March 2012
Volume 53, Issue 14
ARVO Annual Meeting Abstract  |   March 2012
A Continuous Classification System for Glaucomatous Visual Field
Author Affiliations & Notes
  • Xian Zhang
    Psychology and Radiology, Ophthalmology,
    Columbia University, New York, New York
  • Ali S. Raza
    Psychology Department,
    Columbia University, New York, New York
  • Gustavo D. Moraes
    New York Eye & Ear Infirmary, New York, New York
  • Jeffrey M. Liebmann
    Ophthalmology, NYU School of Medicine, New York, New York
  • Robert Ritch
    Psychology and Radiology, Ophthalmology,
    New York Eye & Ear Infirmary, New York, New York
  • Donald C. Hood
    Psychology and Ophthalmology,
    Columbia University, New York, New York
  • Footnotes
    Commercial Relationships  Xian Zhang, None; Ali S. Raza, None; Gustavo D. Moraes, None; Jeffrey M. Liebmann, None; Robert Ritch, None; Donald C. Hood, None
  • Footnotes
    Support  NIH Grant EY-02115
Investigative Ophthalmology & Visual Science March 2012, Vol.53, 195. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Xian Zhang, Ali S. Raza, Gustavo D. Moraes, Jeffrey M. Liebmann, Robert Ritch, Donald C. Hood; A Continuous Classification System for Glaucomatous Visual Field. Invest. Ophthalmol. Vis. Sci. 2012;53(14):195.

      Download citation file:

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

  • Supplements

To develop a continuous classification system of glaucomatous defects that represents both the pattern of the visual field (VF) defects, as well as the prevalence of these patterns, principal component analysis (PCA) was performed on 45,643 VFs.


The 24-2 VFs (Carl Zeiss Meditech) of 45,643 individual eyes (glaucoma patients and suspects) collected over 10 years at a single glaucoma clinic were used. To be included, the VF had to be reliable (indices <30%). The pattern deviation (PD) values were analyzed separately for the upper and lower hemifields. PD values greater than 0 dB were set to 0. The data from left eyes were flipped to correspond to right eyes. PD dB values were transformed into linear units, but transformed back after the PCA analysis. Using PCA, PD values were projected into a 2-dimensional VF space, where the two axes were the first 2 PCs and each point is associated with a particular pattern of PD values and the number of glaucoma patients/suspects sharing that pattern.


A 2-D representation of the results for the lower field is shown in the fig., where each axis is a PC. The PD probability values for each pattern are coded from white (p>0.05) to light gray (p≤0.05) to black (p<0.005). The dash contours are iso-lines for the prevalence from highly likely (inner-dash rings) to less likely (outer-dash rings). The labels show locations of typical patterns such as arcuate or ceco-central patterns seen in current discrete classifications schemes [1].


The proposed continuous classification scheme provides information about the relationship among different patterns of glaucomatous defects, as well as about the prevalence of these patterns. It also has the advantage that it can be compared to a similar continuous scheme for structural damage seen with frequency domain optical coherence tomography. 1. Keltner Johnson, Cello, et al, 2003  

Keywords: visual fields • perimetry • optic nerve 

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.