Abstract
Purpose :
To introduce an integrated multi-parameter representation of changes in visual field data time series obtained with threshold Amsler grid tests performed at various grid contrast levels.
Methods :
The Ceeable Visual Field Analyzer (CVFA; Adams et al., SPIE 9836, 2016; Fink & Sadun, JBO 2004), a threshold Amsler grid test, records visual field changes either as missing areas (i.e., scotomas) or distorted areas (i.e., metamorphopsia) on a user-defined number of Amsler grids of varying contrast levels. Frequent retesting of a subject or subjects establishes time series of visual field data that capture the temporal development of visual fields due to disease or treatment. CVFA extracts objectively the following parameters from the raw visual field exam data (You & Fink, ARVO 2010): P1: relative number of Amsler grid test locations affected; P2: relative hill-of-vision lost; P3: slope histogram of contrast sensitivity loss over visual field degrees; P4: lost area grade, i.e., ratio of scotoma area at highest vs. lowest tested contrast level; P5: visual field loss progression, i.e., respective % of visual field area not seen as a color-coded (according to Amsler grid contrast levels tested) horizontal bar of a length proportional to that % (Fink et al., ARVO 2017).
Results :
A CVFA exam of an AMD-related central visual field defect prior to treatment (Fig. 1) and a follow-up exam 6 months after treatment (Fig. 2) were simulated. The importance of looking at the above parameters collectively to express comprehensively the visual field changes is evident: the larger relative scotoma turns into a smaller absolute scotoma (3D scotoma plots in Figs. 1, 2), i.e. in green: P1, P2 down; P3: shallow to steep slopes; P4 up; P5: shorter, equal-sized bars.
Conclusions :
The devised parameters P1-P5 only collectively express/describe, both numerically and graphically, visual field changes over time due to disease or treatment in a comprehensive manner, whereas each parameter by itself can only attest to a part of the temporal visual field development at the expense of the global picture. These parameters can be readily and objectively extracted from the raw data of each visual field examination, and are amenable to machine/deep learning and data mining methods.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.