Related Experiment Videos
Evaluation of a Novel Visual Field Analyzer Application for Automated Classification of Glaucoma Severity
Renato A S Germano1, Carlos Gustavo de Moraes, Remo Susanna
1*Department of Ophthalmology, University of Sao Paulo School of Medicine, Sao Paulo, SP ‡Department of Computer Science, Federal University of Sergipe (UFS), Sao Cristovao, Brazil †Department of Ophthalmology, Columbia University Medical Center, New York, NY.
Purpose:
To describe the Glaucoma Staging Application (GSA), a new application for automated classification of visual field (VF) severity that can be used in any Humphrey perimeters. We also tested and validated its performance in staging glaucomatous VFs defects and assessed its reproducibility.
Materials And Methods:
The GSA is based upon the modified University of São Paulo Glaucoma Visual Field Staging System, and consists of 3 sections: severity based on the visual field index; defect location based on eccentricity; and extent of topographic damage by hemifields. We analyzed 612 VFs of 32 patients. VFs were analyzed by a glaucoma expert and then compared with the output of the GSA. To evaluate the repeatability, the same search criteria were repeated 1 week later using the same GSA. The classification given by the software was considered correct if it completely agreed in all variables (severity, location, and extent of topographic damage) with the classification performed manually by the glaucoma expert.
Results:
One hundred seventy-eight eyes were classified as early VF damage, 148 moderate, and 156 severe. One hundred thirty-one VFs were classified as normal. When repeated, 1 week later, the GSA showed excellent repeatability. The automated classification agreed with the expert's grading in 100% of cases.
Conclusions:
The GSA allows an automated, fast, reproducible, and objective classification system for multiple 24-2 VFs of Humphrey perimetry. This study is a proof-of-concept that could translate into a useful tool to analyze and stage VFs more objectively and thus help detecting progression of different types of VFs defects.