Pilot comparison of monocular and binocular modes of virtual reality field testing in strabismus

Abstract

Virtual reality field (VRF) perimetry is a promising new tool for visual field assessments, offering both monocular and binocular testing modes. We aimed to examine whether ocular misalignment, known as strabismus, influences VRF outcomes between monocular and binocular testing modes, which is currently unknown. Children with non-amblyopic horizontal strabismus were enrolled and completed VRF testing (Olleyes VisuALL) in either the game-based or standard Humphrey visual field (HVF)-equivalent algorithm. Within their respective algorithms, participants completed tests in both monocular and binocular modes. Analyzed parameters included mean deviation (MD), pattern standard deviation (PSD), and foveal sensitivity (FS). Additionally, a masked ophthalmologist graded whether the VRF was abnormal. Twenty-two children (44 eyes) were enrolled with a mean age of 12.1 ± 2.7 years. Regarding MD, participants who took the game-based algorithm compared to the HVF-equivalent yielded better scores (P = 0.048). For PSD, older participants exhibited better scores than younger participants (P = 0.038). Binocular and monocular mode testing were similar across all parameters, and yielded equal rates of abnormal test results (22.7% vs. 22.7%; P = 1.00). Overall, non-amblyopic horizontally strabismic children demonstrated comparable VRF outcomes in both monocular and binocular testing modes, suggesting similar utility between exam options.

Introduction

Virtual reality field (VRF) testing has emerged as a promising technology for pediatric and adult eye care specialists in the evaluation of visual field (VF) defects^1–3. Requiring only a compact headset worn over a patient’s eyes, VRF perimetry offers a more affordable and physically accessible alternative to traditional perimeters that involve large tabletop machinery⁴. Yet, although this new tool has shown strong potential, it still possesses nuances that require further investigation to understand their clinical significance⁵. For instance, VRF platforms can be equipped with various features, such as game-based (targeted toward children) and Humphrey visual field (HVF)-equivalent (standard test) algorithms, as well as the option to conduct exams in either monocular or binocular modes^6–8. Recent studies have shown that game-based VRFs may enhance engagement in children, while game-based and HVF-equivalent VRFs can yield outcomes comparable to tabletop HVF exams in glaucoma^5–8. However, little to no research has been published regarding the implications of testing in binocular mode (common default setting in VRF systems), as opposed to monocular mode.

The current standard for VF testing, the tabletop HVF exam, typically assesses each eye individually in a monocular fashion^1–3. This format is analogous to monocular mode VRF exams where only one eye is examined during a session. In contrast, binocular mode VRF perimetry examines both eyes in a single testing session by sending individual stimuli to each eye in an alternating sequence^1,6. In theory, outcomes from monocular and binocular testing should be equal; however, this may not hold true for patients with strabismus, given the complexities that ocular misalignments potentially add to binocular examinations⁹. This is an important consideration for patients with undiagnosed strabismus, particularly in conditions where it commonly coexists, like glaucoma and neuro-ophthalmic disorders^10–14. To that end, evaluating binocular versus monocular VRF testing modes in strabismic eyes may provide valuable insights into whether monocular testing should be required in those with a higher risk of undiagnosed strabismus.

In this study, we evaluate outcomes between monocular and binocular mode VRF perimetry in a cohort of children with non-amblyopic (normal vision) horizontal strabismus. Additionally, we assess how game-based and HVF-equivalent algorithms influence results.

Methods

Ethics statement

This prospective study was approved by the Institutional Review Board of Duke University Medical Center and was conducted in accordance with the tenets of the Declaration of Helsinki. The parent or guardian of each child provided written informed consent, and children provided verbal informed assent, with those 12 years and older also providing written informed assent.

Enrollment criteria

Inclusion criteria were pediatric patients older than 8 years of age with esotropia or exotropia and near perfect corrected vision (20/25 visual acuity or better in both eyes) presenting for standard-of-care eye examinations to Duke Eye Center between January 2024 and January 2025. Patients were excluded if they had amblyopia, vertical strabismus, nystagmus, any anterior or posterior segment pathologies, developmental delays, or pathologies that could mimic strabismus, such as macular drag or angle kappa.

Virtual reality field testing platform

VRF assessments were conducted with VisuALL (Olleyes Inc, Summit, NJ), a virtual reality system designed specifically for ophthalmic vision assessments. This system offers two algorithms for VRF testing: a game-based algorithm and an HVF-equivalent algorithm, the latter designed to mirror standard tabletop HVF exams. Both algorithms can be administered in either monocular or binocular modes, as VRF headsets include separate screens for each eye without cross-visibility. During monocular mode testing, both eyes can remain open, but only one eye is assessed while the fellow eye does not receive stimuli during an exam. Consequently, eye patching, regularly used in standard tabletop HVF tests, is not required. In binocular mode, both eyes remain open as inputs are individually presented to each eye in a 1:1 alternating pattern; a given stimulus is never displayed to both eyes or screens simultaneously. As a result, fusion is neither required nor elicited during testing, and therefore fusion cues are not a part of the VRF perimetry system. In the present study, all exams were completed using the 24 − 2 VF pattern, size 3 stimulus, and Adaptive Visual Algorithm-Fast setting. Testing was performed with the patient’s own corrective spectacles in place, if appropriate.

VisuALL eye-tracking technology

The VisuALL headset used in this study was equipped with infrared-based eye-tracking technology that checks the actual position of pupillary fixation a few milliseconds before and during stimulus presentation¹. If the user’s gaze shifts within 15 degrees of central fixation during testing, the device adjusts its image to recenter fixation. However, if the device is unable to compensate for the patient’s eye movements or the gaze deviates beyond the allowable fixation range, a fixation loss is recorded.

Of note, unlike most HVF tabletop exams, our VRF testing system does not use blind spot mapping to identify fixation losses. Although the VisuALL system is capable of displaying the physiological blind spot, the grid used in this study does not include test stimuli directly over the blind-spot region, as the manufacturers deemed it redundant to perform additional alignment checks such as the Heijl-Krakau method (derived from non-published communications with Olleyes). However, the locations surrounding the expected blind spot area are actively tested to identify pathologic enlargement of the blind spot area. With that said, blind spots still appear on visual field exam printouts.

Grouping protocol

Upon enrollment, patients were randomly assigned to test with either the game-based or HVF-equivalent algorithm. Within their respective algorithms, patients were asked to complete VRF testing in both monocular and binocular modes, and they were further randomized as to the testing order between the two modes. This protocol is illustrated in Fig. 1.

Fig. 1.

Flow diagram of study grouping protocol. Twenty-two children with non-amblyopic horizontal strabismus (10 esotropia [ET], 12 exotropia [XT]) were randomized to perform virtual reality field testing in either game-based or Humphrey visual field (HVF)-equivalent algorithms. Within each algorithm group, they were further randomized to begin with either monocular or binocular modes; all participants completed both test modes.

Data collection and analysis

Background information was collected on age, gender, strabismus type (esotropia vs. exotropia), fixating eye, and angle of ocular deviation. VRF outcome measures included mean deviation (MD), pattern standard deviation (PSD), and foveal sensitivity (FS). Additionally, the presence or absence of VF abnormality was determined by a masked ophthalmologist (BW) experienced in VF grading, who was masked to patient and testing characteristics. Presence of VF abnormality was defined as 3 clustered points that were more than 5 decibels below normal for age-matched controls, as performed in past VRF studies^6,7. Of note, each VRF test was assessed independently, rather than in pairs from a single individual, so no consideration was given to the relationship between eyes of the same child.

Statistical analysis was conducted using R (version 4.3.2, macOS). A linear mixed-effects model was employed to assess how MD, PSD, and FS were influenced by test algorithm (game-based vs. HVF-equivalent), test mode (binocular vs. monocular mode), test order (first vs. second), strabismus type (esotropia vs. exotropia), absolute angle of ocular deviation, fixating eye (fixating, non-fixating, or alternating fixation), and age. To account for correlation among repeated measurements within individuals and eyes, we included random intercepts, without random slopes, for subject and for eye. Chi-squared tests were used to compare the distributions of normal and abnormal VF outcomes across different testing variables. Values represent mean ± standard deviation unless otherwise stated.

Results

Patient and virtual reality field testing characteristics

During this study period, a total of 22 patients (44 eyes) were enrolled. The game-based algorithm cohort (n = 10 patients) included 5 patients with esotropia (inward misalignment) and 5 with exotropia (outward misalignment), while the HVF-equivalent algorithm cohort (n = 12 patients) included 5 patients with esotropia and 7 with exotropia (Fig. 1). The mean age of participants was 12.1 ± 2.7 years (Table 1). The mean ocular deviation at near and distance was 19.3 ± 10.6 and 17.3 ± 9.9 prism diopters, respectively. Test duration for monocular and binocular modes was 3.28 ± 1.35 min and 3.27 ± 1.35 min, respectively (P = 0.94); for game-based and HVF-equivalent algorithms it was 4.55 ± 0.80 min and 2.20 ± 0.52 min, respectively (P < 0.001, Table 2). Visual fixation losses for tests taken in monocular and binocular modes were 1.97 ± 4.11 and 1.77 ± 3.11, respectively (P = 0.80).

Table 1.

Patient baseline characteristics (n = 22).

Characteristic	Value
Age (years)	12.1 ± 2.7
Sex (n)
Male	8
Female	14
Strabismus (n)
Esotropia	10
Exotropia	12
Deviation at near (prism diopters)	19.3 ± 10.6
Esotropia	24.1 ± 10.3
Exotropia	15.3 ± 9.4
Deviation at distance (prism diopters)	17.3 ± 9.9
Esotropia	14.2 ± 9.9
Exotropia	20.0 ± 9.2

Values represent mean ± standard deviation unless otherwise stated.

Table 2.

Virtual reality field testing metrics across groups.

Test category	Group	MD (dB)	PSD (dB)	FS (dB)	Duration (minutes)
Mode	Monocular	− 0.12 ± 1.39	2.39 ± 1.15	33.70 ± 1.34	3.28 ± 1.35
	Binocular	− 0.50 ± 1.39	2.65 ± 1.5	33.57 ± 1.35	3.27 ± 1.35
Algorithm	Game-based	0.22 ± 1.03*	2.21 ± 0.84	33.67 ± 1.56	4.55 ± 0.80*
	HVF-equivalent	− 0.74 ± 2.03*	2.78 ± 1.6	33.61 ± 1.13	2.20 ± 0.52*
Order	First	− 0.23 ± 1.4	2.46 ± 1.21	33.6 ± 1.66	3.32 ± 1.37
	Second	− 0.39 ± 1.95	2.58 ± 1.47	33.77 ± 1.22	3.28 ± 1.37
Strabismus type	Esotropia	− 0.65 ± 2.00	2.58 ± 1.57	33.73 ± 1.03	3.48 ± 1.35
	Exotropia	− 0.02 ± 1.36	2.56 ± 1.29	33.64 ± 1.35	3.10 ± 1.33
Fixation	Fixating eye	− 0.74 ± 2.08	2.45 ± 1.22	33.17 ± 1.27	2.94 ± 0.95
	Non-fixating eye	− 0.34 ± 2.52	2.18 ± 0.77	33.83 ± 0.94	2.95 ± 0.93
	Alternate fixation	− 0.22 ± 1.46	2.6 ± 1.46	33.69 ± 1.43	3.39 ± 1.46

Asterisk (*) represents comparisons from multivariate analysis that were significantly different between marked groups within a test category; P < 0.05. Values represent mean ± standard deviation.

HVF = Humphrey visual field; MD = Mean deviation; PSD = Pattern standard deviation; FS = Foveal sensitivity.

Virtual reality field testing outcomes following multivariate analysis

Following linear mixed-effects model analysis for MD, test algorithm demonstrated statistically significant effects on outcome (Table 2). Specifically, those who performed the game-based algorithm compared to the HVF-equivalent, exhibited better (less negative) MD scores (P = 0.048). Of note, differences in MD with regard to age approached significance (P = 0.058), with older patients yielding better MD scores. Regarding PSD, age showed a statistically significant association (P = 0.038), with older participants producing better (lower) PSD values. Other variables did not demonstrate significance, although test algorithm showed borderline significance (P = 0.11), exhibiting a trend toward better PSD scores in the game-based group. FS was not significantly different across testing groups. Test mode (monocular vs. binocular), test order (first vs. second), strabismus type (esotropia vs. exotropia), angle of deviation, and fixating eye (fixating vs. non-fixating vs. alternate fixation) did not show any significant differences in MD, PSD, or FS.

Abnormal status of virtual reality field test across testing groups

A total of 88 VRF exams, each corresponding to a single eye, were evaluated for normal or abnormal outcomes, with 44 conducted in monocular mode and 44 in binocular mode. Of these tests, 20 (22.7%) were graded as abnormal by a masked ophthalmologist (Table 3). Chi-squared comparisons of abnormal test distributions for algorithm (game-based [12.5%] vs. HVF-equivalent [31.2%]; P = 0.07), test order (first [29.5%] vs. second test [15.9%]; P = 0.20), strabismus type (esotropia [25.0%] vs. exotropia [20.8%]; P = 0.84), and mode (monocular [22.7%] vs. binocular [22.7%], P = 1.00); did not yield any significant differences (Table 3).

Table 3.

Distribution of abnormal visual field outcomes across test conditions.

Test category	Group	Total tests	Abnormal tests (%)	P value
Algorithm	Game-based	40	5 (12.5%)	0.07
	HVF-equivalent	48	15 (31.2%)
Order	First	44	13 (29.5%)	0.20
	Second	44	7 (15.9%)
Strabismus type	Esotropia	40	10 (25.0%)	0.84
	Exotropia	48	10 (20.8%)
Mode	Monocular	44	10 (22.7%)	1.00
	Binocular	44	10 (22.7%)

Percentages represent the proportion of abnormal results out of the total number of tests conducted within a group.

HVF = Humphrey visual field.

Discussion

Although VRF perimetry shows promise for routine clinical use, its broader integration requires a thorough understanding of the novel features that distinguish it from standard tabletop HVF perimetry^5–8. It is important to evaluate whether binocular mode can be used in the setting of strabismus, which is known to regularly co-occur in glaucoma and neuro-ophthalmic disorders^10–14. Notably, studies have shown that strabismus is common in adult and childhood glaucoma patients prior to surgery, and its prevalence rises in both groups after glaucoma surgery^13,14. In the present study, binocular and monocular VRF testing modes demonstrated comparable outcomes in assessing children with non-amblyopic (normal vision) horizontal strabismus.

With respect to VFs in strabismus, past studies reported normal tabletop monocular VFs in both non-amblyopic and amblyopic subjects^15–17. In contrast, tabletop binocular VFs exhibited blind spots and suppression patterns in one or more eyes as compensatory mechanisms to avoid double vision^15–17. That said, it is important to distinguish that VRF testing in binocular mode actually examines monocular VFs rather than binocular VFs^1,6. Recall that VRF tests provide each eye with a separate screen so that during binocular mode testing, each stimulus is displayed individually to one eye at a time, rather than to both eyes simultaneously. Accordingly, all VRFs in our cohort were expected to be normal, and scoring discrepancies were attributed to test-taking issues rather than true VF defects. However, we wished to verify that in the setting of real-world clinical VRF testing, strabismus would not interfere with binocular mode assessments.

Regarding MD, our multivariate analysis identified the use of game-based rather than HVF-equivalent algorithms, as a predictor of better scores (P = 0.048), while older children demonstrated borderline significance (P = 0.058) in producing better MD outcomes. PSD outcomes were significantly better in older participants (P = 0.038). Similar age-related performance patterns were documented by Alvarez and by Griffin, reflecting the challenges younger children potentially face when taking VF examinations^3,6. This phenomenon presumably stems from reduced attention span and focus among younger patients, which could hinder test proficiency. Given this trend, better MD values with the game-based algorithm highlight an important consideration: younger children may perform better using game-based VRF compared to HVF-equivalent VRF algorithms. This is further supported by the fact that children have a strong preference for game-based VRFs, as documented by both Wang and Groth, which could boost test engagement^7,8. In terms of FS, results were similar for all groups, which was expected as FS was tested in the same way regardless of mode or algorithm. Importantly, there were no significant differences between binocular and monocular modes for MD, PSD, or FS, suggesting similar performances across both settings.

Among variables analyzed for associations with abnormal test results, none reached statistical significance in our cohort. Only algorithm choice approached statistical significance with HVF-equivalent tests displaying a higher rate of abnormal outcomes compared to game-based tests (31.2% vs. 12.5%, respectively; P = 0.07). Although not statistically significant, this may further support the notion that some children may perform better with game-based algorithms. On the other hand, monocular and binocular modes produced the exact same proportion of abnormal results (22.7% vs. 22.7%; P = 1.00), suggesting both may have been similarly effective testing methods for our non-amblyopic strabismic cohort.

Concerning other key findings, esotropia and exotropia did not yield significantly different outcomes in any study measures. Other related studies reported similar findings with respect to monocular tabletop VFs^15–17. In binocular tabletop VF testing, however, esotropic and exotropic eyes exhibited different degrees of nasal or temporal scotomas, depending on the type and angle of strabismus^15–17. Yet, as previously mentioned, tabletop VFs taken with binocular viewing are not particularly relevant to VRF binocular mode testing because the latter system examines each eye’s monocular VF separately, albeit in one testing session. We also recorded nearly identical test durations between monocular and binocular modes, so time should not influence decisions regarding mode selection. Conversely, the game-based algorithm did have a significantly longer test duration compared to the HVF-equivalent (4.6 vs. 2.2 min, respectively; P < 0.001). This has also previously been documented in past studies and presents one of the considerable trade-offs in algorithm choice^7,8.

Lastly, visual fixation losses were also not significantly different between binocular and monocular modes, suggesting that ocular misalignments were compensated for during VRF testing in both modes. We credit the consistency in fixation losses between testing modes to VisuALL’s advanced eye-tracking software. It is plausible that the system’s capacity to realign displays to an eye’s directional gaze before presenting signals was able to minimize or eliminate the “correction” needed for a deviated eye. In addition to misalignment, some studies have reported that strabismic patients experience increased wandering and motor instability in their non-fixating eye during binocular viewing^18,19. Nevertheless, our report suggests that the VRF system used in the present study may effectively manage these potential issues in strabismic patients when both open eyes are tested in binocular mode.

Our results should be considered in light of several limitations. Our relatively small study cohort may limit the generalizability of our findings and the statistical power of our results. Furthermore, only horizontal strabismus patients were included, and most of them had moderate angles of deviation. Therefore, we cannot apply our findings to patients who experience vertical or large-angle eye deviations. Additionally, our findings do not include comparisons to tabletop HVF outcomes, as obtaining three separate tests from pediatric patients in one clinic visit was not feasible from a stamina point of view.

Another limitation to our generalizability was the inclusion of non-amblyopic otherwise normal eyes, which were expected to have full VFs. Also, because of the unique design of the game-based VRF algorithm, false positives and false negatives were not comparable to those of the HVF-equivalent algorithm and, thus, were not included in our analysis. We used a single VRF headset and system to obtain our findings. Although this may limit generalizability to other devices, the Olleyes VisuALL model is designed to replicate standard VF assessment mechanisms, likely comparable to those used in other VRF systems^1–8.

We also want to clarify that our statistical comparisons between binocular and monocular modes report the absence of a significant difference in outcomes for MD, PSD, and FS, but they do not confirm equivalence between testing options. We also acknowledge that our VisuALL system is not equipped with blind spot detection and uses a different algorithm for fixation losses. While this lack of blind spot detection and monitoring differs from standard HVF perimetry, we hope that the integrated eye-tracking technology in our system helped to compensate for this limitation and maintain the reliability of testing outcomes. Lastly, because this study included only pediatric participants, the greater variability typically seen in pediatric perimetry testing and retesting may limit our ability to isolate the pure effect of strabismus on VRF performance.

In this pilot study, children with non-amblyopic horizontal strabismus performed similarly in VRF testing between binocular and monocular modes, suggesting that binocular mode may suffice for most VRF perimetry, even in the setting of moderate ocular misalignments. This finding may provide reassurance to eye care providers, particularly in adult clinics where strabismus may not be routinely screened for, such as in adult glaucoma clinics^10–14. Regarding the use of monocular VRF testing, this mode remains valuable for enhancing efficiency in cases where only one eye requires perimetry, such as in patients with extremely low vision or blindness in their fellow eye. Further study of VRFs in strabismic eyes with known VF defects may shed additional insights into the most effective VRF testing conditions for this population.

Literature search

PubMed was searched in May of 2025 for English-language results under the following criteria: visuall OR virtual reality field AND binocular. A total of 65 results appeared; no studies compared outcomes between testing in monocular and binocular modes in VRFs, let alone in strabismic patients.

Author contributions

Study conception and design: J.L., Z.S., B.W., R.N., M.E., S.F.F. Data collection: Z.S., M.G., N.L.C., Q.W.W., A.F., R.N., M.E., S.F.F. Data analysis and statistical expertise: J.L., Z.S., B.W., S.F.F. Manuscript writing: J.L., B.W., S.F.F. All authors read and approved the final manuscript.

Data availability

The data used in this manuscript is not publicly available due to the protection of personal patient information. However, data may be shared upon reasonable request from clinicians or researchers affiliated with recognized academic or medical institutions. Please email Sharon Freedman, MD, at sharon.freedman@duke.edu for any related inquiries.

Declarations

Competing interests

Sharon Freedman has a device on loan from Olleyes, Inc, the maker of the virtual reality field headset used in this study. Jeffrey Liu, Zara Saleem, and Rizul Naithani were funded as research coordinators through support from LC Industries and Saving Kids’ Sight. Bo Wang is funded by a Knights Templar Eye Foundation career starter grant. The other authors do not have any competing interests to declare.