∆, and strabismic > 5∆ participants using repeated-measures ANOVA. Tukey's test was used for post hoc comparisons. 

We analyzed age differences between healthy controls and amblyopic participants using an unpaired t-test. Differences in visual acuity, suppression, and stereopsis across healthy controls, anisometropic, strabismus < 5∆, and strabismus > 5∆ groups were analyzed using 1-way ANOVA. Suppression levels in strabismic participants with treated strabismus, and those with and without a fixation switch were also analyzed using 1-way ANOVA. Post hoc comparisons were performed with Bonferroni correction. All statistical tests had a critical alpha value of 0.05. Statistical analysis was conducted using SPSS and GraphPad Prism. 

Results

There were no significant age differences between controls (13 ± 10.23) versus amblyopia (14.7 ± 11.79) participants (P = 0.12). The visual acuity (log MAR) was not different in anisometropic (0.43 ± 0.30), strabismic < 5Δ (0.35 ± 0.26), and strabismic > 5Δ (0.39 ± 0.43) participants (P = 0.12). There was no difference between the suppression (log AUC) across groups (anisometropia = 2.9 ± 0.29, strabismic < 5Δ = 2.4 ± 0.52, and strabismic > 5Δ = 2.9 ± 0.24 (P = 0.06). On the other hand, the stereo-acuity deficits (log arcsec) were least in anisometropic and worst in strabismus > 5Δ (anisometropia = 2.0 ± 0.19, strabismus < 5Δ = 3.0 ± 0.8, and strabismus > 5Δ = 3.6 ± 0.39, P< 0.0001). Post hoc comparisons showed differences between anisometropia versus strabismus < 5Δ (P = 0.0021), anisometropia versus strabismus > 5Δ (P < 0.0001), and strabismus < 5Δ and strabismus > 5Δ (P = 0.03). 

Temporal Coordination of Eyes During Viewing of Dichoptic Videos

We aimed to examine how the FE and AE attends to the dichoptic video stimuli as the FE contrast is lowered (Fig. 3). Significant differences were found in synchronous time across groups for 100% FE (F = 9.79, P < 0.001), 50% FE (F = 17.1, P < 0.001), 25% FE (F = 16.68, P < 0.001), and 10% FE contrasts (F = 16.6, P < 0.001) with healthy controls, anisometropic, and strabismic < 5∆ participants spending more time in synchronous viewing compared with strabismic > 5∆ participants. No significant differences were observed in asynchronous viewing time for most FE contrasts (100% FE [F = 0.82, P = 0.49], 50% FE [F = 0.51, P = 0.67], and 25% FE [F = 1.77, P = 0.17]), except at 10% FE contrast (F = 3.52, P = 0.03). Strabismic > 5∆ participants had significantly increased durations in asynchronous out times than other groups at all FE contrasts (100% FE [F = 6.93, P = 0.007], 50% FE [F = 7.87, P = 0.006], 25% FE [F = 11.18, P = 0.002], and 10% FE [F = 8.08, P = 0.002]). Overall, controls, anisometropic, and strabismic < 5∆ participants viewed the stimulus with both eyes throughout most of the trial, whereas strabismic > 5∆ participants mostly viewed the stimulus with one eye at a time. 

Figure 3.

 

Depicts the temporal synchronization of both eyes in terms of percentage durations spent in each of the three classifications at each FE contrast level (A) 100% FE, (B) 50% FE, (C) 25% FE, and (D) 10% FE, for each of the four groups. Controls (green), anisometropia (yellow), < 5Δ (blue), and > 5Δ (red). Asterisk (*) indicates statistically significant difference between groups evaluated at P < 0.05 for post hoc multiple pairwise comparisons with Bonferroni correction.

Eye Tracking Behavior While Viewing Dichoptic Videos Per The Type of Amblyopia

Figure 4 illustrates the fixation behavior of two anisometropic participants during dichoptic video viewing. Participant 7, with more suppression (log AUC = 3.45), had both eyes fixating in the FE region at 100%, 50%, and 25% FE contrasts, but shifted to the AE region at 10% FE. Participant 8, with less suppression (log AUC = 2.7), had both eyes fixating in the FE region at 100% FE, but shifted to the AE regions at 50%, 25%, and 10% FE. Synchronized time, indicated by square symbols, shows both eyes in the same region at a given time. 

Figure 4.

 

Illustrates fixation behavior of participants 7 and 8 (anisometropic amblyopia) during the viewing of dichoptic videos. The snapshots were taken at each FE contrast level and overlaid with amblyopic eye and fellow eye positions. Subject 7 is fixating with both eyes on the stimulus presented to the fellow eye (FE region) at 100%, 50%, and 25% FE contrasts but fixates on stimulus presented to amblyopic eye (AE region) at 10% FE contrast. Subject 8, however, fixates on the stimulus presented in the fellow eye (FE region) only at 100% FE contrast but fixates on the stimulus presented in the amblyopic eye (AE region) at 50%, 25%, and 10% FE contrasts. = synchronous time (amblyopic eye and fellow eye are in the same region). Fixation of the amblyopic eye within the amblyopic eye stimulus region is indicated by a red arrow pointing to the gaze position with a red shape surrounding the gaze position whereas fixation of the fellow eye within the fellow eye stimulus region is indicated by a black arrow pointing to the gaze position with a black shape surrounding the gaze position.

Figure 5 shows the fixation behavior of two strabismic < 5∆ participants during dichoptic video viewing. Participant 11, with more suppression (log AUC = 3.09), had their AE fixating in the FE or transition region but not in the AE region for all contrasts. Participant 12, with less suppression (log AUC = 2.43), had their AE fixating in the AE regions at 25% and 10% FE contrasts. Synchronized time is indicated by the square symbols whereas asynchronous time is indicated by the hexagon symbols. 

Figure 5.

 

Illustrates fixation behavior participants 11 and 12 (strabismic (< 5Δ) amblyopia participants) during viewing of dichoptic videos. Their fellow eye and amblyopic eye positions are overlaid on a video frame from each contrast of the dichoptic videos. Subject 11 has amblyopic eye fixing in the FE or transition region and not in the AE region for all 4 contrasts. Subject 12 had amblyopic eye fixing in the AE regions at 25% and 10% FE contrasts. = synchronous time (amblyopic eye and fellow eye are in the same region); = asynchronous time (amblyopic eye and fellow eye in different regions on the stimulus). Fixation of the amblyopic eye within the amblyopic eye stimulus region is indicated by a red arrow pointing to the gaze position with a red shape surrounding the gaze position whereas fixation of the fellow eye within the fellow eye stimulus region is indicated by a black arrow pointing to the gaze position with a black shape surrounding the gaze position.

Figure 6 shows the fixation behavior of two strabismic > 5∆ amblyopia participants during dichoptic video viewing. Participant 19, with less suppression (log AUC = 2.34), had the FE fixating in the FE region at 100% FE contrast while the AE was outside the video frame. At 50%, 25%, and 10% FE contrasts, the AE fixated in the AE regions, and the FE was in the out region. Participant 23, with greater suppression (log AUC = 3.50), consistently had the SE in the out region, while the FE fixated in the FE or transition regions for all FE contrasts. Asynchronized out time is indicated by triangle symbols around the eye position overlays. 

Figure 6.

 

Illustrates fixation switch behavior of subjects 19 and 23 (strabismic > 5Δ amblyopia participants) during viewing of dichoptic videos. Their fellow eye and amblyopic eye positions are overlaid on a video frame from each contrast of the dichoptic videos. Subject 19 had their fellow eye fixating in the FE region at 100% FE contrast and amblyopic eye outside the video frame. However, at 50%, 25%, and 10% FE contrasts, the amblyopic eye is fixating in the AE regions and fellow eye is outside the video frame. Subject 23 has the amblyopic eye fixating outside of the video frame while fellow eye is fixating in the FE region or transition region and not in the AE region for all 4 FE contrasts. Fixation of the amblyopic eye within the amblyopic eye stimulus region is indicated by a red arrow pointing to the gaze position with a red shape surrounding the gaze position whereas fixation of the fellow eye within the fellow eye stimulus region is indicated by a black arrow pointing to the gaze position with a black shape surrounding the gaze position. △ = asynchronous out (one eye on the screen and the other eye outside the video frame).

We analyzed the cumulative %durations spent by the FE and AE in FE, AE, transition, and out regions at various FE contrasts (Figs. 7A–D) with group-wise comparisons. 

Figure 7.

 

Illustrations of percentage durations both the fellow eye and amblyopic eye spent within each of the four regions (FE, AE, transitions regions, and outside) at each FE contrast level (A) 100% FE, (B) 50% FE, (C) 25% FE, and (D) 10% FE, for each of the 4 groups. Controls (green), anisometropia (yellow), < 5Δ (blue), and > 5Δ (red). Asterisk (*) indicates statistically significant difference between groups evaluated at P < 0.05 for post hoc multiple pairwise comparisons with Bonferroni correction.

At 100% FE contrast, no differences were found in the FE’s time spent in the FE, transition, and out regions. There was a borderline significance where the FE spent less time in the AE region for participants with coexisting strabismus compared with controls and anisometropic subjects. The AE showed significant differences in FE and out regions with borderline significance in the AE region. Strabismic > 5Δ participants spent less %duration in the FE, AE, and transition regions and more time in the out region compared with the other groups. 

At 50% FE contrast, we found no significant differences in %durations spent by the FE in all 4 regions. However, there were significant differences in the AE’s time spent in the transition region between strabismus > 5Δ and the other groups (controls [P = 0.003], anisometropia [P = 0.029], and strabismus < 5Δ [P = 0.013] groups). There were also significant differences between the AE’s time spent in the out region between strabismus > 5Δ and the other groups (controls [P = 0.050], anisometropia [P = 0.027], and the strabismus < 5Δ [P = 0.029] groups). 

At 25% FE contrast, the strabismic participants’ FEs spent more time in the FE region and less in the AE region compared with controls. The AE showed significant differences in FE and out regions, with strabismic > 5∆ participants spending less time in the FE region and more time in the out region. 

At 10% FE contrast, there were no significant difference in the %durations the FE spent in the AE, transition, or out regions, but a significant difference in the FE region and the AE showed significant differences in the FE, AE, and out regions. Strabismic > 5∆ participants spent less time in the FE and AE regions, with the AE spending more time in the out region compared with the other groups. 

For anisometropic and strabismic < 5∆ participants, at 100% FE contrast, the AE spends less time in the AE region compared with the controls. However, as the FE contrast is lowered, the time spent by the AE in the AE region was similar to that of controls. For strabismus > 5∆ participants, there is a consistent increase in the %duration the FE spends in the out region as the FE contrast is lowered from 100% to 10% (see Figs. 7A–D). This indicates a fixation switch behavior where the AE fixates on the stimuli while the FE is in the out Region. These eye tracking results suggest that lowering the FE contrast allows the SE to attend to the stimuli, helping to overcome suppression. Although differences in eye movement behavior were noted based on the type of amblyopia, it is crucial to understand how suppression influence eye movement patterns during the viewing of dichoptic videos. 

Eye Tracking Behavior While Viewing Dichoptic Videos Per the Functional Deficits (Table 2)

When evaluating these durations per stereopsis, we found that at 100% FE contrast, those with greater stereopsis deficit had shorter durations of FE fixing within the AE region and AE spent shorter duration fixing within the FE, AE, and transition regions with more time in the out region (Table 2). At 50% FE contrast, those with greater stereopsis deficit, the FE spent more time fixing within the FE region and less time in the AE region and the AE spent less time in the transition region with more time in the out region. At 25% FE contrast, those with greater stereopsis deficits, the FE spent more time in the FE region and less time in the AE region, whereas the AE spent more time in the out region with less time in the FE region. At 10% FE contrast, those with greater stereopsis deficits, the FE was in the FE region for a shorter duration and the AE was in for a longer period in the out region, with shorter durations in the FE and AE regions. 

When evaluating these durations per AE visual acuity deficit, we found that at 100% FE contrast, those with greater visual acuity deficit had significantly less duration, whereas the AE was in the transition region but had greater time when the AE was in the out region. At 50% FE contrast, we found that those with greater visual acuity deficit of the AE, the FE spent less time in the out region and the AE spent less time in the AE region. At 25% FE contrast, those with greater visual acuity deficit, the FE spent more time in the FE region and less time in the AE and out regions, whereas the AE spent less time in the AE and transition regions and more time in the out region. At 10% FE contrast, those with greater visual acuity deficit, the FE spent more time in the transition region, whereas the AE spent less time in the FE, AE, and transition regions but more time in the out region. 

When evaluating these durations per suppression, we found that at 100% FE contrast, individuals with greater suppression spent significantly shorter durations whereas the AE was in the AE region, with longer durations in the out region. As the FE contrast was reduced, no significant differences were observed in the durations of the FE and the AE in various regions relative to suppression levels. This suggests that decreasing FE contrast allows the AE to attend to stimuli presented in the AE region, providing evidence that such stimulus modification facilitates overcoming suppression. 

We grouped strabismic participants into those that had treated strabismus either with glasses or surgery = 5. For the remaining strabismic participants with manifest deviation, we evaluated eye position traces during viewing of dichoptic stationary dot to look for a fixation switch and grouped them into those with fixation switch = 9 where the AE is fixing on the target in at least one dichoptic stationary dot stimuli and without fixation switch = 5 where the amblyopic eye is not fixating on the target for any of the dichoptic dots. In agreement with prior studies,^22,23 we found that strabismic participants without fixation switch had greater suppression (3.4 ± 0.12) than those with a fixation switch (2.5 ± 0.3), treated strabismus (2.5 ± 0.6), anisometropic (2.9 ± 0.2), and controls (2.3 ± 0.17, F = 6.2, P = 0.01). Post hoc comparisons demonstrated differences between controls versus anisometropic (P = 0.2), control versus without fixation switch (P < 0.0001), treated strabismus versus without fixation switch (P = 0.01), and without versus with fixation switch (P = 0.001). Thus, we compared total %durations spent by FEs and AEs in different regions in controls, anisometropic, strabismus participants with treated strabismus, and with fixation switch (combined together) and those without fixation switch as FE contrast is varied (Table 3). We find that in all strabismic participants regardless of the fixation switch, the FE and AE spent less time in the AE region at 100% FE contrast compared with controls and anisometropic participants. At 50% FE contrast, the FE spent more time fixing in the FE region with less time in the AE region in those without fixation switch. At 25% FE contrast, the FE fixated more in the FE region in those without fixation switch. In addition, in participants without fixation switch, the AE spent less time in the AE region even at 10% FE contrast compared with the other groups. Thus, fixation switch behavior observed during viewing of dichoptic stationary dots are associated with the durations the AE and FE fixates on in different regions while passively viewing dichoptic videos. 

Table 3.

 

Percentage Durations of Fellow and Amblyopic Eye in Different Regions of the Dichoptic Video Stimuli Across Groups

 

Eye Deviation Changes During Viewing of Dichoptic Stationary Dots and Videos

This study compared eye deviation changes in different groups while viewing a dichoptic stationary dot stimulus and dichoptic videos. Figure 8 shows cumulative sum histograms and Table 4 show percentiles of eye deviation at various FE contrasts for controls, anisometropic, strabismic < 5∆, and strabismic > 5∆ groups. Controls had the smallest eye deviation, whereas strabismic > 5∆ participants had the largest deviation shown as the rightward shift of the cumulative sum histograms. No significant differences were seen within controls, anisometropic, and strabismic < 5∆ groups when comparing dichoptic stationary dots versus videos. However, strabismic > 5∆ participants showed increased eye deviation during viewing of dichoptic videos than stationary dots at all FE contrasts. 

Figure 8.

 

Depicts cumulative sum histograms of eye deviations (degrees) at each contrast level (FE 100%, FE 50%, FE 25%, and FE 10%) for each group for both stationary dot (dotted lines) and dichoptic videos (solid lines) trials. (A) Controls (green), (B) anisometropia (yellow), (C) < 5Δ (blue), and (D) > 5Δ (red). The colors are shaded to show a reduction in FE contrasts within each group. DM = dichoptic video stimuli; DG = dichoptic stationary dot stimuli.

Table 4.

 

Eye Deviation Percentiles for Free Viewing of Dichoptic Video Stimuli and Fixation on Dichoptic Stationary Dot

 

Discussion

Newer dichoptic treatments aim to reduce suppression by presenting different images to each eye. One approach involves reducing the contrast of the FE stimuli while keeping the AE at 100% contrast.^22,34–40 Despite theoretical benefits, dichoptic treatment outcomes are mixed, indicating a need for further research.^16–18 Eye tracking is valuable for quantifying visual attention,^41–43 but most studies have been limited to simple fixation tasks.^22,23,44 These studies do not fully capture the complexity of viewing dichoptic videos as used in amblyopia therapies. The current study addresses these gaps by using dichoptic videos and recording eye movements to obtain “ground truth” visual attention data, closely approximating naturalistic viewing scenarios. We found systematic differences in binocular coordination among the control, anisometropic, and strabismic groups as the FE contrast was reduced. Participants with strabismus > 5Δ exhibited more asynchronous viewing compared with other groups. For anisometropic and strabismic amblyopia < 5Δ, the AE’s time in the AE region increased as the FE contrast was lowered. There was also a consistent increase in the %duration spent by the FE in the outside region for strabismic > 5Δ as the FE contrast decreased, indicating a fixation switch to the AE. 

Our study found that eye tracking patterns during passive viewing of dichoptic videos are influenced by functional deficits. Individuals with greater suppression, visual acuity deficits in the AE, and stereoacuity deficits showed different eye tracking patterns than controls. At 100% FE contrast, those with greater suppression, visual acuity deficits, and stereoacuity deficits had the AE less in the AE region and more in the out region. As FE contrast was reduced, differences in eye movement patterns were less influenced by suppression levels, consistent with previous studies suggesting that lowering the non-AE’s contrast can temporarily disrupt established suppression patterns.^45,46 Notably, even when suppression differences were insignificant, individuals with greater visual acuity and stereoacuity deficits still demonstrated less fixation in the AE region. In our study, stereopsis was measured using the Titmus Fly Test, which could potentially give false results due to monocular and non-stereoscopic binocular cues.⁴⁷ However, consistent with other studies,^2,48,49 participants with anisometropic amblyopia in our cohort had better stereopsis compared with those with strabismus < 5Δ, with the poorest stereopsis found in participants with strabismus > 5Δ. Thus, the correlation analysis between visual function deficits and eye movement patterns highlights the importance of considering individual factors, such as suppression, stereoacuity, and visual acuity deficits of the AE, along with the clinical type of amblyopia, which may impact the ability of the amblyopic eye to attend to the dichoptic videos. 

Prior studies from our laboratory have shown that dichoptic stationary dots can result in fixation switch and change the eye deviation. Thus, we investigated whether fixation switch behavior, observed under varied FE contrasts with dichoptic stationary dots, is associated with greater visual attention of the AE during viewing of dichoptic videos. Strabismic participants’ AE spent less time in the AE region at 100% FE contrast compared with controls. However, at 10% FE contrast, strabismic participants with treated strabismus or with fixation switch showed comparable AE region time to controls. Conversely, strabismic participants without fixation switch during viewing of dichoptic stationary dots spent less time in the AE region and more time in the out region across all FE contrasts. This suggests that the ability to alternate fixation between eyes may predict a subject’s capacity to overcome suppression and attend to stimuli presented to the AE, potentially improving visual function recovery and leading to more favorable dichoptic treatment outcomes.^2,45–53 

Previous studies have suggested that strabismic amblyopia may be less responsive to dichoptic treatments, particularly for improving stereopsis, compared to anisometropic amblyopia.^50–52 Our study showed significant shifts in eye deviation as the FE contrast decreased during dichoptic video viewing, especially in the strabismic group. This can influence the effectiveness of dichoptic treatments by inducing fixation disparity, potentially leading to diplopia⁵³ and limiting visual acuity and stereopsis recovery. Thus, eye-tracking data highlight the impact of strabismus severity and associated functional deficits on the binocular integration of dichoptic stimuli. Further, monitoring ocular alignment during dichoptic treatment protocols involving contrast rebalancing is essential. 

Despite the valuable insights, this study has several limitations. The sample size was relatively small, restricting the generalizability of our findings. This smaller cohort size precluded detailed analyses of how suppression, stereoacuity, and visual acuity deficits within each clinical type of amblyopia influence treatment response. Additionally, the study was observational and conducted at a single time point. Future studies should involve larger cohorts and longitudinal designs to observe changes over time. 

Moreover, this study involved passive viewing rather than active training with dichoptic stimuli. Whereas it provides insights into fundamental oculomotor and perceptual factors, dynamic binocular combination and visuo-motor adaptations may differ when participants are actively engaged in dichoptic tasks.⁵¹ 

Despite limitations, these findings offer significant potential to enhance amblyopia treatment practices, improve patient outcomes, and drive further research into neural mechanisms and optimal therapies. Future studies should integrate behavioral measures of visual function with eye tracking during dichoptic viewing to create personalized treatment protocols, potentially incorporating contrast modulation and engaging video stimuli that require active subject participation. In conclusion, whereas dichoptic treatments show promise, mixed results underscore the need for ongoing research. Advanced methods like dynamic dichoptic video stimuli and eye tracking could yield deeper insights and help develop more effective amblyopia therapies. 

Acknowledgments

Supported by the NEI T32: 5 T32 EY 24236-4 (J.M.), VA SPiRE I21RX003878-02 (A.G.S. and F.G.), Blind Children's Foundation grant (F.G.), and Research to Prevent Blindness Disney Amblyopia Award (F.G.), CWRT CTSC Pilot Grant Program (F.G.), Cleveland Clinic RPC Grant (F.G.), Lerner Research Institute Artificial Intelligence in Medicine (F.G.), Departmental Grants from Research to Prevent Blindness, Unrestricted Block Grant CCLCM, NIH-NEI P30 Core Grant Award, and the Cleveland Eye Bank. 

Disclosure: I.M. Quagraine, None; J. Murray, None; G.B. Cakir, None; S.B. Beylergil, None; A. Kaudy, None; A.G. Shaikh, None; F.F. Ghasia, None