Abstract
Purpose :
To test the hypothesis that the agreement between structural (OCT) and functional (visual field) damage is improved by topographically comparing deviation/probability maps,1 an objective, automated, topographical method was compared to more traditional measures of agreement.
Methods :
Wide-field, swept-source OCT scans (Topcon, Inc), which included the disc and macular regions, and 24-2 and 10-2 visual fields (VF) were obtained from 45 healthy control eyes/individuals with normal fundus exams and IOP <22 mmHg, and 53 eyes/patients with 24-2 MD better than -6dB. The referring glaucoma specialist diagnosed these 53 eyes as “definite glaucoma”, based upon all information for that eye. All individuals had a spherical refractive error between ±6D. In addition to standard metrics of the VF (MD, PSD, GHT) and OCT (global, quadrant and clock hour RNFL thickness), the topographical (T) agreement between the 24-2 and 10-2 VF total deviation probability maps and OCT deviation/probability maps was assessed. To obtain an objective, automated assessment of the latter, a custom program was developed in R. The following criteria defined agreement: 2 abnormal points on the combined 24-2 and 10-2 probability map at ≤5% that fell upon a region ≤10% on the RGC+IPL or RNFL probability maps.
Results :
For the T-method, only 7 (13.2%) of the 53 eyes with early glaucoma failed to show agreement (“misses”), while 2 (4.4%) of the 45 healthy eyes showed agreement. Both of the latter were easily identified as mistakes, as the abnormal region of the VF did not replicate. Without the 10-2, the misses for the T-method increased to 18 (34.0%) eyes. The best VF standard metrics (i.e., 24-2 GHT or PSD, or 10-2 PSD) combined with best OCT (quadrants) had 16 (30.2%) misses, while omitting the 10-2 VF resulted in 20 (37.7%) misses.
Conclusions :
First, agreement between structural and functional damage can be improved by obtaining 10-2, as well as 24-2, VFs and comparing abnormal locations on the VFs to abnormal RGC andRNFL OCT regions. In this way, the addition of a 10-2 test adds significant, clinically-useful information about early glaucoma. Second, an objective, automated approach provides a novel way of providing structure-function corroboration of damage and serves as the underpinning for future artificial intelligence applications. 1. Hood DC, De Moraes CG. (2018) J. of Glaucoma. 27:657.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.