Purchase this article with an account.
Lei Liu, Ellen Bowman; Comparability of Visual Performance of Individuals with Low Vision in Real and Virtual Street Intersections. Invest. Ophthalmol. Vis. Sci. 2013;54(15):4067. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Low Vision Orientation & Mobility (O&M) evaluation and training are typically performed in streets and street intersections. While effective, the efficiency of the practice can be greatly improved by employing computer-generated virtual street scenarios. To evaluate the feasibility of virtual O&M training, we compared visual performance of low vision patients in real and virtual street intersections.
A semi-CAVE virtual reality simulator was built. It consisted of a game computer, 3 digital projectors, three 2.4x1.9 m screens (>180o deg horizontal) and surrounding sound. Four virtual street intersections were built based on 4 real intersections in an urban environment. Google Street, Geographic Information System and on-site photos/videos were used to match the physical layout and surrounding, traffic controls, and vehicle and pedestrian traffic of the real and virtual intersections. Multiple crossing scenarios were set at each intersection. Ten legally blind subjects with adequate prior O&M training performed 10 visual tasks in both real and virtual intersections. These tasks included identifying traffic elements, determining intersection layout and choosing the safest time to cross. The performance was rated pass or fail by two certified O&M specialists. Agreement between performance in corresponding real and virtual intersections was compared.
Data from corresponding real and virtual intersections was organized into a 2x2 contingency table. The overall, positive and negative agreements were 83.3%, 89.4% and 68.6%, respectively, indicating that if a subject could perform a task in a real intersection, it was highly likely that she could also perform the same task in a virtual version of the intersection, and vice versa. The kappa coefficient was 0.59, bordering moderate and substantial agreement. A McNemar analysis of paired binary data showed an exact significance of 0.824, indicating no difference in performance between real and virtual intersections.
Persons with low vision exhibited similar abilities while performing visual tasks in real and virtual street intersections. The agreement suggested that computer-generated traffic scenes provided sufficient visual and auditory information for low vision patients to perform O&M tasks. Virtual reality has the potential to become a useful supplement to tradition low vision rehabilitation training.
This PDF is available to Subscribers Only