Walking safely through highly crowded environments is a significant challenge for individuals with cerebral visual impairment (CVI). Yet current ophthalmic examinations do not capture functional visual difficulties related to safe mobility. We developed an immersive virtual reality (VR)-based task that tracked eye gaze behaviors within dynamic areas of interest to assess pedestrian collision detection, avoidance, and associated visual scanning in CVI (n = 12) compared to control (n = 14) participants. Subjects walked through a simulated shopping mall populated with crowds of varying densities. The testing scenario was presented using a head-mounted display with integrated eye tracking, and locomotor, behavioral, and visual scanning responses were recorded. Compared to controls, CVI participants exhibited a slower mean preferred walking speed. They were also less likely and slower to detect target (colliding) pedestrians and were more likely to make a collision. CVI participants were also slower in making their first fixation and followed a larger visual scan path to find the target pedestrian. They also spent more time fixating on non-target compared to target pedestrians. Finally, CVI participants showed greater variability in their performance (including pathing deviations), reflecting a range of individual strategies, and maintained a larger walking safety margin (spatio-temporal envelope). These results provide objective evidence of mobility and associated gaze behaviors in CVI during navigation through highly crowded environments.

PURPOSE:To compare novel multi-periscopic prisms (MPP; 42° field expansion) to commercial Fresnel peripheral prisms (FPP; 30° field expansion) as mobility devices to assist with pedestrian collision hazard detection when walking. METHODS:Participants with hemianopia without neglect or cognitive decline were recruited at 7 sites in a single-masked crossover trial, allocated by minimization to receive either MPP or FPP first. Pedestrian collision detection was evaluated after 4-weeks of home-use of each type of prisms. The test simulated walking through a busy shopping mall with multiple pedestrians; in each trial, one pedestrian could be on a collision course, bearing angle of ±20° or ±40°, or there could be no colliding pedestrian. The main outcome, improvement with prisms (yes/no), was defined as blind-side detection rates that were significantly higher with than without prisms at the same visit. RESULTS:Of 64 participants screened, 38 were eligible and 63% (24/38) showed improvement with at least one type of prism glasses. There were no differences between the proportions of participants showing improvement with MPPs and FPPs for either 40° blind-side pedestrians (MPPs 31%, 12/38 vs. FPPs 26%, 10/38; difference 5%; 95% CI -14% to 24%, p = 0.75) or 20° blind-side pedestrians (MPPs 37%, 14/38 vs. FPPs 42%, 16/38; difference -5%; 95% CI -24% to 14%, p = 0.75). Of the 32 participants who completed both crossover periods, 13 (41%) preferred MPPs whereas 19 (59%) preferred FPPs (p = 0.38). MPPs were preferred for image clarity and extent of field expansion and FPPs for cosmetic appearance. Both MPPs and FPPs reduced (p < 0.01) perceived difficulty when walking in unfamiliar areas, crowded situations, detecting objects and people on the blind side, the most difficult situations without prisms. CONCLUSIONS:Both MPPs and FPPs improved blind-side collision detection at large (40°) as well as smaller (20°) bearing angles. These positive results strengthen the evidence base for FPPs and establish the potential of MPPs as mobility devices for patients with hemianopia. TRIAL REGISTRATION:ClinicalTrials.gov NCT04827147, registered on 04/01/2021.

Homonymous hemianopia (HH) patients report difficulties in avoiding collisions with other pedestrians. We evaluated pedestrian collision detection and avoidance behaviors in HH patients and healthy controls using a novel virtual reality (VR) walking with pedestrians, which enables natural walking behavior in an empty real-world corridor while viewing an immersive VR environment (shopping mall with colliding and other pedestrians) presented in a head-mounted display (HMD). Critically, it measures avoidance maneuvers in addition to collision detection. Colliding and non-colliding pedestrian scenarios were developed for Meta Quest 2 using Unity. Ten normal vision (NV) subjects and 12 HH subjects detected and avoided collisions with virtual approaching and overtaken pedestrians initialized at bearing angles of 20, 40, and 60 degrees, with planned time-to-collision of 6 seconds in each trial. HH subjects were less likely to detect and more likely to collide with pedestrians than NV, particularly for blind-side targets. Response times did not differ between groups but were faster for overtaken pedestrians. HH subjects also biased their head rotations toward the blind side and more after detection compared to before. Collision avoidance difficulties as reported by HH subjects, which clinical measures fail to capture, were recorded and analyzed with objective measures. These metrics may offer further insights into the underlying mechanisms driving collision avoidance behaviors. Our HMD-VR collision detection and avoidance paradigm enables natural walking behaviors and offers an affordable, objective assessment tool that may be adopted by clinicians for mobility enhancement and rehabilitation.

SIGNIFICANCE Performance-based outcome measures are crucial for clinical trials of field expansion devices. We implemented a test simulating a real-world mobility situation, focusing on detection of a colliding pedestrian among multiple noncolliding pedestrians, suitable for measuring the effects of homonymous hemianopia and assistive devices in clinical trials. PURPOSE In preparation for deploying the test in a multisite clinical trial, we conducted a pilot study to gather preliminary data on blind-side collision detection performance with multiperiscopic peripheral prisms compared with Fresnel peripheral prisms. We tested the hypothesis that detection rates for colliding pedestrians approaching on a 40° bearing angle (close to the highest collision risk when walking) would be higher with 100Δ oblique multiperiscopic (≈42° expansion) than 65Δ oblique Fresnel peripheral prisms (≈32° expansion). METHODS Six participants with homonymous hemianopia completed the test with and without each type of prism glasses, after using them in daily mobility for a minimum of 4 weeks. The test, presented as a video on a large screen, simulated walking through a busy shopping mall. Colliding pedestrians approached from the left or the right on a bearing angle of 20 or 40°. RESULTS Overall, blind-side detection was only 23% without prisms but improved to 73% with prisms. For multiperiscopic prisms, blind-side detection was significantly higher with than without prisms at 40° (88 vs. 0%) and 20° (75 vs. 0%). For Fresnel peripheral prisms, blind-side detection rates were not significantly higher with than without prisms at 40° (38 vs. 0%) but were significantly higher with prisms at 20° (94 vs. 56%). At 40°, detection rates were significantly higher with multiperiscopic than Fresnel prisms (88 vs. 38%). CONCLUSIONS The collision detection test is suitable for evaluating the effects of hemianopia and prism glasses on collision detection, confirming its readiness to serve as the primary outcome measure in the upcoming clinical trial.

Among various specifications of near eye display (NED) devices, a compact formfactor is essential for comfortable user experience but also the hardest one to accomplish due to the slowest progresses. A pinhole/pinlight array based light-field (LF) technique is considered as one of the candidates to achieve that goal without thicker and heavier refractive optics. Despite those promising advantages, however, there are critical issues, such as dark spots and contrast distortion, which degrade the image quality because of the vulnerability of the LF retinal image when the observer's eye pupil size changes. Regardless of previous attempts to overcome those artifacts, it was impossible to resolve both issues due to their trade-off relation. In this paper, in order to resolve them simultaneously, we propose a concept of multiplexed retinal projections to integrate the LF retinal image through rotating transitions of refined and modulated elemental images for robust compensation of eye pupil variance with improved conservation of contrast distribution. Experimental demonstrations and quantitative analysis are also provided to verify the principle.

Binocular double vision in strabismus is marked by diplopia (seeing the same object in two different directions) and visual confusion (seeing two different objects in the same direction). In strabismus with full visual field, the diplopia coexists with visual confusion across most of the binocular field. With visual field loss, or with use of partial prism segments for field expansion, the two phenomena may be separable. This separability is the focus of this review and offers new insights into binocular function. We show that confusion is necessary but is not sufficient for field expansion. Diplopia plays no role in field expansion but is necessary for clinical testing of strabismus, making such testing difficult in field loss conditions with confusion without diplopia. The roles of the three-dimensional structure of the real world and the dynamic of eye movements within that structure are considered as well. Suppression of one eye's partial view under binocular vision that develops in early-onset (childhood) strabismus is assumed to be a sensory adaption to diplopia. This assumption can be tested using the separation of diplopia and confusion.

Avoiding person-to-person collisions is critical for visual field loss patients. Any intervention claiming to improve the safety of such patients should empirically demonstrate its efficacy. To design a VR mobility testing platform presenting multiple pedestrians, a distinction between colliding and non-colliding pedestrians must be clearly defined. We measured nine normally sighted subjects' collision envelopes (CE; an egocentric boundary distinguishing collision and non-collision) and found it changes based on the approaching pedestrian's bearing angle and speed. For person-to-person collision events for the VR mobility testing platform, non-colliding pedestrians should not evade the CE.

When walking, we detect possible collisions with other pedestrians and avoid them, estimating body volumes and safety margins. This safety margin, collision envelope, has been measured in limited conditions: only parallel approaching collisions with lateral offset were tested while subjects watched a walking video. These scenarios limit generalizability for real-world walking where walkers navigate freely with various approaching directions of pedestrians. To evaluate realistic and dynamic collision envelopes in a risk-free environment, we developed a virtual reality walking scenario using the Meta Quest 2 head-mounted display (HMD). While a subject walking with gaze movement in an empty real-world corridor, a corresponding virtual shopping mall with pedestrians approached from 20°, 40°, or 60° bearing angles on a collision course face-to-face or overtaken were shown on HMD. 10 non-colliding pedestrians on various walking paths were also present. Subjects were asked to freely and naturally avoided potential collisions (walking path or speed change). Subjects with homonymous hemianopia (HH; n=6) and subjects with normal vision (NV; n=8) avoided 20 face-to-face and 20 overtaken pedestrians. As a result of the collision avoidance behavior, the trajectories of pedestrians relative to the subjects were changed, and thus the safety margins in various paths were collected. Dynamic collision envelope was calculated as the area kept as the safety margin in more than 50% of trials. HH subjects had larger envelopes (0.95m2, SD=0.60) than NV subjects (0.71m2, SD=0.42; p=0.044) and envelopes were larger when colliders were approaching (1.14m2, SD=0.53) compared to overtaken (0.49, SD=0.19; p<0.001). These results may suggest a more conservative safety margin in HH than NV when avoiding potential collisions. Since the relative walking speeds of the approaching pedestrians were faster than the overtaken pedestrians, estimated time-to-collision may also affect the size and structure of the collision envelope.

Prism field expansion is a common treatment for patients with peripheral field loss, shifting images from the blind field into the seeing field. The shifted image originates from a new viewpoint translated and rotated from the original viewpoint by the prism. To understand such viewpoint changes, we simulated two field expansion methods in virtual reality: 1) angular (i.e., rotational) field expansion and 2) linear field expansion via image crop-and-shift. Changes to object locations, sizes, and optic flow patterns by those methods were demonstrated and analyzed in both static and dynamic conditions, which may affect navigation with such field expansion devices.

Oblique Fresnel peripheral prisms have been used for field expansion in homonymous hemianopia mobility such as walking and driving. However, limited field expansion, low image quality, and small eye scanning range limit their effectiveness. We developed a new oblique multi-periscopic prism using a cascade of rotated half-penta prisms, which provides 42° horizontal field expansion along with 18° vertical shift, high image quality, and wider eye scanning range. Feasibility and performance of a prototype using 3D-printed module are demonstrated by raytracing, photographic depiction, and Goldmann perimetry with patients with homonymous hemianopia.

Recent head-mounted displays and smart glasses use vision multiplexing, an optical approach where two or more views are superimposed on each other. In vision multiplexing, augmented information is presented over an observer’s natural field of view, providing field expansion and critical information during mobility situations like walking and driving. Yet despite its utility, vision multiplexing may produce visual rivalry, a phenomenon where perception alternates between the augmented information and the background scene for seconds at a time. To investigate, we compared the effect of different peripheral vision multiplexing configurations (unilateral opaque, unilateral see-through and bilateral see-through) on the detection of augmented information, incorporating at the same time real-world characteristics (target eccentricity, depth condition, and gaze movement) for a more realistic assessment. Results showed a persistently lower target detection rate in unilateral configurations than the bilateral configuration, suggesting a larger effect of binocular rivalry on target visibility. Nevertheless, this effect does become attenuated when more naturalistic elements are incorporated, and we discuss recommendations for vision multiplexing design and possible avenues for further research.

A single superimposed image containing two image views causes visual confusion for both human vision and computer vision. Human vision needs a "develop-then-rival" process to decompose the superimposed image into two individual images, which effectively suppresses visual confusion. However, separating individual image views from a single superimposed image has been an important but challenging task in computer vision area for a long time. In this paper, we propose a human vision-inspired framework for single superimposed image decomposition. We first propose a network to simulate the development stage, which tries to understand and distinguish the semantic information of the two layers of a single superimposed image. To further simulate the rivalry activation/suppression process in human brains, we carefully design a rivalry stage, which incorporates the original mixed input (superimposed image), the activated visual information (outputs of the development stage) together, and then rivals to get images without ambiguity. Experimental results show that our novel framework effectively separates the superimposed images and significantly improves the performance with better output quality compared with state-of-the-art methods. The proposed method also achieves state-of-the-art results on related applications including single image reflection removal, single image rain removal, single image shadow removal, and illumination correction, etc., which validates the generalization of the framework.

SIGNIFICANCE:Veridical depictions of scene appearance with scotomas allow better understanding of the impact of field loss and may improve the development and implementation of rehabilitation. Explanation and depiction of the invisibility of scotoma may lead to patients' understanding and thus better compliance with related treatments. PURPOSE:Simulations of perception with scotomas guide training, patient education, and rehabilitation research. Most simulations incorrectly depict scotomas as black patches, although the scotomas and the missing contents are usually invisible to patients. We present a novel approach to capture the reported appearance of scenes with scotomas. METHODS:We applied a content-aware image resizing algorithm to carve out the content elided under the scotomas. With video sequences, we show how and why eye movements fail to increase the visibility of the carved scotomas. RESULTS:Numerous effects, reported by patients, emerge naturally from the scotoma carving. Carving-eliminated scotomas over natural images are barely visible, despite causing substantial distortions. Low resolution and contrast sensitivity at farther eccentricities and saccadic blur reduce the visibility of the distortions. In a walking scenario, static objects moving smoothly to the periphery disappear into and then reemerge out of peripheral scotomas, invisibly. CONCLUSIONS:Scotoma carving provides a viable hypothetical simulation of vision with scotomas due to loss of neurons at the retinal ganglion cell level and higher. As a hypothesis, it generates predictions that lend themselves to future clinical testing. The different effects of scotomas due to loss of photoreceptors are left for follow-up work.

Visual confusion occurs when two dissimilar images are superimposed onto the same retinal location. In the context of wearable displays, it can be used to provide multiple sources of information to users on top of the real-world scene. While useful, visual confusion may cause visual rivalry that can suppress one of the sources. If two different images are projected to each eye (i.e., monocular displays), it provokes binocular rivalry wherein visual perception intermittently switches between the two images. When a semi-transparent image is superimposed (i.e., see-through displays), monocular rivalry results, causing perceptual alternations between the foreground and the background images. Here, we investigated how these rivalries influence the visibility of the peripheral target using three configurations of wearable displays (i.e., monocular opaque, monocular see-through, and binocular see-through) with three eye movement conditions (i.e., saccades, smooth pursuit, and central fixation). Using the HTC VIVE Eye Pro headset, subjects viewed a forward vection of a 3D corridor with a horizontally moving vertical grating at 10° above the center fixation. During each trial (~1 min), subjects followed a fixation cross that varied in location to induce eye movements and simultaneously reported whether the peripheral target was visible. Results showed that the binocular display had significantly higher target visibility than both monocular displays, and the monocular see-through display had the lowest target visibility. Target visibility was also higher when eye movements were executed, suggesting that the effects of rivalry are attenuated by eye movements and binocular see-through displays.