June 2022
Volume 63, Issue 7
Open Access
ARVO Annual Meeting Abstract  |   June 2022
Estimating Pedestrian Detection with peripheral prisms in Homonymous Hemianopia
Author Affiliations & Notes
  • Sailaja Manda
    Opthalmology, Massachusetts General Hospital, Boston, Massachusetts, United States
  • Nish Mohith Kurukuti
    Biomedical Engineering, Northwestern University, Evanston, Illinois, United States
  • Alex D Hwang
    Opthalmology, Massachusetts General Hospital, Boston, Massachusetts, United States
  • Jae-Hyun Jung
    Opthalmology, Massachusetts General Hospital, Boston, Massachusetts, United States
  • Eli Peli
    Opthalmology, Massachusetts General Hospital, Boston, Massachusetts, United States
  • Footnotes
    Commercial Relationships   Sailaja Manda None; Nish Kurukuti None; Alex Hwang None; Jae-Hyun Jung None; Eli Peli Chadwick Optical, Code P (Patent)
  • Footnotes
    Support  : NIH R01 EY023385
Investigative Ophthalmology & Visual Science June 2022, Vol.63, 2453 – F0030. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Sailaja Manda, Nish Mohith Kurukuti, Alex D Hwang, Jae-Hyun Jung, Eli Peli; Estimating Pedestrian Detection with peripheral prisms in Homonymous Hemianopia. Invest. Ophthalmol. Vis. Sci. 2022;63(7):2453 – F0030.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose : Fresnel(F, 57PD) and Multiperiscopic(M, 100PD) prisms are peripheral prisms designed to expand the visual field for persons with Homonymous Hemianopia (HH). Prism segments are placed above(U) and below(L) the pupil in a horizontal(H) or oblique(O) configuration. We constructed a mathematic model to simulate pedestrians walking toward a patient wearing any combination(U/L/Both(U&L)) of (F/M) prisms to compare detection rates and estimate if wearing single prism(U/L) is as good as both(U&L).

Methods : The dataset simulated 6-second events of a patient wearing F/M looking straight ahead without eye movement and walking toward a collision point. A pedestrian approached at 0°-90° to this point (at 1° intervals). We modelled 6 pedestrian forms based on physical anthropometry measurements. Each form used 80 data-points to model patient and pedestrian attributes such as walking speed, height, torso, arm/leg width and span. A pedestrian was categorized detected if the height/width of the pedestrian’s image within the prism was >3°(based on experiment data collected from 4 HH patients). Detection rates were computed for 540 events per HH model wearing a U/L/U&L prism combination of F/M prisms. Differences in detection rates were analyzed using a 2(F/M) x 2(H/O) x 3 (U/L/U&L) ANOVA.

Results : The detection rate with any prism(F:0.40±0.38, M:0.62±0.32) was better than without prisms(0.05±0.2). There were significant main effects(all ps<0.001) of the type of prism design(F vs M), configuration(H: 0.62±0.4, O: 0.37±0.27) and number of segments(U:0.15±0.03, L: 0.67±0.33, U&L:0.71±0.33). All interactions between these variables were significant(all ps<0.001). The LH configuration(LHF:0.82±0.01, LHM:1) had the best detection rates and these were not significantly different from three of the U&L prisms performance(U&L HF:0.824±0.01, U&L HM:1±0, U&L OM: 0.782±0.02).

Conclusions : A mathematic model was used to simulate if a single prism supported detection as well as dual peripheral prism configurations in the absence of observer head tilt or eye movement. Use of a single prism configuration would provide reduced cost and better cosmesis for patients. Mathematic modeling to estimate detection can combine real world physical measurements with theoretical prism expansion calculation. Further analyses with head and gaze shift modelling could be a useful tool to support design of peripheral prisms and aid clinical decision making.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×