Measuring Pedestrian Collision Detection With Peripheral Field Loss and the Impact of Peripheral Prisms

Abstract

Purpose: Peripheral field loss (PFL) due to retinitis pigmentosa, choroideremia, or glaucoma often results in a highly constricted residual central field, which makes it difficult for patients to avoid collision with approaching pedestrians. We developed a virtual environment to evaluate the ability of patients to detect pedestrians and judge potential collisions. We validated the system with both PFL patients and normally sighted subjects with simulated PFL. We also tested whether properly placed high-power prisms may improve pedestrian detection.

Methods: A virtual park-like open space was rendered using a driving simulator (configured for walking speeds), and pedestrians in testing scenarios appeared within and outside the residual central field. Nine normally sighted subjects and eight PFL patients performed the pedestrian detection and collision judgment tasks. The performance of the subjects with simulated PFL was further evaluated with field of view expanding prisms.

Results: The virtual system for testing pedestrian detection and collision judgment was validated. The performance of PFL patients and normally sighted subjects with simulated PFL were similar. The prisms for simulated PFL improved detection rates, reduced detection response times, and supported reasonable collision judgments in the prism-expanded field; detections and collision judgments in the residual central field were not influenced negatively by the prisms.

Conclusions: The scenarios in a virtual environment are suitable for evaluating PFL and the impact of field of view expanding devices.

Translational relevance: This study validated an objective means to evaluate field expansion devices in reproducible near-real-life settings.

Keywords: peripheral field loss; prism; retinitis pigmentosa; tunnel vision; vision rehabilitation; visual field expansion; walking simulator.

Figure 1

Near-collision (a–c) and center-to-center collision (d–f) that can be encountered by patients with PFL (illustrated as a gray diamond). (a) Top view of a near-collision when the two pedestrians are on sidewalks or in corridors, where their trajectories are parallel. The solid gray diamond and blue circle show the initial positions of the patient and another pedestrian, respectively. The angle formed by the gray and black solid lines, , is the pedestrian's bearing angle relative to the patient's heading direction. The patient and pedestrian are presumed to reach where the arrows point after 6 seconds, where the distance between their central points becomes the closest (0.6 m). The shaded gray triangle illustrates the residual central field visible to a patient with a 20° diameter residual central field. (b) Top view of the patient and pedestrian pair 5 seconds after (a). The bearing angle changes to , which gradually moves out of the patient's residual central field (the shaded gray triangle). (c) The changes of the pedestrian's bearing span relative to the patient's heading direction as a function of time ((a) to (b) and to the 6th second). The gray area indicates the patient's visible field on the right side over time. (d) A center-to-center collision case, which is likely in an open space where pedestrians' paths are less regulated. The pedestrian shown as an orange circle is outside the patient's residual central field. The black dot indicates the collision point where the patient (gray diamond) and pedestrian (orange circle) would arrive simultaneously. (e) Top view of the same patient and pedestrian pair 5 seconds after (d). (f) The bearing span as a function of time relative to the field of the patient shown in gray. The pedestrian is only visible to the patient's residual central field after 5.8 seconds.

Figure 2

Simulated walking in an open space environment. (a) An open park scene rendered on the driving simulator (only three out of five screens are shown). (b) A front screen image showing a pedestrian approaching from the left at 30° bearing. The string marking the center of the screen for aligning the heading direction is illustrated using a black vertical line. The participant maintains the planned walking path by steering to align the intended path traced by the basketball (small orange circle) with the string.

Figure 3

Diagrams of the various trials with pedestrians appearing initially at various bearings () and crossing the participant's path at various distances from the participant (). The participant is shown as a gray diamond and the pedestrian is a blue circle if its bearing is within the simulated residual central field or orange circle if it is outside. The shaded triangle in light gray indicates the simulated residual central field with 20° diameter. (a) A center-to-center collision with . The participant and the pedestrian arrive at the collision point simultaneously (). The pedestrian is invisible to the participant's residual central field most of the time. (b) A center-to-center collision with . (c) The participant and the pedestrian walk on parallel paths but toward each other with and the closest distance 1.69 m (noncollision). In (b) and (c), the pedestrian is initially visible within the participant's residual central field. (d) A near- or noncollision trial with , where the pedestrian passes in front of the participant. The corresponding condition with the center-to-center collision is shown with the dashed orange line as a comparison. Given the same initial bearing (), the same pedestrian heading () is used for various path crossing distances. (e) , where the pedestrian passes behind the participant.

Figure 4

Goggles for simulating PFL. (a) Front diagram of the goggles with the mask to simulate the 20° diameter residual central field and the oblique Fresnel prisms. The arrows indicate the direction from the prism apex to base, thus the prisms are placed base-out and base-up in front of the right eye, and base-out and base-down in front of the left eye. (b) Picture of the goggles with the prisms attached.

Figure 5

Illustration of the simulated PFL and prism effects. (a) Binocular perimetry through the simulated PFL using goggles with the prisms. A scene from the virtual scenario from the simulator front screen is overlaid by the perimetry result, which was measured with the prisms in place using a 1.5° black square target (on white screen) from 735 mm viewing distance. Note that without prisms, the expanded areas to the left and right would not be present. The measured expansion is close to the calculated theoretical values (between 11° to 39°). The black rectangular box presents the simulator rear-view mirror images, which was outside the field of simulated PFL. (b) Illustration of percepts for left and right eyes (field of view as two apertures). The vertical middle portion in both apertures is seen by both eyes and is fused, but the left and right expanded fields, as shown in (a), were each seen only through one eye. The left expanded field appears in the upper portion of the left eye view, and the right expanded field appears in the lower portion of right eye view. The far ends of expanded fields are highly compressed (note the laterally compressed head and torso of the pedestrian in the left eye's view). Areas of TIR (Supplementary Appendix “Details of prism positioning”) shown in black can be seen in the left and right ends of the left and right eyes' views, respectively. Under binocular viewing the upper and lower un-shifted fields of view will be seen by the other eye.

Figure 6

The grouped detection, decision, and response time (RT) results for the residual central field (hatched blue bars) and the prism-expanded field or the unseen field for patients (solid orange), and for the three viewing conditions or patients: NV, normal vision; PFL, simulated peripheral field loss; PFL+PR, simulated PFL with prisms; PAT, patients with PFL. (a) The detection rate. (b) The collision decision accuracy. (c) The RT for detection (bottom and darker) and decision (stacked on top and lighter color). Error bars represent standard error of the mean.

Figure 7

The average detection rate, detection RT, decision of collision, and decision RT results from the normally sighted subjects (a–d) and the PFL patients (e–h). The three panels in (a–d) are for the three viewing conditions as marked. The blue symbols indicate the conditions when pedestrians initially appear with (triangles, ; diamond, ; star, ), the orange circles are for the conditions with . The x-axis shows the path crossing distance, = −2 m, 0, +2 m, +12 m, and symbol ∞ for indicates the sidewalk-like encounters. (a) The pedestrian detection rate of the normally sighted subjects for the individual conditions. (b) The RT to detect the pedestrians. (c) The percentage of trials perceived as collisions. (d) The RT to make the collision judgment (after the pedestrian detection). Error bars represent standard error of the mean. (e–h) The average detection rate, detection RT, decision of collision, and decision RT of the eight PFL patients. Their behavioral patterns are similar to those acquired in simulated PFL with the normally sighted subjects (middle column in a–d).

Figure 8

Visual cues for possible collision with an approaching pedestrian and their relationships with the collision judgment by the subjects. (a) Accumulated bearing deviations of pedestrian's central bearing from a constant bearing. The straight dashed gray line shows the constant central bearing representing the bearing of a center-to-center colliding pedestrian. The gray area indicates the bearing span of that colliding pedestrian increasing as a function of time. The dark blue curve shows the central bearing of a pedestrian passing-in-front, which starts at the same initial bearing. The black vertical arrows are samples of the accumulated bearing deviations. (b) The accumulated bearing deviations show a negative correlation with the subjects' perceived collision. (c) The bearing span overlap of the dark blue pedestrian (from (a)) with the center-to-center colliding pedestrian. The percent of overlap was calculated as the blue-gray overlapping area divided by the gray area. (d) The subjects' collision judgment shows a positive correlation with the bearing span overlap. (e) The collision judgment is negatively correlated with the closest distance between the subject and pedestrian. Data in the scatterplots are from the NV condition (same colors and icons as in Fig. 7). Additional data (black asterisks) are from (Qiu C, et al. IOVS. 2017;58:ARVO E-Abstract 3287).