March 2012
Volume 53, Issue 14
Free
ARVO Annual Meeting Abstract  |   March 2012
Towards an Improved Estimation of Pattern-Deviation Maps Based on Total-Deviation Rank Curves
Author Affiliations & Notes
  • Ivan Marin-Franch
    School of Optometry, Indiana University, Bloomington, Indiana
    Optometry and Visual Science, City University London, London, United Kingdom
  • William H. Swanson
    School of Optometry, Indiana University, Bloomington, Indiana
  • Lyne Racette
    Eugene and Marilyn Glick Eye Institute, Indiana University, Indianapolis, Indiana
  • Footnotes
    Commercial Relationships  Ivan Marin-Franch, None; William H. Swanson, None; Lyne Racette, Haag-Streit (R)
  • Footnotes
    Support  NIH EY007716
Investigative Ophthalmology & Visual Science March 2012, Vol.53, 208. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Ivan Marin-Franch, William H. Swanson, Lyne Racette; Towards an Improved Estimation of Pattern-Deviation Maps Based on Total-Deviation Rank Curves. Invest. Ophthalmol. Vis. Sci. 2012;53(14):208.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose: : Pattern-deviation (PD) analysis relies on the assumption that eyes from healthy subjects of the same age have similar shapes in their hill of vision. PD maps are derived from a specific estimate of the general height (GH), the 85th percentile of the total-deviation (TD) values, and are known to fail under certain conditions. The long-term goal of the present study is to develop improved PD plots using Bebie TD rank curves. A step towards this goal was to determine how inter-subject differences in shape of the TD rank curves, which can reflect differences in the shape of the hill of vision, affect estimates of GH.

Results: : Both models yielded good fits with average root-mean-square-errors of 0.21 dB for Model 1 and 0.28 dB for Model 2. Taking the GH estimates by Model 1 as reference, GH estimates by Model 2 had a mean bias of -0.02 dB. The standard deviation (SD) of bias was 0.1 dB. By comparison, GH estimates as the 85th TD percentile had a mean bias of -0.1 dB, five times larger than that for Model 2, with a SD of 0.3 dB, three times larger. Estimate bias, however, depended on the shape of the TD rank curve. The slope of a linear regression of GH estimate by Model 1 on the slope of the GLM linear predictor was 0.1 (p = 0.02). Dependence was much stronger for GH estimates as the 85th TD percentile; with a negative slope of -1.0 (p < 0.0001) .

Conclusions: : The shape of the TD rank curves affects estimates of GH based on the 85th TD percentile. Estimates that are more robust to differences in shape can be obtained if the whole TD rank curve is considered and may lead to improved PD maps.

Keywords: visual fields • perimetry 
×
×

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.

×