Visual Attention Measures Predict Pedestrian Detection in Central Field Loss: A Pilot Study

Concetta F Alberti,Todd Horowitz,P Matthew Bronstad,Alex R Bowers

doi:10.1371/journal.pone.0089381

Abstract

PurposeThe ability of visually impaired people to deploy attention effectively to maximize use of their residual vision in dynamic situations is fundamental to safe mobility. We conducted a pilot study to evaluate whether tests of dynamic attention (multiple object tracking; MOT) and static attention (Useful Field of View; UFOV) were predictive of the ability of people with central field loss (CFL) to detect pedestrian hazards in simulated driving.Methods11 people with bilateral CFL (visual acuity 20/30-20/200) and 11 age-similar normally-sighted drivers participated. Dynamic and static attention were evaluated with brief, computer-based MOT and UFOV tasks, respectively. Dependent variables were the log speed threshold for 60% correct identification of targets (MOT) and the increase in the presentation duration for 75% correct identification of a central target when a concurrent peripheral task was added (UFOV divided and selective attention subtests). Participants drove in a simulator and pressed the horn whenever they detected pedestrians that walked or ran toward the road. The dependent variable was the proportion of timely reactions (could have stopped in time to avoid a collision).ResultsUFOV and MOT performance of CFL participants was poorer than that of controls, and the proportion of timely reactions was also lower (worse) (84% and 97%, respectively; p = 0.001). For CFL participants, higher proportions of timely reactions correlated significantly with higher (better) MOT speed thresholds (r = 0.73, p = 0.01), with better performance on the UFOV divided and selective attention subtests (r = −0.66 and −0.62, respectively, p<0.04), with better contrast sensitivity scores (r = 0.54, p = 0.08) and smaller scotomas (r = −0.60, p = 0.05).ConclusionsOur results suggest that brief laboratory-based tests of visual attention may provide useful measures of functional visual ability of individuals with CFL relevant to more complex mobility tasks.

Highlights

Central field loss (CFL) is the presence of scotomas within the central visual field including the fovea
In order to compensate effectively for the CFL, the preferred retinal locus (PRL) has to be used consistently for fixation, saccades need to be directed to the PRL rather than the fovea, and scanning of a scene with the PRL has to be accomplished in such a way as to minimize occlusion of objects by the scotoma
Results of a prior study in the dynamic environment of a driving simulator suggest that drivers with CFL might not be able to fully compensate for their scotomata, as responses to pedestrian hazards that appeared in scotoma areas were much slower than responses to hazards in non-scotoma areas [7]

Summary

Introduction

Central field loss (CFL) is the presence of scotomas (blind areas) within the central visual field including the fovea. Despite advances in treatments [1], the majority of individuals with macular degeneration have irreversible vision loss that causes difficulties in a range of activities including reading and mobility (walking and driving) [2,3]. People with CFL almost always use a preferred retinal locus (PRL), an extra-foveal location near the scotoma, to fixate targets that would normally be foveally fixated [4,5]. They have a blind area in central vision, and impaired visual acuity and contrast sensitivity [6]. Results of a prior study in the dynamic environment of a driving simulator suggest that drivers with CFL might not be able to fully compensate for their scotomata, as responses to pedestrian hazards that appeared in scotoma areas were much slower than responses to hazards in non-scotoma areas [7]

Methods

Results

Conclusion