A comparison of algorithms for calculating glaucoma change probability confidence intervals

Area of Science:

• Ophthalmology
• Medical Technology
• Data Analysis

Background:

• Automated perimetry is crucial for diagnosing and monitoring glaucoma.
• Accurate detection of visual field change is essential for timely intervention.
• Glaucoma Change Probability (GCP) is a metric used to assess visual field progression.

Purpose of the Study:

• To evaluate the efficacy of four distinct methods for calculating Glaucoma Change Probability (GCP).
• To compare the ability of these methods to detect changes in standard automated perimetry data.
• To assess the sensitivity and specificity of each GCP calculation method.

Main Methods:

• Utilized a database of visual fields from 35 glaucoma patients and 15 normal individuals.
• Developed confidence intervals for GCP using four methods: defect-based vs. mean threshold-based, and test-retest vs. baseline-less-follow-up data.
• Measured specificity using simulated stable visual field data and sensitivity using simulated progressive visual field data.

Main Results:

• Defect-based methods with test-retest differences showed higher specificity compared to mean threshold-based methods.
• Mean threshold-based methods demonstrated greater sensitivity in detecting localized visual field decreases.
• Employing baseline-less-follow-up data for confidence intervals enhanced the detection of visual field progression compared to test-retest differences.

Conclusions:

• Stratifying baseline visual field measurements by defect and eccentricity yields high specificity (98%).
• Stratifying by mean threshold and using baseline-less-follow-up increases sensitivity by 14.1% at a minor specificity cost (2.2%).
• Optimized GCP calculation methods can improve the detection of glaucoma progression in automated perimetry.