Recent Conference Abstracts

2019

Dylan Rose and Peter J. Bex (2019) “Re-re-considering Yarbus: Predicting observer “taskiness” from eye movement patterns.” Submitted to Vision Sciences Society. Abstract

Greene, Liu and Wolfe reported a negative result for identifying task using eye movements (2012). Others have challenged this finding, often by substituting dynamic for static features (e.g. Boisvert & Bruce, 2016). We propose instead to map eye movements onto a latent dynamic manifold with connected, overlapping regions of related task effects. This manifold’s structure is defined using features extracted from a variational auto-encoder trained on multi-dimensional time-series of values for two different salience maps at observer fixation locations (GBVS: Harel et al., 2007; modified semantic similarity: Rose & Bex, 2018). This approach helps reconcile Greene’s findings with others’ by treating eye movement as less “task driven” than “tasky”: affected by current task while remaining influenced by those performed in the past. We therefore hypothesized that observers’ sequential performance of different tasks–as in Greene and colleagues’study–may deflect behavior across manifold regions, decreasing classifiability. Performing the same task should result in more regular trajectories within a manifold, increasing classifiability. We tested this hypothesis on three datasets first by comparing classification accuracy per task/trial using twenty-five manifold features and a SVM classifier. Second, we compared sequential determinism and entropy in trial-to-trial sequences of feature distances to separating hyperplanes using multidimensional recurrence quantification analysis (Wallot, 2016). For two sets (Koehler et al.,2014; Borji&Itti, 2014,Expt.2), subjects always performed the same task. In the third (created in-lab), tasks were randomized across trials. Consistent with our predictions, classification performance was significantly above chance and equivalent to that reported elsewhere for random order studies (Borji & Itti, 2014:24.2%, ours:24.6%, chance:14.2%; Boisvert & Bruce,2016:53.4%; ours:51%, chance:33.3%) but not the third (ours:39.5%, chance:33.3%). Determinism was also significantly higher and entropy significantly lower for the random task order experiment than for the others (Kruskal-Wallis, p<0.001). Our method therefore both retains high discrete classification peformance while connecting task effects to other efforts concerning serial dependencies (Fischer & Whitney,2014).

Concetta F. Alberti, Anna Kosovicheva and Peter J. Bex (2019) “Measuring the contrast sensitivity function in the periphery via reflexive saccades”. Submitted to Vision Sciences Society. Abstract

The visual system’s response to complex stimuli can be estimated from the contrast sensitivity function (CSF), which varies across the visual field. Measuring this field of contrast sensitivity (Watson, 2018) has significant value, but requires too many measurements for practical applications. However, simplification may be possible via a spatial scaling factor (Strasburger et al, 2011). We measured detection thresholds in 6 normally-sighted observers with Laplacian of Gaussian (LoG) stimuli at 3 eccentricities (4°, 9°, 19°) and 8 angular locations in two 8AFC tasks. In a behavioral task, observers reported the location of the target via button press; in a continuous- eye movement task, observers foraged for targets and each saccade landing point was scored as a directional response and became the fixation for the next trial. For faster evaluation, two points on the CSF were measured and then the contrast sensitivity function was fit with an asymmetric log-parabola (Watson & Ahumada, 2005) with upper and lower bandwidths fixed (Chung & Legge, 2016). First, visual acuity was measured at each location, then contrast sensitivity was measured 1.16 log10 units below the measured acuity. There were no significant differences between thresholds measured with behavioral and eye-movement methods, suggesting that these methods are interchangeable and the ease of comprehension of the eye movement task may favor its use in naive populations. Log Acuity decreased linearly with eccentricity (p<0.001). Peak contrast sensitivity was constant as a function of eccentricity (p>0.05), which we attribute to the scaled nature of the LoG stimuli. The CSF in healthy eyes was therefore determined by visual acuity. Therefore, in healthy visual systems and scaled visual stimuli, the field of contrast sensitivity function can be estimated from the field of acuity resolution and this may provide a simple reference to rapidly detect visual field deficits in visually impaired populations.

Yukai Zhao, Luis A. Lesmes, Michael Dorr,  Peter Bex and Zhong-Lin Lu (2019) “Accuracy and Precision of the ETDRS Chart, E-ETDRS and Bayesian qVA Method” Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

Purpose Visual acuity (VA) remains a fundamental measure of visual function. The accuracy and precision of VA assessment are extremely important for its use in disease management, therapeutic development, and occupation qualification. Although the ETDRS chart (Ferris III, et al., 1982) with different termination rules and E-ETDRS (Beck et al., 2003) provide the standard VA assessment in clinical trials, different termination rules may yield different VA scores for the same observer (Carkeet, 2001). Recently, Lesmes (2018) introduced a Bayesian adaptive qVA test that estimates the threshold and range of the VA psychometric function (PF) via higher sampling resolution of optotype size and a rich model of row-based PFs (Figure 1). In this study, we use Monte Carlo simulations to evaluate the accuracy and precision of VA assessment using ETDRS with 6 termination rules in current practice, E-ETDRS, and qVA. Methods: Observers with three different “true” VA thresholds (-0.3, 0.25, and 1 logMAR) and range (0.15, 0.3 and 0.6 logMAR) were simulated. The row-based PFs in qVA were used to simulate observer performance. The six termination rules were: reading the whole chart, or stopping at the line with at least 1, 2, 3, 4, or 5 mistakes. Each qVA run consisted of 45 trials with a row of 3 optotypes in each trial. Each observer was assessed 1000 times by each method. Results: The qVA generated the most accurate (bias: -0.004 to 0.004 logMAR) and precise (SD: 0.010 to 0.037 logMAR) assessment of VA thresholds, across observers (Figure 2). The ETDRS chart with different termination rules yielded VA scores with biases between -0.228 and 0.173 logMAR and SDs between 0.025 and 0.126 logMAR. Among the 6 termination rules, the ETDRS with the 3-mistake termination rule yielded the smallest bias (-0.018 to 0.079 logMAR), and the ETDRS with the whole-chart termination rule yielded the smallest SD (0.026 to 0.071 logMAR). The bias (0.096 to 0.057 logMAR) and SD (0.025 to 0.090 logMAR) of the E-ETDRS were similar to those of the ETDRS with the 5-mistake termination rule. Conclusions: The ETDRS with the 6 termination rules and E-ETDRS do not converge to the true acuity of the simulated observers. The qVA provides unbiased and most precise VA assessment. 

Srinivas Sridhar, Craig Versek, Ali Banijamali, Kameran Lashkari, and Peter J. Bex (2019) “Portable VEP Diagnostics for NeuroVisual Disorders” Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

Purpose Visual Evoked Potentials (VEPs) provide objective neuro-opthalmologic assessments that avoid patient task performance but utilizes invasive and cumbersome apparatus. We have developed a system that combines a scalp neuroelectric potential and field sensor with a smartphone in a portable wireless display headset called the NeuroDotVR. The system records VEPs and Fields (VEPF) in response to dichoptic stimuli presented on the smartphone display for a range of neuro-oplthalmologic disorders. We evaluate the NeuroDotVR for Dark Adaptation Recovers (DAR), a key biomarker for age-related macular degeneration (AMD). Methods DAR was measured simultaneously in both eyes of # patients with AMD and # age-matched controls. Following a 60s photobleach (TODOcd/m2 white cellphone screen), recovery of visual sensitivity was recorded with VEPFs to pattern reversal checkerboard stimuli presented to the   central (x-y degrees) and peripheral (> z degrees) macula of each eye. Results The amplitude of the DAR response is attenuated and delayed in patients with AMD. The dominant DAR response is centered around the N70-P100-N150 complex in subjects without AMD, but around approximately P350 in AMD patients. We developed a metric for DAR that can classify patients with AMD and provide a diagnostic marker for AMD based on VEG  responses. Conclusions NeuroDotVR is a comfortable, objective, portable wireless system that measures monocular and binocular VEPFs at multiple locations of the visual field. DAR can be measured with the system without requiring task performance and can classify deficits in DAR in patients with AMD.

Zhong-Lin Lu, Yukai Zhao, Luis A. Lesmes, Michael Dorr and Peter Bex (2019) “Unbiased Threshold Estimates in Bayesian Adaptive qCSF and qFC with Mismatched Psychometric Function Slopes.” Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

Purpose: To improve data quality in basic and clinical applications, Bayesian methods have been developed to adaptively assess thresholds on single [1,2] or multiple psychometric functions (e.g., the contrast sensitivity function [3,4]). To simplify these procedures – reduce model parameters and increase estimation efficiency – the slope of the psychometric function can be fixed [3,4]. However, a model mismatch occurs when the assumed slope differs from observer’s true slope. What is the impact of this mismatch on the accuracy, precision, and efficiency of adaptive estimation? In this study, we used Monte Carlo simulations to show that, for methods with fixed slopes, the qFC [2] in m-alternative forced choice tasks (m=2, 4, 8, and 10) and qCSF [3,4]: (1) there exists a d’ performance level at which the estimated threshold is unbiased, and (2) precision and efficiency increase with the observer’s true slope. Methods: For qFC, seven simulated observers, one with matched (3.05, 3.45, 3.90, and 4.05 for m=2, 4, 8, and 10, respectively) and six with mismatched slopes (0.5, 1, 2, 5, 6, 8) were simulated in each m-alternative task. The thresholds (d’=0.5 to 3.5) of each simulated observer were estimated with the qFC (100 trials) method 500 times. For qCSF, six observers, one with matched (4.05), four with a single mis-matched (1, 2, 6, and 8, respectively) across all spatial frequencies (SFs), and one with two mis-matched slopes (8 when SF < 4 cpd; 1 when SF ≥4 cpd), were simulated. The CSFs (d’=0.5 to 3.5) of each simulated observer were estimated with the qCSF (200 trials) method 1000 times. Results: The results are shown in Table 1. We found that the value of d’ where bias = 0 depended on the number of alternatives in forced choice tasks. Precision and the 68.2% half-width confidence intervals (HWCI) of the estimated thresholds increased with slope. Efficiency increased with slope and with the number of alternatives in forced choice tasks. Conclusions: Even under mismatched conditions, Bayesian adaptive methods with a fixed slope can generate unbiased threshold estimates in certain d’ performance levels. The results provide the theoretical basis to use psychometric functions with fixed slopes in parametric Bayesian adaptive procedures. REFS: [1] Kontsevich & Tyler, 1996; [2] Lesmes et al., 2015; [3] Lesmes et al., 2010; [4] Hou et al., 2015. 

Tina Y. Gao, Lisa M. Hamm, Joanna Black, Philip R.K. Turnbull, Peter J. Bex, Shuan Dai and Steven C. Dakin. (2019) “Automated assessment of ocular deviations using a consumer-grade eye tracker and 3D display.” Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

 Purpose The gold-standard for detecting strabismus and measuring ocular alignment are the cover/uncover test and the prism alternate cover test (PACT) respectively. Administration of these tests requires skilled clinicians and considerable time, and results are variable across examiners. We investigated whether a digital alternating cover test (ACT) consisting of a consumer-grade eye tracker and 3D display could provide an automated, efficient and objective assessment of eye alignment. Methods We tested 23 adult participants (age 20-51 years, including 4 with strabismus) on both a 9-point digital ACT and a standard clinical PACT. Both tests were conducted at a 55cm viewing distance. For the digital ACT, participants wore stereo glasses and fixated/followed a target on the 3D monitor. The target was presented monocularly and binocularly at a series of locations spanning ±15° horizontally and ±12° vertically from primary gaze. Both eyes were recorded by the eye tracker. Re-fixation eye movements (made in response to stimulus change) were used to detect strabismus and to quantify deviation angle. At least 3 measurements were taken for left and right eye fixing at each target location (total test time: 3-4 min) and the average of these values quantified deviation angle for each combination of target location and fixing eye. The digital ACT was performed twice for each participant to examine internal reliability. Results At primary gaze, mean horizontal angle for left eye fixing was 0.24 ± 9.84 Δ (SD) for the digital ACT and -2.70  ± 9.60 Δ  for the clinical PACT (Spearman’s r = 0.86, p<0.0001), and for right eye fixing was -0.05  ± 9.09 Δ for the digital ACT and -2.74 ± 9.32 Δ for the clinical PACT (Spearman’s r = 0.87, p<0.0001). Comparing the digital and clinical tests, the digital test measurements were on average 2.9 Measurements were consistent between the two repeats of the digital ACT (mean differences and 95% limits of agreement were -0.4 ± 4.6 Δ for left eye fixing and -0.2 ± 5.5 Δ for right eye fixing). Conclusions Low-cost eye-tracking technologies can be used to quickly assess ocular deviations at multiple gaze directions, and results are strongly correlated with standard clinical measures. This type of test may be useful in clinical settings, such as for regular repeated measurements or for children. 

Aigbe, S., Panorgias, T., Jeong, E., Otero, C., Bex, P.J. and Vera-Diaz, F. A. (2019) “Comparison of Monocular Microperimeter and Binocular Eye-tracking Methods for Assessment in Low Vision.” Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

Purpose: Sensitivity to blur is decreased in the peripheral visual field of myopic eyes, which may be associated with retinal zones within the near periphery that function as a retinal decoding system for optical blur. We used pattern electroretinograms (pERG) to evaluate whether this system is disrupted in the near-peripheral retina of myopes. Methods: ERGs were used to measure retinal responsivity in n=7 subjects with myopia (mean SE 2.59± 1.38D, mean age 24±1 years) and n=5 subjects with emmetropia (mean SE 0.03± 0.04D, mean age 26±2 years). Retinal activity was measured while subjects viewed naturalistic images with simulated optical blur. Images were convoluted to: (1) four different types of blur (Defocus, Spherical Aberration [SA], Astigmatism [J45], and a combination of Astigmatism and Defocus [J45+Defocus]; (2) blur from the periphery and up to one of three eccentricities (12, 6, or 0 degrees); and (3) three levels of blur (0.1, 0.3, or 0,5μm for Defocus, SA and J4, and 0.071, 0.213, or 0.354μm for J45+Defocus). Subjects viewed nine images for each type of blur and one no blur condition in random order, for a total of 37 conditions. The N95 amplitude of the pERG was analyzed using a mixed model ANOVA to each parameter (within subjects factors: type of blur{Defocus, J45, SA, J45+Defocus}, blur amount{0.1, 0.3, 0.5}, eccentricity{12, 6, 0} and between subjects factor: refractive group{Myopes, Emmetropes}. Results: A main effect of amount of blur was found for each type of blur (Defocus: F=36.07, p<0.01; SA: F=12.58, p<0.01; J45: F=6.76, p<0.01; J45+Defocus: F=11.84, p<0.01) in both myopes and emmetropes. There were significant differences between responses to 0.1μm and 0.3μm blur for SA (p=0.01) and between 0.1μm and 0.5μm for all types of blur. A main effect of blur eccentricity was also found for Defocus (F=11.41, p<0.01), SA (F=14.25, p<0.01) and J45+Defocus (F=4.02, p=0.02). Significant interactions were found for eccentricity and the amount of blur for SA only. No main effect was found for refractive error group. The only difference between emmetropes and myopes in this study was found for 0.3μm of J45 blur (p=0.04). Conclusions: Differences in retinal responses with blur eccentricity suggest distinct retinal zones for blur decoding in both myopes and emmetropes. The effect of blur amount was larger than the effect of eccentricity or type of blur. No differences in retinal responses were found between emmetropes and myopes. 

Arango, T., Martire, J., Ross, N. and Bex, P.J. (2019) “Comparison of Monocular Microperimeter and Binocular Eye-tracking Methods for Assessment in Low Vision.” Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

Purpose Fundus-oriented microperimetry measures visual sensitivity and fixation behavior in retinal coordinates, but utilizes expensive equipment ,which provides only monocular measurements. We evaluated a low-cost approach comprising a consumer-grade eye tracker and stereoscopic 3D display that may provide monocular and binocular functional assessment in retinal coordinates for patients with low vision. Methods Microperimetry and fixation assessment was performed on a NIDEK MP1 microperimeter in 12 eyes with central vision loss (ages, 18 to 90 years). The same subjects completed fixation assessment with a 500Hz eye tracker. Gaze positions of the left and right eyes were recorded while observers fixated a binocular target with their Preferred Retinal Locus (PRL). Fixation stability was calculated from the bivariate contour ellipse area (BCEA, 68%). To provide a retinal landmark, observers reported when a monocular target, independently controlled with active shutter glasses, disappeared into the optic nerve head (ONH) of their left and right eyes. Typical anatomical dimensions were used to compute the locations of the diseased fovea and the PRL relative to the ONH in each eye. The areas of absolute and relative vision loss were computed with iso-sensitivity contour fits to the perimetry data and their locations were recorded relative to the ONH. Results There was no significant difference between estimates of fixation stability measured with the microperimeter (OD median = 8.28°2; OS median = 7.1°2) and the consumer system (OD median = 59.6°2; OS median = 14.72°2) (OD, ranksum = 30, p = 0.18; OS, ranksum = 35, p = 0.59). There were significant differences between the location of the PRL used under binocular conditions with the consumer system and monocular conditions with the microperimeter. The PRL reported by the microperimeter fell within regions with a manifest scotoma, 50% of the time (6 eyes), but outside a scotoma in the consumer system, suggesting that task instructions may influence behavior. Our estimate of scotoma area, density and location provide new quantitative metrics that may help track the progression of eye disease. Conclusions Low cost consumer hardware can provide useful quantitative metrics of binocular visual function in patients with low vision that are not currently possible with high cost clinical apparatus. We were found comparable fixation stability in the consumer system using a binocular fixation assessment compared to Nidek’s monocular assessments. However, there were significant differences in predicted PRL locations between systems, which may be task specific. Our estimate of scotoma characteristics may elucidate relative differences between scotoma and visual function in patients with low vision. 

Arango, T. and Bex, P.J. (2019) “Oculomotor characteristics under binocular eye-hand coordination.” Submitted to Vision Sciences Society  Abstract

 Patients with central vision loss often adopt eccentric viewing strategies using a preferred retinal location (PRL) in the absence of functional foveae. Little is known about the oculomotor characteristics of the PRL under binocular eye-hand coordination. We examined PRL location, fixation stability and smooth pursuit under binocular viewing using simulated gaze-contingent scotomas in normally-sighted observers. Participants (N=7) completed two eye-hand coordination tasks in two conditions with gaze-contingent central scotomas. Scotomas were updated at 60Hz refresh rate based on a 500Hz eye tracker. In a fixation task, participants touched a target in the center of the screen with their hand. In a pursuit task, participants followed a horizontally oscillating target with their hand. In the second condition, participants viewed a picture of a hand that moved in same touch positions as previous trials in the fixation and smooth pursuit tasks. Participants completed two runs of each condition. Outcome measures were fixation stability (BCEA, 68%), PRL eccentricity and stability of smooth pursuit. PRL eccentricity increased with scotoma size (p < .001), but BCEA did not vary as function of scotoma size, task or condition (ps > 0.05). There was a significant interaction between run and condition (hand vs picture): PRLs were closer to the target in second run of the hand but not picture condition (p = 0.05). Smooth pursuit stability decreased with increasing scotoma size and was greater in the own hand than picture condition (p < .01). Smooth pursuit but not fixation stability varied as function of scotoma size and hand condition. There was a training effect for PRL location in the hand condition, suggesting that PRL use may adapt under eye-hand coordination. These differences offer insight into how the coupling of eye and hand movement may influence binocular visual function in patients with central vision loss. 

Dian Li, Lucy Q. Shen, Louis R. Pasquale, Michael V. Boland, Sarah R. Wellik, Carlos Gustavo de Moraes, Jonathan S. Myers, Peter J. Bex, Osamah Saeedi, Neda Baniasadi, Hui Wang, Jorryt Tichelaar, Tobias Elze, Mengyu Wang (2019) Baseline Visual Field (VF) Patterns are Predictive of Global and Central Loss in 24-2 Visual Fields. Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

 Purpose: To study whether and how visual field (VF) patterns are related to global and central loss in 24-2 VFs. Methods: Eyes with both ≥ 5 reliable 24-2 VFs and 5-year follow-ups (intervals ≥ 6 months) were selected. One eye was included randomly if both eyes qualified. 2 feature types from 2 baseline VFs were extracted: (1) Pattern features, including 16 VF pattern coefficients quantified by 16 VF archetypes (ATs) derived by machine learning in our prior work (Fig. 1), and the mean squared error (MSE) of VF archetype reconstruction; (2) VF global indices, including baseline mean deviation (MD)/pattern standard deviation (PSD), MD/PSD differences between the 2 baseline VFs, glaucoma hemifield test (GHT), and the total deviation (TD) difference (MSE) between 2 baseline tests. Linear regression was applied to predict the MD slope and average central 4 TD slope adjusted by the baseline age and follow-up time. Stepwise regression via the Bayesian information criterion (BIC) was used to select the optimal features. Results: For the 7,360 eyes from 7,360 patients, the baseline age, MD, follow-up time, and the number of VFs were 65.2±12.7 years, -4.4±5.2 dB, 7.5±1.9 years and 6.6±1.7, respectively. The Pearson correlation coefficient between MD slope and central TD slope was 0.79 (P<0.0001). Eyes with higher coefficients of the temporal wedge (AT 4), near total loss (AT 6), central scotoma (AT 7), concentric peripheral defect (AT 11) and TD difference were less likely to develop MD worsening (Fig. 2a). MD worsening was related to higher AT MSE, abnormal baseline GHTs, and worsened MD/PSD between 2 baselines. Central TD worsening was more associated with higher coefficients of superonasal step (AT 3), inferior altitudinal defect (AT 13), inferior
paracentral defect (AT 16) and AT MSE, and was less likely for eyes with higher coefficients of near total loss (AT 6), central scotoma (AT 7) and TD difference (Fig. 2b). Abnormal baseline GHTs, and worsened MD between 2 baselines were associated with central TD worsening. Compared to VF global indices alone, adding VF pattern features improved the prediction of MD slope and mean central TD slope by a decrease in BIC with 18 and 78 (lowered BIC > 6: strong model improvement). Conclusions: Adding VF pattern features improves the prediction of the global and central 24-2 VF
worsening, which might result in better glaucoma progression recognition. 

Mengyu Wang, Lucy Q. Shen, Louis R. Pasquale, Michael V. Boland, Sarah R. Wellik, Carlos Gustavo de Moraes, Jonathan S. Myers, Peter J. Bex, Osamah Saeedi, Neda Baniasadi, Hui Wang, Dian Li, Jorryt Tichelaar, Tobias Elze (2019) Central Visual Field Loss Patterns in Glaucoma. Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

Purpose: To illustrate and quantify the central visual field (CVF) loss patterns in glaucoma. Methods: Total deviation values from reliable 10-2 VFs from 5 academic glaucoma services in the US were used to determine the CVF loss patterns for eyes with all glaucoma severities using the most recent 10-2 VFs. Furthermore, the CVF loss patterns were identified for mild, moderate and severe glaucoma separately. The 10-2 VFs tested within 3 months of or before any 24-2 VF with mean deviation (MD) ≥ -6 dB were assigned to mild glaucoma, were assigned to severe glaucoma if tested within 3 months of or after any 24-2 VF with MD < -12 dB, and were assigned to moderate glaucoma if they were tested within 3 months of or between 24-2 VFs with -12 dB ≤ MD < -6 dB. The most recent 10-2 VFs of each eye were selected for each glaucoma stage. An unsupervised machine learning method of archetypal analysis was applied to determine the CVF loss patterns. Cross-validation was performed to determine the optimal number of patterns for archetypal analysis. Results: 13,951, 3,529, 1,528 and 3,066 VFs were selected to determine the CVF patterns for eyes with all glaucoma severities, mild, moderate and severe glaucoma, respectively, for which 17, 11, 11 and 16 CVF loss patterns were identified. For eyes with all glaucoma severities (Fig. 1), the most common pattern is intact field (38.1%). There were 4, 4, and 5 CVF patterns related to superior loss, inferior loss and overall loss with island of vision, which accounted for 17.2%, 12.5% and 19.8% CVF loss, respectively. The total loss pattern accounted for 5.2% CVF loss. Notably, 4 of the 5 CVF loss patterns with island of vision preserved the less vulnerable zone proposed by Hood and co-workers. When comparing the 3 glaucoma severities, similar CVF defects include superonasal defects and temporal island patterns (Fig. 2a). Unique patterns include central defect (6.0%) and superior-peripheral loss (10.0%) for mild glaucoma, central-nasal loss (5.2%) for moderate glaucoma, and total loss (8.6%) and inferotemporal loss (4.1%) for severe glaucoma (Fig. 2b). Conclusions: Various CVF loss patterns in glaucoma were illustrated and quantified for the first time. Some CVF defects were present in eyes with all severities, while some were specific to particular glaucoma severities. Our quantified CVF loss patterns are promising to advance understanding of CVF loss in glaucoma. 

Jorryt Tichelaar, Mengyu Wang, Lucy Q. Shen, Louis R. Pasquale, Michael V. Boland, Sarah R. Wellik, Carlos Gustavo de Moraes, Jonathan S. Myers, Peter J. Bex, Osamah Saeedi, Neda Baniasadi, Dian Li, Hui Wang, Tobias Elze (2019) Defect Classes of Visual Field Loss in Glaucoma. Submitted to the Association for Research in Vision and Ophthalmology.  Abstract

 Purpose: To date, glaucomatous visual field (VF) loss is diagnosed and quantified by summary measures that ignore specific glaucomatous patterns. Here, we computationally trace the evolution of VF loss patterns over time to develop a quantitative model of functional glaucoma subtypes (defect classes). Methods: To determine representative patterns, total and mean deviations (MD) of the most recent reliable VFs (fixation loss≤33%, false negatives/positives≤20%, SITA Standard 24-2) were retrospectively selected among all patients from five clinical glaucoma practices. VFs were grouped by severity into MD bins of 1dB width. Separately for each MD bin, archetypal analysis, an unsupervised machine learning technique, was applied to autonomously determine severity-specific patterns (archetypes [ATs]). To trace their evolution over time, all eyes with multiple VFs were selected. Each VF was quantitatively decomposed into the ATs of each MD bin by weighting the AT coefficients according to the measurement uncertainty distribution of the respective measured MD (Fig. 1A). For each baseline-follow up VF pair, an evolution matrix of the respective AT coefficients is calculated (Fig. 1B). Pooling over all evolution matrices, for each AT within each MD bin, the most probable follow-up AT for each next-more-severe stage was calculated, which yielded pattern trajectories that define defect classes. Results: Based on 179,936 VFs from 179,936 eyes, between 7 and 14 ATs were identified for each MD bin. Based on baseline-follow up evolution matrices from 278,445 VFs from 74,806 eyes, for moderate severity glaucoma (-3dB≥MD>-12dB), seven distinct defect classes could be determined that contain features which are compatible with nerve fiber defects (Fig. 2A): A singular upper peripheral class as well as upper-lower hemifield pairs of (para-)central, nasal, and temporal patterns. Nasal, but not temporal to the blind spot, the classes respect the horizontal midline. All defect classes apart from one upper paracentral class spare central vision. Distinctive features like presence/absence of central VF loss are preserved within the respective defect classes over the whole severity range (e.g. Fig. 2B). Conclusion: We introduce VF loss defect classes which allow quantitative defect classifications of arbitrary VF measurements, which may not only help to improve glaucoma progression diagnosis but also enhance subtype-specific structure-function modeling. 

2018

Rose, D. & Bex, P.J. (2018) Semantic Scene Statistics Using a Novel Computational Method. Presented at the European Conference on Visual Perception.  Abstract

A number of computational salience models exist that describe low-level image features responsible for gaze position selection. Though powerful, there is increasing interest in extending these models to capture additional types of information, such as top-down effects of scene context or object-level features. We propose a novel method for modeling joint linguistic/visual content in this domain that we call semantic saliency. This process evaluates the semantic closeness or similarity of scene objects with respect to each other (object-object semantic saliency) and to a computationally generated scene context label (object-scene semantic saliency) using vector-space language models. These scores are then aligned with object-specific masks within an image, allowing inferences to be made regarding the effect of semantic salience on gaze guidance to be evaluated directly. We here present a specification of the method, as well as a of descriptive results gathered from applying it to a large corpus of images (~10k). We also present results examining the relative degree of gaze guidance both types of semantic saliency and Itti-Koch image saliency express for a subset of these images (210) using eye movements made in response to one of three tasks (free viewing, object counting and scene description).
Correlations between semantic and image saliency maps, averaged across images in this subset, were modest but highly significant for both semantic saliency map types. The average correlation coefficient for image to object-object saliency maps was 0.35 (one-sample t-test, t=26.5,df =209, p<0.0001);  for image to object-scene it was 0.33 (t=25.9, df =209,p<0.0001). ROC analyses of gaze position alignment with salience maps of both types indicate that object-object or object-semantic saliency were significantly better predictors of gaze than image salience across tasks (Tukey HSD post-hoc test, p<0.0001 for both image/object-object and image/object-scene), with equivalent performance between the semantic saliency map types (p=0.68).
Taken together, we believe these results support the use of our method for calculating semantic saliency, alone or in combination with other top-down and bottom up salience models, for predicting gaze. Object-scene semantic saliency in particular represents a novel method for including potentially important information regarding the role of novel or contextually unlikely scene content on gaze guidance. 
View poster pdf

Wang, M. , Pasquale, L. R., Shen, L. Q., Boland, M. V., Wellik, S. R., De Moraes, C. G., Myers, J. S., Baniasadi, N., Li, D., Wang, H., Bex, P. J. & Elze, T. (2018) “Predicting Functional Progression in Glaucoma from Baseline Visual Fields.” Association for Research in Vision and Ophthalmology. Abstract

Purpose To predict glaucomatous visual field (VF) worsening after at least 5 years using features of 3 baseline VFs. Methods In this retrospective multi-center cohort study, eyes with ≥ 5 reliable automated VFs and ≥ 5 years follow-up were selected. We also restricted the time between each follow-up VFs to be ≥ 6 months, the time between the first and third VFs to be ≤ 3 years, and glaucoma hemifield test (GHT) to be within/outside normal limits or borderline. VF features are extracted: (1) age, follow-up time, mean deviation (MD) and pattern standard deviation (PSD) of the 3rd VF, (2) GHT results, MD slope, PSD slope and intraclass correlation for the 3 baseline VFs, and (3) worsening VF locations using permutation of pointwise linear regression (PoPLR, slope<-1 dB/year and p<0.01) and the Collaborative Initial Glaucoma Treatment Study (CIGTS) score of the 3rd VF based on the 3 baselines. Worsening VFs were determined by CIGTS criteria, MD regression (minus slope, p<0.01) and PoPLR (≥3 progressed locations). The time of progression was defined as the follow-up time when progression was first detected with progression confirmed on the full VF series. Cox regression was applied to model risk of worsening from VF features with stepwise model selection. 10-fold cross validation was applied to evaluate the model by the area under the receiver operating characteristic curve (AUC). Results 7,522 eyes were selected. The mean±standard deviation of age, MD and PSD at the 3rd baseline VF were 65.9±12.4 years, -4.4±5.5 dB and 4.4±3.8 dB. The median of total follow-up time and number of VFs were 6.7 years and 7. The median time to worsening for the methods of CIGTS, MD regression and PoPLR were 6.21 (671 eyes), 6.79 (793 eyes) and 5.91 (1,035 eyes) years, respectively. CIGTS progression was positively associated with age, MD, PSD slope, ICC and CIGTS score, and negatively associated with follow-up time at the 3rd VF and MD slope (Fig. 1 (a)). Abnormal GHT results of the second and 3rd VFs were associated with progression. The AUCs to predict 5-year progression defined by CIGTS (142 eyes), MD (95 eyes) and PoPLR (240 eyes) were 0.912, 0.830 and 0.819, respectively. Fig. 2 (a) and (b) show the optimal models to predict progression defined by MD and PoPLR. Conclusions Our model using 3 baseline VFs with mean follow-up time of 2.1 years accurately predicts the 5-year progression risk using various progression algorithms.

Li, D., Pasquale, L. R., Shen, L. Q., Boland, M. V., Wellik, S. R., De Moraes, C. G., Myers, J. S., Baniasadi, N., Wang, H., Bex, P. J., Elze, T. & Wang, M. (2018) “A New Method to Detect Visual Field Progression based on Spatial Pattern Analysis.” Association for Research in Vision and Ophthalmology.  Abstract

Purpose  Current methods for detecting visual field (VF) change only provide global assessments of functional deterioration, while clinician’s judgement of VF progression heavily relies on examination of VF spatial patterns. We aim to develop a new algorithm to detect VF worsening to provide both status outcome and spatial patterns of VF progression. Methods In this retrospective cohort study, we selected eyes from multiple sites with at least 5 reliable automated VFs and at least 5 years follow-up using the SITA Standard strategy and 24-2 pattern. The time between each VF was restricted to be at least 6 months. Each VF was decomposed into a weighed sum of 16 VF archetypes (ATs) including 1 normal AT and 15 VF loss ATs (Fig. 1) identified previously [1]. For each eye, linear regressions were applied from follow-up time to the 16 AT weights of VFs. If any of the weights substantially changes for the 15 VF loss ATs (AT slope≥0.01/year and p<0.01) and the normal AT (AT slope≤-0.01/year and p<0.01), the eye is determined to be worsening. Our new algorithm was compared to existing algorithms including MD slope, Advanced Glaucoma Intervention Study (AGIS) scoring, Collaborative Initial Glaucoma Treatment Study (CIGTS) scoring and the permutation of pointwise linear regression (PoPLR). The concordance between those algorithms was evaluated by Kappa coefficient. Results 12,217 eyes were selected for analyses. The mean±standard deviation of age, MD and PSD at the first VF were 63.8±12.7 years, -4.1±5.2 dB and 3.9±3.5 dB. The median of follow-up time and number of VFs was 7.1 years and 6. The prevalence of VF progression by AT slope, MD slope, AGIS scoring, CIGTS scoring and PoPLR were 10.3%, 9.3%, 3.9%, 9.4% and 9.2%. The progression detection by AT slope was in fair agreement (Kappa 0.2 to 0.4) with MD slope (0.38), AGIS (0.23), CIGTS (0.25) and PoPLR (0.27). The overall Kappa coefficient between existing algorithms was 0.34. Among the progressed 1,262 eyes determined by AT analysis, 89.6%, 9.8% and 0.6% of the eyes had 1, 2 and 3 ATs progressed, and the 3 most frequent progressed ATs are AT 8 (17.3%), 6 (15.2%) and 3 (8.3%) (Fig. 2). Conclusions Our new algorithm based on AT analysis can provide information of spatial pattern of VF progression in addition to status outcome. The spatial patterns of VF progression can be used to assist clinicians to assess progression. [1] Elze et al., 2015, J. R. Soc. Interface

McCormack, G., Ferrucci, J., Guarino, A. & Bex, P.J. (2018) “The Shape of the Psychometric Function in Blur Discrimination Judgments.” Association for Research in Vision and Ophthalmology.  Abstract

Purpose The effect of motion on blur sensitivity (“motion sharpening”) could affect the detectability of blur caused by accommodative error in 3D displays. Discrimination of blur during motion has been tested by rapid adaptive psychophysical methods which assume that the shape of the psychometric function is a logistic curve (“PEST”) or a Weibull curve (“QUEST”). Because the accuracy of these adaptive methods might depend on the actual shape of the psychometric function, we have tested whether a logistic curve or a Weibull curve more accurately fits blur discrimination data. Methods Six healthy young adults wearing full refractive corrections viewed images on an Asus 3D display at 1M. The test images were 1°w x 2°h bars of 50% contrast on a dark gray background, with 42 minute of arc gaussian blurs on the vertical edges. The bars moved leftward or rightward at 1.0°/s, either above or below a small fixation point. Motion sharpening reduced perceived edge blur, with the trailing edge more sharpened than the leading edge. We evaluated the shape of the psychometric function for matching the leading edge blur to the trailing edge blur. The approximate match blur was initially determined by PEST, after which the psychometric function was sampled by a constant stimulus method with 20 repeated trials at each of nine leading edge blur values centered on the initial PEST match value. Levenberg-Marquardt algorithms in LabView software fit logistic and Weibull curves to the constant stimulus data. Match values and chi-square goodness of fit values were determined for each curve. Results In leftward motion the logistic curve more accurately fit the data than the Weibull curve in all subjects, and for rightward motion the logistic curve more accurately fit the data in two thirds of subjects. To determine differences between the logistic and Weibull match-values, two separate paired t-tests of match values were conducted with alpha adjusted by the Bonferroni technique (p < .025) to control for inflated Type I error rates. The logistic analysis produced statistically significantly lower match values than the Weibull analysis in both leftward motion (p<.001) and rightward motion (p<.021). Conclusions The logistic curve more accurately represented the psychometric function in blur discrimination judgments than did the Weibull curve.

Vera-Diaz, F.A., Otero, C. & Bex, P.J. (2018) “Effect of monochromatic aberrations on suprathreshold and threshold discrimination tasks.” Association for Research in Vision and Ophthalmology.  Abstract

Purpose Each form of optical aberration degrades retinal image quality differently, which may be critical for assessing image quality for optical correction and refractive error etiology. We therefore evaluated the effect of different levels and combinations of aberrations on suprathreshold and threshold visual discrimination. Methods Images were computationally convolved with different combinations of monochromatic aberrations at levels 0.1 to 0.5µm: defocus (C2,0), spherical aberration (SA) (C4,0), vertical astigmatism (C2,2), and oblique astigmatism (C2,-2); or one of these combinations: (C2,2)+ (C2,0), (C2,-2)+ (C2,0), (C2,0)+ (C4,0), all in random order. Aberrations were applied to the images’ point spread functions for a 5mm pupil. Two visual tasks were performed at 5m distance: (1) an ETDRS letter acuity (VA) threshold task, and (2) a suprathreshold task where subjects matched the blur of dead-leaves naturalistic stimuli. Seven healthy subjects with best corrected VA of 0.0LogMAR or better participated. For each visual task and subject, a linear fit was computed and the area under the curve obtained integrating over the range [0.1 0.5] µm. Results Repeated measures ANOVA for both tasks showed statistically significant effects among types of aberrations (VA: F=74.70, p<.001, Matching: F=93.07, p<.001). Reduction in VA was greater with defocus and SA than oblique astigmatism (C2,0 – C2,-2: p<0.01; C4,0 – C2,-2: p<0.01). Combinations of aberrations decrease VA less than any single aberration (F=213.96, p<.001). This effect was greater for defocus added to astigmatism than defocus added to SA. For the matching task, oblique astigmatism produced perceptually higher levels of blur compared to the other single aberrations (p<0.01). The perceived blur caused by any single aberration was greater than the combinations of aberrations. Similarly, defocus added to SA was perceived as less blurred when compared to defocus than when compared to SA (p<0.01). Conclusions Even though the optical quality of the retinal image varies with aberration order, functional measures of perceived blur and VA are relatively invariant of aberration order. For both matching suprathreshold and threshold discrimination tasks, single aberrations caused greater functional degradation than combinations of aberrations, suggesting that aberrations are not linearly summed in the human visual system.

Kosovicheva, A., Hawlader, O. &. Bex, P.J. (2018) “Localization errors following saccadic adaptation to a dichoptic step.” Vision Sciences Society.Abstract

In saccadic adaptation, saccade amplitudes change to account for errors in saccade landing location. Saccadic adaptation also produces mislocalizations around the adapted target location (Bahcall & Kowler, 1999; Awater, Burr, Lappe, Morrone, Goldberg, 2005). While saccadic adaptation is typically tested with identical target steps in each eye, previous work has demonstrated that the oculomotor system can adaptively recalibrate saccade amplitudes in response to opposite target shifts in each eye (Maiello, Harrison, & Bex, 2016). We investigate the perceptual effects of adaptation to a dichoptic step: are localization errors consistent with changes in saccade amplitude in different directions in the two eyes? Subjects made repeated rightward saccades to a Gabor target at 8º eccentricity, and shutter glasses were used to introduce a dichoptic step. On saccade onset, the saccade target stepped 0.8º inward in the left eye and 0.8º outward in the right eye, remaining onscreen following saccade completion. After 75 adaptation trials, subjects were shown probe trials (interleaved with top-up adaptation trials) to measure perceptual shifts following adaptation. During probe trials, subjects performed a post-saccadic Vernier discrimination task with a pair of brief (100 ms) lines, each shown to a different eye. Following adaptation, in order to appear collinear, each Vernier line needed to be shifted in the same direction as the adapting step in the corresponding eye. The perceived misalignment was significantly larger compared to a control condition in which subjects adapted to identical inward steps in both eyes (mean total misalignment of -0.44º vs. -0.03º; p = 0.008). Similar mislocalizations were observed when comparing the apparent locations of pre- and post- saccadic probes, indicating that each line was shifted relative to the pre-saccadic goal. Together, these results are consistent with previously reported mislocalizations following conjugate target steps, and demonstrate novel dichoptic visual errors following oculomotor adaptation. 

McKay, R.A.  Piano, M.E.F, Bex, P.J., Preston, J.A., Stansfield, B.W. and Simmers, A.J. (2018) “Persistent Visual Impairment in Multiple Sclerosis: Prevalence and functional consequences” Presented at Vision Sciences Society. Abstract

Impaired visual function is a concern for up to 80% of people who have received a diagnosis of multiple sclerosis (Balcer 2015), and may result from involvement of structures in the afferent visual pathways, ocular motor systems or from cerebral deficits. Although visual symptoms in pwMS may precede, occur simultaneously with, or follow the development of other neurologic manifestations, they may represent the most prominent symptoms from the person living with MS’s point of view. In a mixed methods study 110 participants completed a series of quality of life questionnaires (VFQ-25 with 10item neurological add-on, MSIS29) and a full visual (HCVA, Contrast Sensitivity Function – spatial and temporal, ocular alignment and motility, colour vision, confrontational visual fields, stereoacuity and Pulfrich’s) and functional assessment using the Multiple Sclerosis Functional Composite and the King-Devick Test. A small subgroup (n=10) undertook a further assessment of visual fields, motion coherence, OCT and balance using the Biosway Portable Balance Test. In Stage One pwMS reported greater persistent visual dysfunction impacting on daily life when compared with a non-MS, eye disease free population. Clinical assessment in Stage Two found that while visual deficits previously reported as cause for concern in pwMS – high contrast acuity, visual fields, colour and binocular vision did not appear to affect pwMS longitudinally, contrast sensitivity function was a significant cause of persistent visual loss impacting on daily life in pwMS. In Stage Three motion coherence was also found to be impaired causing difficulties with daily life. The study findings demonstrate a significant loss in visual perception in pwMS when testing in both the temporal and spatial domain. These persistent visual losses correlated with self-reported visual function and functional assessment having implications for guiding future rehabilitation strategies.

Arango, T. & Bex, P.J. (2018) “Fixation Patterns to Celebrities and Selfies following Image and Task Modification.” Vision Sciences Society.  Abstract

Normally-sighted participants (N=6) viewed 150 images of famous faces or selfies under 4 conditions: upright, inverted, Mooney or Canny line drawings. Face images were 17×23 degrees, and normalized by aligning the pupils at a separation of 7 degrees. After viewing each image for 5 seconds, in separate blocks, observers identified either the written name or an image feature (right eye, left eye, left ear, right ear, nose, chin,
mouth and forehead, at random across trials) in a 2AFC task. The name or feature of the test face was paired with a name or same feature from a different face. Binocular gaze direction was tracked using a 60Hz eye tracker. Face identification was significantly worse for line drawings than upright faces (p< .0001). Gaze direction varied as a function of face feature (p< 0.001) with more fixations towards the eyes and nose than any other features (all p< 0.05), and more fixations to left than right eye (all p< 0.05). Participants made fewer fixations on the eyes of all non-upright faces or selfies compared to upright faces (all p< 0.05), but fixation patterns were not significantly different for face naming than feature recognition (p<0.87). In agreement with previous research, gaze direction is not uniformly distributed across faces. However, changes in image type can modify the pattern of eye movements at no expense to identification accuracy. Furthermore, changes in task type failed to modify fixation patterns. These findings are not consistent with a single information-driven goal of eye movements.

Arango, T., Martire, J., Ross, N. & Bex, P.J. (2018) “Oculomotor behavior in monocular versus binocular gaze-contingent scotomas.” Association for Research in Vision and Ophthalmology.  Abstract

Purpose Patients with bilateral central vision loss often adopt a preferred retinal location (PRL) in the absence of functional foveae. Fixation stability as quantified by the bivariate contour ellipse area (BCEA) of PRL gaze positions is correlated with functional performance and used as a clinical endpoint. Fixation stability is typically measured under monocular viewing. Little is known about the properties of the PRL under binocular asymmetric scotomas. We explored PRL location and fixation stability under monocular and binocular viewing conditions using simulated independent gaze-contingent scotomas in normally-sighted observers. Methods Normally-sighted participants completed 3 tasks with monocular (n=12) and binocular (n=12) gaze-contingent pink noise scotomas (0.5, 1, 2 deg radius). Scotomas were updated at 120Hz refresh rate based on a 500Hz eye tracker. For binocular scotomas, the scotoma size and location in each eye was independently controlled with active shutter glasses in 3 symmetric and 3 asymmetric combinations. Participants fixated at a dot at nine locations (8X8 deg grid), and performed a contrast detection task and a word identification task at 2 locations. Outcomes were fixation stability (BCEA, 68%) and PRL location.Results BCEA varied as a function of gaze posture in the dot fixation task under both binocular (p<.0001) and monocular (p=.03) viewing conditions. BCEA increased with scotoma size in the monocular (p=0.08), but not in the binocular condition. There was no effect of scotoma size or symmetry on BCEA or PRL location in monocular or binocular viewing for the contrast task. Under monocular viewing for the word recognition task, BCEA increased with scotoma size (p=.01). There was a significant interaction between testing location and scotoma size (p=.04): PRLs were closer to target with the largest scotoma. PRLs were closer to the target in monocular than binocular conditions (p=.04). Conclusions For simulated scotomas, estimates of PRL location and stability depend on gaze direction, scotoma size and viewing condition. For a simple fixation task, location influenced fixation stability more than scotoma characteristics. Simulated monocular gaze contingent scotomas capture some, but not all of the oculomotor characteristics observed in patients. The nuanced differences in oculomotor control with monocular compared to binocular scotomas emphasize the need to measure visual function binocularly.

Martire, J., Arango, T., Bex, P.J. & Ross, N.  (2018) “Visual performance at the PRL in monocular versus binocular viewing using simulated gaze-contingent scotomas.” Association for Research in Vision and Ophthalmology. Abstract

Purpose The majority of low vision patients are affected by macular pathologies where foveal function is lost in both eyes and a preferred retinal locus (PRL) is adopted. Much research has explored properties and visual performance of the PRL in monocular viewing, but little is known about the PRL under more realistic binocular viewing conditions. We used simulated gaze-contingent independent scotomas in normally sighted participants to explore visual performance of the PRL under binocular and monocular viewing. Methods Gaze contingent scotomas (0.5, 1.0, 2.0 degree radius) were simulated in each eye of participants in monocular (n=12) and binocular (n=12) viewing conditions. Scotomas were updated at 120Hz refresh rate based on a 500Hz eye tracker. Size and location of binocular scotomas were independently controlled with active shutter glasses. Visual performance was assessed using contrast sensitivity function (CSF) and word identification. For CSF, observers identified band-pass filtered letters, and a quick algorithm controlled spatial frequency and contrast. For word identification, participants identified whether a word or non-word was presented and a quick algorithm controlled font size and duration. Outcomes were high-frequency CSF threshold (CSF acuity), area under the logCSF curve (AULCSF), threshold reading acuity and reading speed. Results Under monocular viewing conditions, contrast sensitivity decreased with increasing scotoma size, with declining CSF acuity (p<0.01) and decreasing AULCSF (p <0.01). For word identification, under monocular conditions, there was no effect of scotoma size on reading acuity but a significant effect for maximum reading speed (p<0.01). Under binocular viewing conditions there was no significant effect of scotoma size on visual performance for any of the tasks. CSF acuity was significantly better when scotomas were symmetric than asymmetric (p=.01) across all sizes. Conclusions With simulated independent binocular gaze-contingent scotomas, visual performance at the PRL differs under monocular versus binocular viewing conditions. Performance depends on whether the scotomas are symmetric or asymmetric and the method used to assess visual function. The results stress the need for the development and use of binocular visual assessments to explore properties and visual performance of the PRL in low vision patients.

Zhao, Y., Lesmes, L.A., Dorr, M., Bex, P.J. & Lu, Z.L. (2018) “Psychophysical validation of a novel Bayesian method
for measuring visual acuity.” Association for Research in Vision and Ophthalmology. Abstract

Purpose The ETDRS chart (Ferris III, et al., 1982) and its computerized version, e-ETDRS (Beck et al., 2003), remain the standard for testing visual acuity in clinical trials, but testing acuity with precision remains a challenge. Lesmes (2018) introduced a novel Bayesian visual acuity test (Bayesian VA) that provides the advantages of high-density sampling of optotype size (.02 logMAR), adaptive stimulus optimization (Lesmes, et al, 2006), and the post-hoc analyses of Bayesian credible intervals (68.2% half-width credible interval – HWCI). Here, a proof-of-concept psychophysical study evaluates the accuracy and precision of the Bayesian VA method.Methods For six subjects with normal or corrected-to-normal vision, monocular visual acuities were measured for both eyes in four acuity conditions (three with different levels of Bangerter foil to degrade vision and one with no foil). In each session, subjects were tested monocularly with Bayesian VA and e-ETDRS methods 4 times in each eye in one of four foil conditions, with random ordering of the two methods and two eyes in different blocks. For each trial of Bayesian VA testing, three optotypes were presented, with their size selected by an adaptive maximization of information gained about the threshold and range of the full acuity function.Results There was excellent agreement between the estimated thresholds from the novel VA approach and the e-ETDRS standard, with a correlation coefficient of .99 (p<0.001) across all subjects and foil conditions. Across conditions with average acuities ranging from -0.02 to 0.50 logMAR, the average HWCI of the estimated acuity from e-ETDRS was 0.096 ± 0.019 decimal log units. In comparison, the precision of Bayesian VA increased with the number of trials. The average HWCI of estimated acuities after 10, 15, and 45 trials were 0.091 ± 0.028, 0.070 ± 0.022, and 0.037 ± 0.011, respectively. Because of different stimulus sampling resolutions (.02 vs .10 logMAR), we compared the test-retest precision of the two methods using Fractional Rank Precision (FRP, where higher values indicate higher precision; Dorr, et al., 2017): for Bayesian VA, FRP=0.852 ± 0.025; for e-ETDRS: FRP=0.826 ± 0.027.Conclusions After 10 trials (30 letters~40 sec), the precision of Bayesian acuity estimates matched that of the e-ETDRS in 30.8 ± 5.4 letters. With more trials, the precision of Bayesian VA continued to improve over that of e-ETDRS.

Bennett, C., Bailin, E., Gottlieb, T., Bauer, C., Bex, P. & Merabet, L. (2018) “Virtual Reality Based Assessment of Static Object Visual Search in Ocular Compared to Cerebral Visual Impairment.” HCI International 2018.

Lu, Z.L., Zhao, Y., Lesmes, L.A., Dorr, M. & Bex, P.J. (2018) “Accuracy in detecting vision changes with visual acuity and contrast sensitivity tests.” Association for Research in Vision and Ophthalmology. Abstract

Purpose
Accurate detection of vision change is important in diagnosing and monitoring visual disease progression and treatment effects. In this study, we evaluated the accuracy of three tests in detecting vision changes: the e-ETDRS (Beck et al., 2003), a new Bayesian visual acuity test (Lesmes, 2017), and qCSF assessment of the contrast sensitivity function (Lesmes, et al, 2010; Hou, et al, 2015).Methods
Monocular visual acuities and contrast sensitivities were measured with the three tests in both eyes of six subjects with normal or corrected-to-normal vision in four simulated vision conditions (no foil and three different levels of Bangerter foil that degraded vision). Each session tested one foil condition. Subjects were tested monocularly with both the new Bayesian visual acuity and e-ETDRS methods 4 times, with random test order of the two methods and two eyes in different blocks. At the end of each session, the contrast sensitivity function of each eye was measured with the qCSF method. An ROC analysis (Hou et al., 2016) was conducted to evaluate the accuracy of the methods in detecting changes in visual acuity and area under the log CSF (AULCSF) associated with different foil conditions. The analysis consisted of computing (1) the average acuity from repeated e-ETDRS measures in each eye and foil condition for each subject; (2) posterior distributions of the estimated VA from the new visual acuity and e-ETDRS tests and AULCSF from the qCSF, (3) the accuracy of detecting acuity changes (averaged across all 16 possible combinations of the four repeats in each condition) and AULCSF changes between each pair of foil conditions; (4) the average accuracy in three ranges of mean acuity changes: 0.02-0.05, 0.05-0.10, and 0.10-0.20 logMAR.Results
For detection of visual function change, the qCSF with 50 trials had the highest accuracy (0.996 ± 0.008, 0.996 ± 0.007, and 0.987 ± 0.020 in the three ranges), followed by the new Bayesian visual acuity test with 45 trials or 135 letters (0.874 ± 0.038, 0.910 ± 0.047, and 0.971 ± 0.036), and lastly by the e-ETDRS test with 30.8 ± 5.4 letters (0.826 ± 0.036, 0.808 ± 0.029, and 0.0.905 ± 0.042). Conclusions Bangerter foils produce deficits in visual function beyond visual acuity. The qCSF was the most accurate test for these deficits, and the new Bayesian VA test is more accurate than e-ETDRS in detecting associated visual acuity changes. 

Jin, Q. Thein, T., Pasquale, L. R., Shen, L. Q., Boland, M. V., Wellik, S. R., De Moraes, C. G., Myers, J. S., Bex, P. J., Baniasadi, N., Li, D., Wang, H Wang, M. & Elze, T. (2018) “Effect modification of refractive error on visual field pattern deviation in glaucoma.” Association for Research in Vision and Ophthalmology. Abstract

Purpose Since glaucoma tends to produce specific visual field defects, the pattern deviation (PD) plot, which shows relative light sensitivity at each VF location, is crucial for the functional diagnosis of glaucoma. We studied effect modification of refractive error on PD values as a function of glaucoma severity.Methods In this retrospective study, the most recent reliable automated VFs (fixation loss≤33%, false negative/positive rates≤20%, SITA Standard 24-2) were selected among all patients from five clinical glaucoma practices. PD values at each VF location were modeled by linear regression with spherical equivalent (SE), mean deviation (MD), and their interaction (SE×MD) as regressors. P values of the slopes were adjusted for multiple comparisons. Results 120,019 eyes of 70,495 patients were selected (medians/[5th;95th] percentiles of age: 66.7 [34.7;86.5] years; MD: -2.7 [-20.2;0.9] dB; SE: -0.4 [-5.75;3.00] D). Fig. 1A shows the isolated effect of SE (i.e. unexplained by MD) on PD. Among all VF locations, 34.6% (predominantly central/infero-temporal) and 36.5% (predominantly peripheral/nasal) showed higher respectively lower PD values with increasing myopia. Fig. 1B illustrates significant interaction effects (SE×MD) on 82.7% of VF locations. With increasing myopia and increasing glaucoma severity, central/nasal VF areas become more depressed and peripheral/temporal areas less depressed. Fig. 2 quantifies the effect of SE on PD separately for the group with at most mildly depressed VFs (MD within ±1dB) vs. severely depressed VFs (MD < -12dB). Significant SE slopes range from -0.11 to +0.04 in mildly depressed VFs and from -0.23 to +0.19 in severely depressed VFs. This means, for severe glaucoma, high myopes (SE: -6D) have up to 2.28 dB lower and up to 2.76 dB higher PDs values than high hyperopes (SE: +6D).Conclusions Ametropia in isolation (Fig. 1A) as well as the interaction between ametropia and glaucoma severity (1B) were found to have significant effects on PD values. The two major retinal nerve fiber bundles, known to be vulnerable to glaucomatous damage, are closer to the fovea with increasing myopia (1C). The interaction results (1B) are consistent with a relative increase of glaucomatous VF loss at the individual locations of these bundles compared to the rest of the visual field. In severe glaucoma, single PDs differ by more than 2 dB between high myopes and hyperopes.

2017

Bex, P.J., Dougherty, L., Forton, A., Knežević, I., Romero-Suarez, L.,  Tella, J. & Welsh, E. (2017) “Generalization of the Flash Distortion Effect.” Presented at the Applied Vision Association.  Abstract

A sequence of faces presented in the peripheral visual field may appear grotesquely distorted (Tangen et al, 2011, Perception, 40, 5, 628), an effect that has been attributed to adaptation of opponent mechanisms in face-space. We examine this phenomenon and explanation with 5 aligned image categories: upright celebrity faces, inverted faces, animal faces, face-like car-front images and different images of the same celebrity; and 4 metrics: distortion occurrence, distortion onset time, detection of physical distortion and distortion matching. Distortion occurrence depended on image category (F(4,20)=20.2, p<0.01) and were less likely to occur in animal or car images  (all p<.01). Onset times were significantly different across image category (F(4,20)=18.2, p<0.01), but were not significantly different for upright, inverted or same celebrity images (all p>0.05). There was no significant difference in the detection of physical distortions (F(4,20)=2.5, p=0.07), however the magnitude of matching distortions differed across image category (F(4,20)=9.5, p<0.01) and were significantly greater in upright faces, inverted faces, and same identity faces than animal or car faces (all p<.05). The spatial scale of perceived distortion corresponded to approximately 1 cycle at the inter-pupillary distance of the faces. These results show that flashed image distortions occur in human, but not non-human face-like images, consistent with the representation of faces in a unique space. However, the presence of distortions in inverted faces and images of the same person is not consistent with opponent dimensions in this space.

Harrison, W. & Bex, P.J. (2017) “On the heterogeneity of visual crowding. “Presented at Vision Sciences Society Abstract

Our ability to identify a visual object in clutter is far worse than predicted by the eyes’ optics and nerve fiber density. Although the ubiquity of such visual impairment, referred to as crowding, is generally well accepted, the appearance of crowded stimuli is debated due in part to the fact that the patterns of perceptual errors made under crowded conditions depend somewhat on the specific task. For example, using stimuli that do not easily combine to form a unique symbol (e.g. letters or objects), observers typically confuse the source of objects and report either the target or a distractor. Alternatively, when continuous features are used (e.g. orientated gratings or line positions), observers often report a feature matching the average of target and distractor features. To help reconcile these empirical differences, we developed a method of adjustment that allows detailed analysis of multiple error categories occurring within the one task. We presented a Landolt C target oriented randomly at 10° eccentricity in the right peripheral visual field in one of several distractor conditions. To report the target orientation, an observer adjusted an identical foveal target. We converted each perceptual report into angular distances from the target orientation and from the orientations of the various distractor elements. We applied new analyses and modelling to these data to quantify whether perceptual reports show evidence of positional uncertainty, source confusion, and featural averaging on a trial-by-trial basis. Our results show that observers reported a distractor orientation instead of the target in more than 50% of trials in some conditions. Our data also reveal a heterogeneous distribution of perceptual reports that depends on target-distractor distance. We conclude that aggregate performance in visual crowding cannot be neatly labelled, and the appearance of a crowded display is probabilistic.

Bauer, C.M. , Bailin, E.S., Bex, P.J. & Merabet, L.B. (2017) “Dorsal attention network activation and diminished attention in Cerebral Visual Impairment.” Presented at the Society for Neuroscience. Abstract

.

Hawlader , O., Kosovicheva, A. & Bex, P.J. (2017) “Sensory Effects of Dichoptic Saccadic Adaptation”. Presented at the 26th Newfoundland Electrical and Computer Engineering Conference.Abstract

.

McKay, R.A., Piano, M.E.F, Bex, P.J., Preston, J.A., Stansfield, B.W., & Simmers, A.J. (2017) “Persistent Visual Impairment in people with Multiple Sclerosis.” Presented at the 13th European Neuro-Ophthalmology Society.Abstract

.

Wang, M, Pasquale, L, Shen, L., Boland, M., Wellick, S. de Moraes, G., Myers, Wang, H. Banaisada, N. Bex, P.J. & Elze, T.  (2017) “Predicting False-Positive Glaucoma Hemifield Test Results by Representative Glaucomatous Visual Field Patterns.” Association for Research in Vision and Ophthalmology.Abstract

 Purpose: Glaucoma Hemifield Test (GHT) is an important functional measure for glaucoma diagnosis. Assuming that glaucomatous visual field (VF) loss is irreversible, a reversal from 2 consecutive GHT outside normal limits (ONL) to within normal limits (WNL) is considered a false-positive result. In this retrospective cohort study, we investigate the relationships between such GHT reversals and the VF loss patterns of the baseline in early-stage glaucoma patients. Methods:From 5 large glaucoma practices, eyes with reliable baseline and two annual follow-up Humphrey VFs (fixation loss≤33%, false negative/positive rates≤20%, SITA Standard 24-2), baseline mean deviation (MD) ≥-3dB and GHT ONL for the first 2 VFs were selected. For each baseline VF, the pattern coefficients are calculated for the 11 glaucomatous VF loss archetypes (ATs) and the normal AT, chosen from all 17 VF ATs identified by [1] (see Fig. 1; AT indices according to [1]). Logistic regression was applied to model the relationship between AT coefficients, MD, pattern standard deviation (PSD), and false-positive GHT. Stepwise regression was performed to select the best predictive model for false-positive GHT. 10 times repeated 10-fold cross validation was applied to evaluate the predictive model performance by the area under the receiver operating characteristic curve (AUC). Results:2,506 eyes were selected for data analysis. 12.1% of the GHTs reversed from 2 consecutive ONLs to WNL at follow-up. Fig. 2a shows the best predictive model (relevant parameters: 8 ATs, MD, and PSD): Higher values of MD and lower values of PSD and all ATs predict GHT false positives with an AUC of 0.717 (95% confidence interval: [0.715, 0.719]; Fig. 2b). A lower similarity to glaucomatous ATs as well as to the normal AT, reflected by negative coefficients, implies a higher probability of false positive GHTs. Conclusions:12.1% of eyes with baseline MD≥-3dB and two consecutive GHT ONL reverse to normality and may therefore be false-positives. Using a previously proposed computational VF decomposition method [1], we show that these false-positives can be statistically predicted by the lack of similarity to typical glaucomatous and typical normal VF patterns at baseline. Our model can be potentially used to recommend patients for VF retest.[1] Elze et al., J. R. Soc. Interface 12:20141118, 2015.

Ferrucci, J., Bex, P. J. & G. McCormack (2017) “Motion Sharpening Processes in Stereoscopic Motion-in-Depth.” Presented at the Association for Research in Vision and Ophthalmology. Abstract

Purpose: Blurred images moving in stereoscopic depth are neurally sharpened more than laterally moving equally blurred binocular images having the same retinal image velocities. We examined whether asymmetric motion sharpening – a retinal process that sharpens the trailing edges of moving ocular images but not the leading edges – caused the difference between stereoscopic and lateral motion sharpening. Methods:10 normally binocular young adults viewed images on an Asus 3D display. The test images were 1°w x 2°h bars of 50% contrast on a dark gray background, with 42’ edge blurs. The bars moved at velocities of 0.5°/s or 2.0°/s. Experiment 1: Monocular test bars appeared for 0.5 sec above or below a fixation cross, at random. To measure asymmetric motion sharpening, the leading edge of a test bar was matched to its 42’ trailing edge by a PEST staircase technique. Matches were obtained for all combinations of eye, motion direction, and velocity. Experiment 2: Binocular images, moving laterally or in stereoscopic depth, were constructed from the perceptually edge-matched ocular images derived from experiment 1, thus removing asymmetric sharpening. Moving test bar blur was then measured by matching its blur, via PEST, to an adjustable-blur static comparison bar on the opposite side of fixation. Experiment 3: Performed like experiment 2, but with test bars constructed from ocular images having physically equal 42’ left and right edge blurs. Results:Mixed ANOVAs evaluated the effects of motion direction, subject, eye, and velocity on match blur. Experiment 1: Leading edge blurs had to be reduced by 19.6% at 2°/s to match trailing edge blurs (F=92.5, p≤0.0001). Experiment 2: Stereoscopic motion sharpening was stronger than conjugate lateral motion sharpening at 2°/s (F=6.68, p=0.036). Experiment 3: Stereoscopic motion sharpening was again stronger than lateral motion sharpening at 2°/s (F=7.11, p=0.032). The relative strength of stereoscopic motion sharpening did not differ between experiments 2 & 3. (F=1.2, p=0.37). Conclusions:(1) Asymmetric motion sharpening does occur with the type of stimuli we used. (2) Motion sharpening is stronger in stereoscopic motion-in-depth than in conjugate lateral motion, with or without edge-matched ocular images. (3) Stronger stereoscopic motion sharpening must be a property of a symmetric sharpening process in cortical stereoscopic motion-in-depth cells.
View poster pdf 

Kosovicheva, A., Kazlas, M., Hunter, D. G. & Bex, P.J. (2017) “What can we learn about strabismus from a 90 second gaze recording?: Rapid evaluation of oculomotor deficits in strabismus.” Presented at the Association for Research in Vision and Ophthalmology.Abstract

Purpose:Clinical evaluation of ocular alignment can be performed with a variety of assessments, including ocular imaging and prism cover tests. However, obtaining precise strabismus measurements with these procedures at a range of gaze postures can require skill or be time-consuming. The purpose of this study was to test the feasibility of using readily available eye tracking equipment for this purpose. We developed and tested a rapid eye tracking based procedure for evaluating ocular alignment and oculomotor deficits in strabismus. Methods:Gaze position was recorded binocularly using an eye tracker in a total of 11 observers (7 strabismic subjects, including 4 with amblyopia, and 4 controls, ages 7 – 50). Subjects were instructed to track a crosshair target that moved between locations on an invisible grid spanning 18° vertically and 24° horizontally for a total gaze recording time of approximately 90 seconds. At each location, subjects viewed the target monocularly for 1000 ms through a pair of shutter glasses. Halfway through each interval, the presentation of the target switched between the two eyes, producing a measurement of differences in gaze position between left eye and right eye targets. In addition, on half the trials, the target remained in the same eye, allowing for a longer measurement of fixation stability for monocular targets. Results:For pairs of monocular targets presented in the same physical location, we observed larger differences in gaze position between left and right eye targets in strabismic subjects (2.84° ± 0.62°) than controls (0.27° ± 0.05°), p = .014. For strabismic subjects, the angular shift in gaze position between pairs of targets correlated with the errors made in an independent dichoptic alignment task (Spearman’s r = 0.82, p = .034). Subjects with strabismic amblyopia additionally showed reduced fixation stability while viewing targets presented to the fellow eye (larger BCEA in the amblyopic compared to fellow eye, p = .010). Conclusions:Together, our results demonstrate that assessment of ocular alignment at a range of gaze postures can be quickly performed using an eye tracking procedure, and that strabismus measurements from this objective gaze recording method show good agreement with subjective measures based on perceptual alignment.
View poster pdf

Shepard, T.G., Hou, F., Bex, P.J., Lesmes, L.A., Lu, Z-L. & Yu, D. (2017) “The quick reading method: its efficiency and accuracy in assessing reading performance in the periphery.” Presented at the Association for Research in Vision and Ophthalmology. Abstract

Purpose:Patients with central vision loss have to rely on their peripheral vision for reading. Accurate assessment of reading performance can help prescribe suitable adaptive devices to the patients. In this study, we develop an adaptive method, quick reading (qR), to measure reading speed in the periphery. While the conventional method is adequate, qR utilizes a Bayesian adaptive framework to select optimal stimuli, thus allowing for an efficient assessment of reading speed in the periphery. Methods:Eight normally-sighted observers participated. We used a rapid serial visual presentation (RSVP) paradigm where words were serially presented at 10° in the lower field. The conventional method involved measuring reading accuracy as a function of exposure duration. Reading speed at a given print size is defined as the duration at which subject’s response is 80% correct. The reading speed versus print size function was estimated by measuring reading speed at five print sizes (a total of 180 trials). In the qR procedure, reading speed versus print size was described by an exponential function with three parameters (asymptotic performance level, print size corresponding to a reading speed of 6 wpm, and a decay constant). Following each trial (50 trials total), posterior distributions of the parameters were updated based on subject’s response, and a stimulus condition (print size and exposure duration) was selected to provide the maximal expected information gain for the upcoming trial. Results:Reading curves (reading speed vs. print size) estimated using the two methods were comparable across observers (area under curve: t(7)=1.87, p=0.10). The conventional data was analyzed using the Bayesian fitting component of qR. A paired-samples t-test was conducted to compare 68.2% credible intervals between the qR and conventional methods. The qR method was more precise (i.e. smaller credible intervals) than the conventional method when considering only 50 conventional trials (p=0.0004) and comparable when 180 conventional trials were included (p=0.11). Conclusions:The current investigation demonstrates that the qR method can adequately measure reading function in the periphery but with higher precision than the conventional method.
View poster pdf

Bex, P.J., Kosovicheva, A., Ferreira, A. & Vera-Diaz, F.A. (2017) “Effects of temporal frequency on binocularity and contrast sensitivity in amblyopia.” Presented at the Vision Sciences Society.Abstract

Amblyopia is associated with a range of visual deficits, including reduced contrast sensitivity, interocular suppression, and impaired stereopsis. Binocular interactions are modulated by temporal frequency in normally-sighted observers; for example continuous flash suppression can be induced by monocular flicker. We examined the effects of temporal frequency on contrast sensitivity and binocular interactions in adults with amblyopia (n = 5) and normally-sighted control subjects (n = 16). For each observer, we estimated the temporal contrast sensitivity function (tCSF) monocularly in each eye using a modified quick CSF procedure (Lesmes et al., 2010) with bandpass-filtered letters at four spatial frequencies (1, 2, 4, and 8 cpd). Results showed greater interocular differences in the area under the tCSF in amblyopes compared to controls, but only at high (4 and 8cpd) spatial frequencies (F = 2.82, p = .047), indicating that amblyopes have a preserved ability to process temporal information at low spatial frequencies. We also evaluated binocularity by measuring interocular suppression and stereoacuity thresholds across four temporal (0, 4, 7.5, and 12 Hz) and spatial (1, 2, 4, and 8 cpd) frequencies. Interocular suppression was estimated by varying the contrast ratio of two dichoptic letters to produce perceptual reports of each letter with equal frequency (Kwon et al., 2014). Stereoacuity thresholds were measured by determining the minimum disparity at which subjects identified a front-depth target with 75% accuracy in a 3AFC task. Across both groups, interocular suppression was lower at higher temporal frequencies (F = 3.50, p = .02), whereas stereoacuity thresholds were unaffected by temporal frequency (F = 0.43, p = .73). Our results point to a temporal dependence of amblyopic deficits, a dissociation between the effects of flicker on interocular suppression and stereopsis, and suggest that temporal modulation may be used to attenuate interocular suppression.
View poster pdf

Ferreira, A., Kosovicheva, A., Bex, P.J. & Vera-Diaz, F.A. (2017) “Temporal Contrast Sensitivity in Myopia.” Association for Research in Vision and Ophthalmology.Abstract

 Purpose:A person’s ability to process and use temporal visual information may have an effect on emmetropization as this affects the spatial image content across the retina. We hypothesize that differences in retinal profile may affect temporal contrast sensitivity in individuals with myopia. Methods:We conducted a cross-sectional study to evaluate the temporal contrast sensitivity function (tCSF) at four spatial frequencies (1, 2, 4 and 8 cpd). Subjects performed a monocular quickCSF procedure (Lesmes et al., 2010) in each eye where they identified a flickering band-pass filtered letter in a 26AFC task. Subjects were 28 young adults (22-31yrs) with best-corrected VA 0.0 LogMAR (20/20) or better in each eye and no binocular or accommodative dysfunction measured by clinical tests. Refractive error was determined by objective and binocular subjective refraction. Subjects were classified as: myopes (SE -1.00 to -8.13D, n=15) or emmetropes (SE +0.63 to -0.25D, n=13). Results:Subjects with myopia showed higher critical flicker frequencies (CFF) than emmetropes (F=6.69, p=0.02), and larger refractive errors were associated with higher CFFs (Spearman’s r=0.42, p=0.03). In addition, we observed a spatial frequency dependent difference in the area under the log contrast sensitivity function (AULCSF) between the refractive groups (F=2.99, p=0.04). Subjects with myopia displayed a higher AULCSF at medium spatial frequencies than emmetropes (2cpd: t=-2.92, p<0.01, 4cpd: t=-1.74, p=0.04). Conclusions:The myopic subjects in our study had low to moderate amounts of myopia. They showed higher CFFs that was related to the amount of myopia. This contrasts with previous results showing lower CFFs in subjects with high myopia (>8.00D; Chen et al, 2000). Myopic subjects in our study also showed high contrast sensitivity at higher temporal frequencies and larger AULCSF for medium spatial frequencies. Overall, our findings suggest that subjects with myopia may have improved, rather than abnormal, temporal vision at higher temporal frequencies than emmetropes.

Vera-Diaz, F.A., Kosovicheva, A., Ferreira, A. & Bex, P.J. (2017) Effect of Temporal Vision on Binocularity in Ametropia.” Presented at the Association for Research in Vision and Ophthalmology Abstract

Purpose:A person’s ability to process and use temporal visual information may affect the binocular balance between the two eyes and their perception of depth (stereopsis). We hypothesize that differences in retinal profile may influence the effect of temporal vision on binocularity in individuals with ametropia. Methods:Forced choice tasks were used to quantify binocularity as a function of flicker (0, 4, 7.5, and 12Hz) and spatial frequency (1, 2, 4, and 8cpd). Interocular Contrast Ratios (ICR) were measured by determining the relative contrast at which subjects reported seeing two dichoptic Sloan letters with equal frequency (Kwon et al, 2014). Stereopsis thresholds were evaluated by determining the minimum disparity at which subjects identified a front-depth target with 75% accuracy. Subjects were 34 young adults (22-31yrs) with best-corrected VA 0.0 LogMAR (20/20) or better in each eye and no binocular or accommodative dysfunction measured by clinical tests. Refractive correction was determined by binocular subjective refraction. Subjects were classified into: Myopes (-0.80 to -9.00D, n=15), Emmetropes (+0.40 to -0.25D, n=13), or Hyperopes (+4.00 to +0.70D, n=6). Results:Deviations from binocularity, quantified as the strength of the difference between the two eyes (difference of the ICR from 50%), were found to be spatial frequency dependent (F=4.24, p<0.01), but not temporal frequency dependent (F=2.07, p=0.11). In contrast, stereoacuity thresholds were both spatial (F=3.00, p=0.04) and temporal frequency (F=3.68, p=0.02) dependent. In addition, we observed temporal and spatial frequency-dependent differences in binocularity between the refractive error groups. A significant three-way interaction indicated that myopes show higher suppression at higher spatial frequencies and lower temporal frequencies than emmetropes (F=2.28, p=0.02). Myopes also showed larger stereoacuity thresholds at 4Hz (t=2.49, p=0.02). Conclusions:Subjects with myopia but otherwise clinically normal vision show a trend towards a temporal frequency-specific reduction in binocularity, as shown by increased stereopsis thresholds and increased suppression. This potential effect of temporal frequency in the ability to perceive depth in myopes may be consistent with previously reported deficits in their magnocellular pathway. However, the specifics of abnormal temporal vision processing in emmetropization are yet to be determined.

Lu, Z-L., Hou, F., Lesmes, L.A., Bex, P.J. & Yu, Deyue (2017) “Bayesian adaptive assessment of reading performance: the quick Reading method.” Presented at the Association for Research in Vision and Ophthalmology. Abstract

Purpose:Reading is a fundamental skill and the reading performance is a key endpoint for quantifying normal or abnormal development and aging. Successful reading performance requires ophthalmic, cognitive and oculomotor proficiency. The deficit or pathology in any of these functions can lead to a deficit in reading performance (Legge et al 1985). Despite its importance for clinical and developmental assessment, existing reading tests are time consuming and difficult to administer. In this study, we propose a novel method, the quick Reading method, for automated measurement of reading speed at multiple letter sizes based on Bayesian adaptive testing (Lesmes, et al., 2010). Methods:A three-parameter exponential function is used to describe the reading speed vs print size function. The quick Reading method selects the optimal test stimulus (print size and presentation duration) by maximizing the expected information gain in each trial and updates the posterior distribution of the parameters of the reading function. The precision and bias of the estimated reading function of a simulated observer obtained using quick Reading were evaluated. Reading functions measured by the conventional (Psi method, Kontzevich & Tyler, 1999) and quick Reading methods in a true/false paradigm (Crossland et al, 2008) were compared in an experiment. Results:The precision of quick Reading method was 0.26, 0.17 and 0.06 log10 unit after 10, 20 and 100 trials, respectively. The bias of the quick Reading method was 0.21, 0.17 and 0.10 log10 unit after 10, 20 and 100 trials, respectively. The estimated reading functions obtained with the conventional and quick Reading methods did not differ significantly (paired t-test, p = 0.184); There were highly correlated (r = 0.969, p = 0.001). The precision of the reading function obtained with 60 quick Reading trials was comparable to that of conventional method with 240 trials. Conclusions:The quick Reading method can be used to precisely and efficiently assess reading performance, with great promise in clinical applications.
View poster pdf

Alberti, C.F. & Bex, P.J.  (2017) “Peripheral depth estimation of disparity-defined targets” Presented at the European Conference on Visual Perception.Abstract

.

Alberti, C.F. & Bex, P.J.  (2017) “Do oculomotor adaptations to a volume scotoma provide functional benefits for binocular vision?” Presented at the Association for Research in Vision and Ophthalmology.Abstract

 Purpose: Binocular eye movements can adjust the projection of a retinal volume scotoma (Arditi, 1988) and modify the retinal disparity of targets in depth. We recently showed (Alberti et al, ARVO 2015) that observers with gaze-contingent simulated independent scotomas make binocular eye movements that move the location of the volume scotoma. We assessed whether such adaptations improve binocular contrast sensitivity in the peripheral visual field. Methods: The contrast sensitivity function was measured with a 26AFC task in which normally-sighted observers (N=6) identified bandpass filtered letters whose spatial frequency and contrast were varied with modified quickCSF algorithm (Lesmes et al, 2010). The letters were positioned 2° in the lower visual field and, in randomly interleaved trials, were either in corresponding retinal locations or displaced horizontally by ±0.25 letter widths to create near or far visual disparity. The gaze contingent scotoma in each eye was a Gaussian windowed (σ=0.5° OS and 1° OD) patch of pink noise, centered on the fovea. Dichoptic presentation of the stimuli was controlled with nVidia 3D glasses synched to a low-latency 144Hz display and eye tracking was measured at 1000Hz with an Eyelink II. Results: The area under the logCSF (AULCSF) was lower for positive or negative disparity stimuli than for stimuli at zero disparity (mean 1.55 vs 1.73, p<0.001), as was peak contrast sensitivity (mean 1.43 vs 1.61, p<0.001). CSF acuity (the highest spatial frequency letter identifiable at full contrast) and other parameters of the CSF did not significantly vary with disparity. Conclusions: In the peripheral visual field, binocular contrast summation requires spatially aligned stimuli and does not occur for disparity-defined targets. Thus oculomotor adaptations that shift the location of a volume scotoma may assist fixation control, but are not associated with functional benefits in contrast sensitivity.

Arango, T., Lu, Z-L., Hou, F., Lesmes, L.A., Yu, D. & Bex, P.J. (2017) “Different reading task measure different reading behaviors.” Presented at the Vision Sciences Society.Abstract

Reading is a primary problem for low vision patients and a functional endpoint for eye disease. However, there is limited agreement on reading assessment methods for clinical outcomes. Many clinical reading tests lack standardized materials for repeated testing and cannot be self-administered, which limit their use for vision rehabilitation and home assessment. We compared three different reading measurements that attempt to address these limitations. Normally-sighted participants (N=13) completed a MNREAD test (Legge et al.,1993), and two different 2AFC reading tasks in counterbalanced order. In one 2AFC task, participants identified whether 5-letter pentagrams, syntactically matched to English, were words or non-words. In the other 2AFC task, participants indicated whether four-word sentences were logically true or false (Crossland et al., 2008). The font size and exposure duration was controlled by a quick Reading algorithm (Lu et al., 2016) that maximized the expected information gain from each trial and updated the posterior distribution of the parameters of the reading function. All lexical stimuli were presented on a computer monitor as black letters on a white background and 2AFC stimuli were pre and post masked with a sequence of Xs. The data from each experiment were fit by an exponential function with parameters for reading acuity (logMAR), acuity reserve and maximum reading speed (words per minute). In all cases, reading speed increased quickly as an exponential function of letter size, in line with previous studies. However, the parameters for the word/non-word, true/false reading and MNREAD methods were significantly different and were not correlated among tasks. These results suggest that these different reading tasks measure different aspects of reading behavior. Evaluating reading performance is an important clinical endpoint and a key quality of life indicator. However, the most effective test that is clinically meaningful is not clear.

Maiello,G.,  Chessa, M., Solari, F. & Bex, P.J.  (2017) “Optimal Combination of Disparity arcoss a log Polar Scaled Visual Field.” Presented at the Vision Sciences Society.Abstract

The human visual system is foveated and the spatial resolution of processing decreases with eccentricity. Sensitivity to stereoscopic disparity varies with spatial frequency and is band-pass tuned in the central visual field. Here, we assess the contributions of different regions of the visual field to disparity processing at different spatial scales. Observers were shown pink noise stereograms, which contained tilted (45° or 135°) sinusoidal corrugations at different modulation spatial frequencies. The corrugation filled a central 21° circular region, or was confined to annular regions spanning 0°-3°, 3°-9° or 9°-21°. Observers indicated the orientation of the corrugation. As expected, disparity sensitivity was band-pass tuned for spatial frequency and the peak shifted to lower spatial frequencies in the peripheral visual field. The envelope of disparity sensitivity for the full field stimulus was the optimal combination of sensitivities across the annular regions. We employed these data to tune and evaluate a foveated log-polar model of disparity processing that mimics the processing stages occurring in primary visual cortex. The model is based on a similar approach we previously developed for the processing of optic flow information directly in the cortical domain [Chessa et al, 2016]. We thus provide a map of disparity sensitivity throughout the visual field and across spatial frequencies, and a biologically plausible model that provides evidence for which neural processing stages are responsible for the topography of disparity sensitivity in man. The common computations of optic flow and stereo disparity in a log polar architecture could be combined to investigate the neural basis of the perception of motion in depth.

Rose, D. & Bex, P.J.  (2017) “Evaluating the Importance of Top-Down “Semantic” Features to Decoding Observer Task from Eye Movements.” Presented at the Vision Sciences Society.Abstract

Since Yarbus (1967), there has been considerable debate over the influence “top-down” factors, such as scene knowledge or task, have on eye movement behavior. Many studies have therefore attempted to decode some feature of an observer’s cognitive state from their eye movements. However, until recently, it has been challenging to embed such “high-level” information directly into images, so few studies have examined the potential role that top-down scene or task features could play in a decoding procedure designed to infer attributes of the observer’s cognition. We evaluated the importance of a novel set of such high-level features with respect to the performance of a classifier built to decode an observer’s task during natural scene inspection. This was achieved by asking subjects to perform one of three tasks while viewing each of 210 natural scene images taken from the LabelMe image database: free viewing, object counting, and inspection made in preparation for a written description of the scene For each subject/trial pairing, three types of features were computed: eye movement features, image salience features of inspected objects and of the entire scene, and a novel set of “semantic” features. These latter described the semantic relatedness of all labeled objects within a scene between themselves, between themselves and a scene gist label, and between the objects sequentially inspected by the subject. Semantic relatedness was calculated using cosine similarity values within a shared vector-space language model (word2vec, Mikolov et al., 2013). A random-forest classifier applied to this data achieved significantly above chance levels of accuracy. Variable importance measures rated these semantic features as among the most important to classifier performance. We therefore suggest that these or similar features should be computed and used in studies that aim to infer an observer’s cognitive state from the pattern of their eye movements in a scene.

Lesmes, L.A., Bittner, A.K., Lu, Z-L., Bex, P.J. & Dorr, M.  (2017) “Distinguishing the contribution of precision and repeatability to vision testing.” Presented at  the Association for Research in Vision and Ophthalmology. Abstract

 Purpose:The promise of visual health monitoring and personalized medicine depends on vision metrics that can precisely track an individual’s vision over time. Common proxies for test precision are based on repeatability, such as the coefficient of repeatability (CoR). However, precision and repeatability are not the same. A test with coarse resolution may be repeatable, but changes in vision within or between individuals are obscured by large steps between test scores. To address this confound, we developed a new Fractional Rank Precision (FRP) metric to evaluate the precision of visual testing, based on concepts of machine learning: how well can an individual be identified in the population distribution of retest measures, based on their initial test measure? We assessed 3 vision tests using FRP: ETDRS visual acuity (VA), Pelli-Robson (PR) contrast sensitivity (CS), and quick Contrast Sensitivity Function (qCSF) testing. Methods:From healthy observers (20-85 years), we obtained 164 monocular and 100 binocular test-retest pairs of qCSF (one week apart). For a broad, scalar summary statistic, we computed the Area Under the Log CSF (AULCSF) from 1.5 to 18 cycles per degree. We also collected 189/180 test-retest pairs from PR CS and ETDRS VA testing. For each test, we computed CoR and FRP: the rank of the retest of a subject when all subjects’ retests are sorted by their similarity to a subject’s initial test, averaged across all subjects. FRP ranges from .5 (chance) to 1.0 (perfect identification of test from retest for each subject). We also recomputed FRP for increasing quantization, i.e. rounding of values to coarse step sizes. Results:CoR and FRP were .214 and .844 (AULCSF), .243 and .721 (PR CS), and .149 and .718 (ETDRS VA), respectively. As expected, increasing quantization reduced FRP. The precision of AULCSF was reduced to that of unmodified (non-quantized) PR CS and ETDRS VA, when strong quantization collapsed the AULCSF population distribution to only 5 step-sizes. Conclusions:The FRP metric is sensitive to a test’s resolution (step-size), variability (CoR), and dynamic range. Despite apparently better repeatability (lower CoR), the precision of ETDRS VA was similar to that of PR CS. The AULCSF provides highest FRP despite intermediate CoR, due to small step-sizes and low variability relative to its range. These features may be useful to detect visual changes in clinical trials and clinical practice.

Vera-Diaz, F. A., Kosovicheva, A., Maiello, G. A., Ferreira, A., Walker, L., & Bex P. J. (2017) “Binocularity in Myopia.” Presented at the International Myopia Conference.

2016

Birch, E.E. Sarah E. Morale, Reed M. Jost, Angie De La Cruz, Krista R. Kelly, Yi-Zhong Wang, & Bex, Peter J.  (2016) “Assessing Suppression in Amblyopia with a Dichoptic Eyechart.” Presented at the American Association for Pediatric Ophthalmology and Strabismus meeting, Vancouver BC, Canada

Walker, L., Maiello, G. Bex, P.J.  & Vera-Diaz, F. A. (2016) “Myopes’ Ability to Perceive Blur Across the Visual Field.” Presented at the Association for Research in Vision and Ophthalmology, 2016.

Aiai K. R., Bex, P., Gardiner, M., Kloek, C., Chang, P., Shen, L., Turalba, A., Pasquale, L.R. and Song, B. (2016) “Comparison of iPad-Based Visual Function Tests for the Detection of Early Manifest Glaucoma Meredith.” Presented at the Association for Research in Vision and Ophthalmology, 2016.

Rose, D. and Bex, P.J. (2016) “Transfer of Peripheral Fixation Training Across Retinal Eccentricities.” Presented at the Vision Sciences Society, 2016.

Kosovicheva, A. and Bex, P.J. (2016) “What color was it? Probing the dynamics of word processing during reading.” Presented at the Vision Sciences Society, 2016.

Carroll, S., Maiello, G., Harrison, W. and Bex, P.J. (2016) “Myopic Eyes See Better in a Crowd.” Presented at the Vision Sciences Society. Abstract

The eyes of myopes are larger, both wider and especially longer, than the eyes of normally-sighted, emmetropic individuals. Even with best-corrected foveal vision, myopic peripheral vision is generally found to be worse than emmetropic peripheral vision. However, the functional vision of the periphery is limited not by acuity or contrast sensitivity but by visual crowding, the phenomenon by which visible targets become harder to identify in clutter. Because of the stretched myopic retina, objects projected onto the peripheral retinal surface of myopic eyes subtend the same retinal angle, but are spaced farther apart than on an emmetropic retina. We ask whether retinal angle or retinal spacing determines crowding. We measured letter acuity as well as radial and tangential crowding zones at 5, 10, and 15 degrees of eccentricity in emmetropic and best-corrected myopic observers. Consistent with previous results, peripheral letter acuity was worse in myopic than emmetropic subjects. We also confirmed the radial/tangential crowding asymmetry in both myopic and emmetropic observers: radial crowding zones were larger than tangential crowding zones by a factor of 2. Critically, our data show that both radial and tangential spatial interference zones in myopic eyes are smaller than in emmetropic eyes. This finding suggests that crowding zones may be determined by retinal spacing rather than by retinal angle. Although myopia is generally thought of as a visual impairment, and clearly hinders sensory aspects of visual function, our data suggest that the elongated retinal surface of myopic eyes may provide a functional benefit by extending the spacing within which peripheral targets are crowded. Our findings raise the possibility that at the supra-threshold object sizes and contrasts that dominate natural vision, interactions between central and peripheral vision might differ between myopes and emmetropes. These differences might play a neglected role in the development and progression of myopia.
 View poster pdf

Elze, T., Shen, L.Q., Wang, M., Boland, M.V., Wellick, S., de Moraes, G., Myers, J., Bex, P. J. and Pasquale, L. R. (2016) “The effect of ametropia on glaucomatous visual field loss.” Presented at the Association for Research in Vision and Ophthalmology, 2016.

Wang, M., Pasquale, L. R. Shen, L.Q., Boland, M.V., Wellick, S., de Moraes, G., Myers, J., Bex, P. J. and Elze, T. (2016) “Impact of natural blind spot location on perimetry.” Presented at the Association for Research in Vision and Ophthalmology, 2016.

Maiello, G., Kwon, M.Y. and Bex P.J. (2016) “3 Dimensional Binocular eye and hand coordination in normal vision and with simulated visual impairments.” Presented at the Vision Sciences Society.Abstract

It is well known that the motor systems controlling the eyes and the hands are closely linked when executing tasks in peripersonal space. We examine how this coordination is affected by binocular and asymmetric monocular simulated visual impairment. In a stereoscopic display, human observers were required to closely track with their gaze a 1 degree Gabor patch moving in three dimensions on a 1/f noise background. The movement of the Gabor patch was either directly controlled by the observer’s unseen hand in real time; or followed their hand movements executed in a previous trial. Hand position was recorded with a Leap Motion hand tracker, and gaze position was recorded with an Eyelink 1000 eye tracker. We simulated visual impairments by Gaussian blurring the visual stimuli independently in each eye. Tracking accuracy was defined as the average correlation coefficient between gaze position and target position along the fronto-parallel plane (pursuit) or the sagittal plane (vergence). We observed a critical blur level up to which pursuit and vergence eye movements maintained fronto-parallel and sagittal tracking accuracy independent of blur level. Monocular blur affected fronto-parallel tracking less than binocular blur, however small amounts of monocular blur impaired tracking in depth much more than binocular blur. Target tracking was more accurate when observers were directly controlling the stimulus than when they tracked a previous hand movement and this benefit was more pronounced with degraded visual input. This suggests that under conditions of visual uncertainty, proprioceptive information is weighed more heavily. Our results confirm that the motor control signals that guide hand movements are utilized by the visual system to plan eye movements. Our findings suggest that hand-eye coordination might be monitored to better understand functional impairments associated with eye disease and may be employed to rehabilitate an array of monocular and binocular visual impairments.

Bex, P. J., Dorr, M., Lashkari, K., Lesmes, L., Lu, Z.L. and Wiecek, E. (2016) “Home-Based Self-Assessment of the Contrast Sensitivity Function in Age-Related Macular Degeneration.” Presented at the Association for Research in Vision and Ophthalmology, 2016. View poster pdf

Hui, W., Lesmes, L., Dorr, M., Lu, Z.L., Elze, T. and Bex, P. J. (2016) “Characterizing monocular and binocular contrast sensitivity in corrected and uncorrected myopia.” Presented at the Association for Research in Vision and Ophthalmology.

Lesmes, L., Lu, Z.L., Bex, P. J. and Dorr, M., (2016) “A Survey of Contrast Sensitivity Deficits in Visual Neuropathology.” Presented at the Association for Research in Vision and Ophthalmology.

Wang, M, Jin, Q, Wang, H, Pasquale, L, Bex, P, Ishikawa, H, Schuman, J, Wollstein, G,  Elze, T.  (2016) “Quantifying Positional Variation of Retinal Blood Vessels in Glaucoma.” Presented at the Association for Research in Vision and Ophthalmology.

Baniasadi, N, Jin, Q, Wang, H, Wang, M, Shen, L, Pasquale, L, Bex, P, Ishikawa, H, Schuman, J, Wollstein, G, Elze, T.  (2016) “The effect of optic nerve head torsion and position of retinal blood vessels on optical coherence tomography (OCT) retinal nerve fiber layer (RNFL) abnormality patterns in myopic glaucoma patients.” Presented at the Association for Research in Vision and Ophthalmology.

Gibaldi, A., Maiello, G., Bex, P. and Sabatini, S. (2016) “Modelling Short-Latency Disparity-Vergence Eye Movements Under Dichoptic Unbalanced Stimulation.” Presented at the MODVIS Workshop, Vision Sciences Society.

Maiello, G., Chessa, M. Bex, P. and Solari, F. (2016) “Can Neuromorphic Computer Vision Inform Vision Science? Disparity Estimation as a Case Study” Presented at the MODVIS Workshop, Vision Sciences Society.

Crucilla, S., Rose, D., Kalia, A., Bex, P.J., and Sinha, P.  (2016) “Mechanisms underlying simultaneous brightness induction: Early and innate.” Presented at the European Conference on Visual Perception.

Maiello, G., Gibaldi, A., Sabatini, S. and Bex, P.J.  (2016) “Vergence Eye Movements to Unbalanced Dichoptic Visual Stimuli.” Presented at the European Conference on Visual Perception 2016


2015

P. J. Bex, (2015) “The Impact of Simulated Central Scotomas on Perception and Eye Movements.” Presented at the Association for Research in Vision and Ophthalmology, Denver.

W. Harrison, M.A. Johnson and P. J. Bex, (2015) “Attentional resolution is not the exclusive limit on visual awareness.” Presented at the Vision Sciences Society.

G. Maiello, W. Harrison, F. Vera-Diaz and P. J. Bex, (2015) “Perceptual Consequences of Elongated Eyes.” Presented at the Vision Sciences Society. View poster pdf

Ayeni, A. J., W. Harrison and P. J. Bex, (2015) “Filling in of Kanizsa-style illusory figures is under top-down control.” Presented at the Vision Sciences Society.

Rose, D. and P. J. Bex, (2015) “Peripheral Oculomotor Control Training in Healthy Individuals: Effects of Training and Training Transfer.” Presented at the Vision Sciences Society. View poster pdf

P.J. Bex, A.J. Ayeni and E. Wiecek, (2015) “Word and Sentence Level Spatial Information In Reading.” Presented at the Vision Sciences Society. View poster pdf

E. Wiecek, and Bex, P. J. (2015) “Binocular Microperimetry in Simulated Asymmetric Bilateral Scotomas.” Presented at the Association for Research in Vision and Ophthalmology.

Alberti, C. and Bex, P. J. (2015) “Simulation of binocular scotoma and compensatory strategies.” Presented at the Association for Research in Vision and Ophthalmology. View poster pdf

M. Dorr, Lesmes, L.A.L., Elze, T., Wang, H., Lu, Z. L. and P. J. Bex, (2015) “Average Precision as a test-­retest reliability measure: a quick CSF study on myopia.” Presented at the Association for Research in Vision and Ophthalmology.

Kerber, K.L., Maiello, G., Thorn, F., Bex, P.J., Vera-Diaz, F.A., (2015) “Myopes’ ability to accurately accommodate to blur cues in virtual 3D images.” Presented at the Association for Research in Vision and Ophthalmology.

Vera-Diaz, F.A., Maiello, G., Kerber, K.L., Thorn, F. and Bex, P.J., (2015) “Myopes’ ability to discriminate and detect blur.” Presented at the Association for Research in Vision and Ophthalmology.

Elze, T., Jin, Q., Wollstein, G., Bex, P.,  (2015) “The impact of blood vessel locations on abnormality profiles of Spectral-Domain Optical Coherence Tomography measurements”. Presented at the Association for Research in Vision and Ophthalmology.

Jin, Q., Wollstein, G., Pasquale, L., Shen, L., Bex, P. J., Elze, T. (2015) “Motion artifacts and missing data in Spectral-Domain Optical Coherence Tomography measurements.” Presented at the Association for Research in Vision and Ophthalmology.

Dorr, M., Wiecek, E., Lashkari, K., Lu, Z. L., Lesmes, L. A., & Bex, P. J. (2015). “Comprehensive mHealth assessment of visual function.” Presented at the German Society for Biomedical Engineering. Lubeck.

Cai, S., Elze, T., Bex, P.J., Wiggs, J.L., Pasquale, L.R., Shen, L.Q., (2015) “Clinical Correlates of Computationally Derived Visual Field Defect Archetypes in Patients from a Glaucoma Clinic.” Harvard Medical School Annual Conference. View poster pdf

Harrison, W. J. and Bex, P. J. (2015) “A unifying model of visual crowding.” Presented at the European Conference of Visual Perception. View poster pdf

Vera-Diaz, F. Weifenback, S., Kwon, M. & Bex, P. J. (2015) “Novel Methods to Clinically Quantify Suppression and Stereopsis Thresholds in Studies of Amblyopia.” Presented at the American Academy of Optometry, New Orleans. View poster pdf

Dakin, S., Bex, P. J. & Turnbull, P. (2015) “Automated Contrast Sensitivity Assessment using the OKN.” Presented at the American Academy of Optometry, New Orleans.

Weifenback, S., Kwon, M. Vera-Diaz, F. & Bex, P. J. (2015) “Methods to quantify spatial frequency dependent suppression and stereopsis.” Presented at the American Academy of Optometry, New Orleans. View poster pdf

M. Dorr, M. Wille, T. Viulet, E. Sanchez, P. Bex, Lu, Z., L. Lesmes (2015) “Next-generation vision testing: the quick CSF.” Presented at the German Society for Biomedical Engineering, Lübeck, Germany.

Bex, P.J. Wiecek, E., Dakin, S.C. and Kwon (2015) “Quantitative Assessment of Inter-Ocular Suppression.” Presented at the Lasker/IRRF Initiative on Amblyopa, J. Erik Jonsson Center, Woods Hole, Massachusetts


2014

T. Elze, P. Benner, L. Shen, and P. J. Bex, “The Spatial Configuration of Visual Field Measurement Locations in Glaucoma.” Presented at the Association for Research in Vision and Ophthalmology, 2014.

M. Bossi, E. J. Anderson, V. K. Tailor, P. J. Bex, J. A. Greenwood, A. Dahlmann-noor, and S. C. Dakin, “An Exploratory Study of a Novel Home-Based Binocular Therapy for Childhood Amblyopia,” Presented at the Association for Research in Vision and Ophthalmology, 2014.

W. J. Harrison and P. J. Bex, “Contour perception across time and eye movements.” Presented at the Vision Sciences Society, 2014. View poster pdf

M. Y. Kwon, E. Wiecek, S. C. Dakin, and P. J. Bex, “A novel method to quantify spatial-frequency dependent  binocular imbalance in amblyopia.” Presented at the Vision Sciences Society, 2014.

G. Maiello, W. Harrison, and P. J. Bex, “Dichoptic saccadic adaptation.” Presented at the Vision Sciences Society, 2014.

G. McCormack, J. Van Cura, and P. J. Bex, “Neural Sharpening of Images Moving in Stereoscopic Depth.” Presented at the Association for Research in Vision and Ophthalmology, 2014.

T. Olsen, T. Langaas, P. J. Bex, S. J. Gilson, and R. C. Baraas, “The effect of Omega-3 status on contrast sensitivity in healthy middle-aged Norwegians.” Presented at the Association for Research in Vision and Ophthalmology, 2014.

M. Piano, A. J. Simmers, and P. J. Bex, “Perceptual Visual Distortions in Juvenile Amblyopes.” Presented at the Vision Sciences Society, 2014.

R. L. Woods, D. R. Saunders, P. J. Bex, and D. Rose, “Measuring information acquisition to evaluate impairment and rehabilitation.” Presented at the Low Vision 2014, Melbourne, 2014.

L.-G. Bogfjellmo, P. J. Bex, and H. K. Falkenberg, “Direction discrimination of global motion low contrast patterns in school aged children.” Presented at the Association for Research in Vision and Ophthalmology, 2014.

[W. J. Harrison, G. Maiello and P. J. Bex, “Illusory objects are altered by saccadic eye movement preparation.” Presented at the Society for Neuroscience, 2014.

T. S. A. Wallis, M. Dorr, and P. J. Bex, “A Bayesian multilevel modeling approach to characterizing contrast sensitivity in naturalistic movies.” Presented at the European Mathematical Psychology Conference, 2014.

Cai, S., Elze, T., Bex, P. J., Pasquale, L., & Shen, L. (2014). “Clinical Significance of Computationally Derived Glaucomatous Visual Field Defects.”

Bex, P. J., Keynote Beibu Gulf Optometry and Vision Forum (2014).


2013

P. J. Bex, “Motion perception in ocular disease and visual dysfunction.” Presented at the Association for Research in Vision and Ophthalmology, Seattle, 2013.

M. Bossi, E. J. Anderson, V. K. Taylor, P. J. Bex, J. A. Greenwood, A. Dahlmann-noor, and S. C. Dakin, “A novel home-­‐based binocular therapy for childhood anisometropic amblyopia.” Presented at the British Congress of Optometry and Vision Science, Glasgow, UK, 2013.

Y. Chee, T. Elze, L. Pasquale, D. Rhee, P. J. Bex, and L. Shen, “Deviation between horizontal and vertical SD-OCT macular thickness values and associated scan errors,” presented at the Association for Research in Vision and Ophthalmology, 2013.

X. Chen, L. A. Lesmes, J. Wallis, T. S. A. Wallis, M. Jackson, and P. J. Bex, “Analyses of contrast sensitivity assessments over time: A pilot study.” Presented at the Association for Research in Vision and Ophthalmology, Seattle, 2013.

M. Dorr, L. A. Lesmes, Z. L. Lu, and P. J. Bex, “Rapid and precise assessment of the temporal contrast sensitivity function on an iPad.” Presented at the European Conference on Visual Perception, 2013.

M. Dorr, L. A. Lesmes, Z. L. Lu, and P. J. Bex, “iPad-Based Quick CSF Implementation to Assess Effects of Dioptric Blur on Contrast Sensitivity.” Presented at the Association for Research in Vision and Ophthalmology, 2013. View poster pdf

M. Dorr, A. Schutz, K. Gegenfurtner, E. Barth, and P. J. Bex, “Eye movements in natural movies under spatio-temporal blur.” Presented at the European Conference on Eye Movements, 2013.

T. Elze, P. Benner, and P. J. Bex, “Novel Approaches to Non-parametric Adaptive Sampling of Psychometric Functions.” Presented at the Asia Pacific Conference on Vision, Suzhou, China, 2013.

T. Elze, P. Benner, L. Pasquale, L. Shen, and P. J. Bex, “A Spatial Model of Visual Fields with Applications to Adaptive Sampling.” Presented at the European Conference on VIsual Perception, Bremen, 2013.

T. Elze, L. Pasquale, L. Shen, A. Turalba, T. Chen, D. Rhee, J. Wiggs, C. L. Grosskreutz, S. Brauner, and P. J. Bex, “Patterns in Glaucomatous Visual Field Loss: Components, Prototypes and Archetypes.” Presented at the 7th Annual Offfice for Research Career Development, 2013.

T. Elze, L. Pasquale, L. Shen, A. Turalba, T. C. Chen, D. Rhee, J. L. Wiggs, C. L. Grosskreutz, S. Brauner, and P. J. Bex, “Finding Patterns in Glaucomatous Visual Field Loss: Components, Prototypes, and Archetypes.” Presented at the Association for Research in Vision and Ophthalmology, Seattle, 2013. View poster pdf

W. J. Harrison, Kwon, M., and P. J. Bex, “The influence of eye movements on contrast sensitivity and gain response in peripheral vision.” Presented at the European Conference on Visual Perception, Bremen, 2013.

M. Jackson, J. Wallis, P. Wicks, and P. J. Bex, “Online Assessment of Contrast Sensitivity and Hallucinations in Parkinson’s Disease.” Presented at the Association for Research in Vision and Ophthalmology, Seattle, 2013.

M. Y. Kwon, L. A. Lesmes, A. Miller, M. Kazlas, M. Dorr, D. G. Hunter,, Z. L. Lu, and P. J. Bex, “Rapid assessment of core visual deficits in amblyopia.” Presented at the Association for Research in Vision and Ophthalmology, 2013. View poster pdf

Kwon, M., L. A. Lesmes, M. Kazlas, D. G. Hunter,, Z. L. Lu, and P. J. Bex, “Novel methods to assess core visual deficits in amblyopia.” Presented at the New England Ophthalmological Society Meeting, 2013.

Kwon, M., L. A. Lesmes, A. Miller, M. Kazlas, M. Dorr, D. G. Hunter,, Z. L. Lu, and P. J. Bex, “Contrast sensitivity deficits in amblyopia.” Presented at the European Conference on Visual Perception, 2013.

Kwon, M., L. A. Lesmes, A. Miller, M. Kazlas, D. G. Hunter,, Z. L. Lu, and P. J. Bex, “The effect of interocular luminance differences on eye movements in amblyopia and normal vision.” Presented at the European Conference on Eye Movements, 2013.

L. A. Lesmes, M. Jackson, J. Wallis, and P. J. Bex, “The Reliability of the quick CSF Method for Contrast Sensitivity Assessment in Low Vision.” Presented at the Association for Reseach in Vision and Ophthalmology, 2013.

G. Maiello, C. Manuella, S. Fabio, and P. J. Bex, “Stereoscopic fusion with gaze-contingent blur.” Presented at the European Conference on Visual Perception, 2013. View poster pdf

G. Maiello, C. Manuella, S. Fabio, and P. J. Bex, “The Contribution of Perspective, Blur and Disparity to Depth Perception in Natural Vision.” Presented at the Association for Reseach in Vision and Ophthalmology, 2013.

M. Piano, A. J. Simmers, and P. J. Bex, “The character of monocular and binocular perceptual visual distortions in adult amblyopia,” presented at the British Congress of Optometry and Vision Science, Glasgow, UK, 2013.

M. Piano, A. J. Simmers, P. J. Bex, and S. Jeon, “Perceptual visual distortions in amblyopia and their stability over time.” Presented at the European Conference on Visual Perception, Bremen, 2013.

D. R. Saunders, P. J. Bex, and R. L. Woods, “Measuring information acquisition during viewing of dynamic scenes using free, natural-language descriptions.” Presented at the Vision Sciences Society, Naples, Fl, 2013.

J. A. Solomon, P. J. Bex, and S. C. Dakin, “Sampling Efficiency and Internal Noise for Summary Statistics.” Presented at the European Conference on Visual Perception, 2013.

T. S. A. Wallis, M. Dorr, and P. J. Bex, “The contrast response function of the human visual system to freely-viewed natural movies.” Presented at the European Conference on Visual Perception, Bremen, 2013.

R. S. Watson, R. Legge, A. J. Simmers, and P. J. Bex, “Illusory Figure Perception in Amblyopes.” Presented at the British Congress of Optometry and Vision Science, Glasgow, UK, 2013.

E. Wiecek, S. C. Dakin, K. Lashkari, and P. J. Bex, “Qualitative and Quantitative Assessment of Metamorphopsia in Retinopathy Patients.” Presented at the Association for Research in Vision and Ophthalmology, 2013.

Vera-Diaz FA, Kerber KL, Thorn F, and Bex PJ. “Peripheral Awareness and Attention in Myopia”. Presented at the 14th International Myopia Conference, Asilomar, CA. August 20th 2013


2012

S. F. Abscar, D. Cyr, A. D. Proia, M. T. Malik, P. J. Bex, and K. Lashkari, “Identifying the Roles of Interferon-Gamma Inducible Chemokines in Progression of Age-related Macular Degeneration (AMD).,” presented at the Association for Research in Vision and Ophthalmology, 2012.

L. Bogfjellmo, H. Falkenberg, and P. J. Bex, “Global motion perception at low speed is reduced due to increased internal noise and reduced sampling efficiency in young, mid-aged and old observers,” presented at the European Conference on Visual Perception, Sardinia, Italy, 2012.

M. Dorr, L. A. Lesmes, L. To, and P. J. Bex, “High-precision psychophysics on the iPad,” presented at the European Conference on Visual Perception, Sardinia, Italy, 2012.

T. Elze, C. P. Taylor, and P. J. Bex, “Organic Light-Emitting Diode Monitors in Vision Science,” presented at the European Conference on Visual Perception, 2012.

T. Elze and P. J. Bex, “Modern Display Technology in Vision Science: Assessment of OLED and LCD Monitors for Visual Experiments,” presented at the Asia Pacific Vision Conference, 2012.

A. Kalia, L. A. Lesmes, M. Dorr, P. J. Bex, T. Gandhi, G. Chatterjee, and P. Sinha, “Measurements of Contrast Sensitivity Functions Show Recovery from Extended Blindness,” presented at the European Conference on Visual Perception, 2012.

L. A. Lesmes, J. Wallis, M. Jackson, and P. J. Bex, “Clinical Application Of A Novel Contrast Sensitivity Test To A Low Vision Population: The Quick CSF Method,” presented at the Association for Research in Vision and Ophthalmology, 2012.

M. Piano, A. J. Simmers, and P. J. Bex, “Assessment of global distortion and perceptual magnification in amblyopia,” presented at the British Congress of Optometry and Visual Science, Bradford, 2012.

D. R. Saunders, P. J. Bex, and R. L. Woods, “A Novel, Objective Measure of Information Acquisition from Video is Correlated with Blur.,” presented at the Association for Research in Vision and Ophthalmology, 2012.

J. Wallis, P. J. Bex, L. A. Lesmes, T. S. A. Wallis, and M. Jackson, “Contrast Sensitivity As A Predictor Of Central Field Loss,” presented at the Association for Research in Vision and Ophthalmology, 2012. View poster pdf

T. S. A. Wallis, M. Dorr, and P. J. Bex, “Sensitivity to gaze-contingent spatial distortions in freely-viewed movies,” presented at the European Coference on Visual Perception, Sardinia, Italy, 2012.

S. G. Wardle, P. J. Bex, J. R. Cass, and D. Alais, “Stereoacuity across the visual field: An equivalent noise analysis,” presented at the European Conference on Visual Perception, 2012.

E. Wiecek, S. C. Dakin, and P. J. Bex, “Metamorphopsia and Visual Acuity,” presented at the European Conference on Visual Perception, 2012.

E. Wiecek, M. Jackson, and P. J. Bex, “Visual Search with Image Enhancements in Age-Related Macular Degeneration,” presented at the Association for Research in Vision and Ophthalmology, 2012.

R. L. Woods, P. J. Bex, and D. R. Saunders, “Free Recall as an Objective Measure of Information Acquisition from Video.,” presented at the Association for Research in Vision and Ophthalmology, 2012.


2011

P. J. Bex, L. Pasquale, and S. C. Dakin, “Glaucoma patients demonstrate elevated neural noise in sectors of the Humphrey Visual Field with normal threshold sensitivity,” presented at the Association for Research in Vision and Ophthalmology, 2011.

Bogfjellmo, H. K. Falkenberg, and P. J. Bex, “How does directional noise limit global sensitivity in ageing?,” presented at the European Conference on Visual Perception, 2011.

J. Cass, S. C. Dakin, and P. J. Bex, “Dichoptic suppression of flanking stimuli breaks crowding,” presented at the European Conference on Visual Perception, 2011, p. 35.

M. Dorr and P. J. Bex, “A gaze-contingent display to study contrast sensitivity under natural viewing conditions,” presented at the Human Vision and Electronic Imaging XVI, 2011.

M. Dorr and P. J. Bex, “Natural vision is geotopic,” presented at the European Conference on Visual Perception, Toulouse, 2011, vol. 40, p. 29.

M. Dorr and P. Bex, “Peri-Saccadic Visual Sensitivity while Freely-Viewing Natural Movies,” presented at the Vision Sciences Society, 2011, vol. 11, p. 517.

H. Galperin, D. Lisitsyn, P. Bex, and J. Fiser, “Perception of Motion in Natural Scenes,” presented at the Vision Sciences Society, 2011, vol. 11, p. 725.

J. Greenwood, V. Tailor, A. Simmers, J. Sloper, G. Rubin, P. Bex, and S. Dakin, “Links between acuity, crowding and binocularity in children with and without amblyopia,” presented at the Association for Research in Vision and Ophthalmology, 2011, vol. 11, p. 405.

L. A. Lesmes, T. S. A. Wallis, and P. Bex, “Response bias contributes to visual field anisotropies for crowding in natural scenes,” presented at the Vision Sciences Society, 2011, vol. 11, p. 1156.

N. Schneider, P. J. Bex, E. Barth, and M. Dorr, “An open-source low-cost eye-tracking system for portable real-time and offline tracking,” presented at the Novel Gaze-Controlled Applications, 2011.

G. Selig, D. Lisitsyn, P. Bex, and J. Fiser, “The diagnostic features used for recognizing faces under natural conditions,” presented at the Vision Sciences Society, 2011, vol. 11, p. 614.

C. Taylor and P. Bex, “Efficient integration of local perceived blur in discrimination and matching,” presented at the Vision Sciences Society, 2011, vol. 11, p. 13.

T. Wallis and P. Bex, “Image Correlates of Peripheral Contour Discrimination in Natural Scenes,” presented at the Vision Sciences Society, 2011, vol. 11, p. 62.

E. Wiecek, L. Pasquale, and P. J. Bex, “Effects of peripheral visual field loss on eye movements during visual search,” presented at the Association for Research in Vision and Ophthalmology, 2011.