Clinical Characteristics of Glaucoma Eyes for AI Diagnostic Validation

370 eyes from 193 patients were used to evaluate two AI systems for glaucoma detection from fundus photographs. The dataset, created by Gilvan Vilarinho da Silva Filho, includes structural and visual field test data for 248 glaucomatous and 122 healthy eyes. Diagnostic accuracy metrics, including AUCs up to 0.916, are provided for the Laguna ONhE and VUNO Med-Fundus AI systems.

Use Cases

• Validate AI diagnostic models for glaucoma based on color fundus photography (CFP) data.
• Compare the performance of combined AI systems (GDF + VUNO) against individual tools.
• Benchmark AI diagnostic accuracy against structural (OCT) and functional (visual field) clinical tests.

Strengths

• Includes data from 370 eyes, with 248 diagnosed with primary open-angle glaucoma and 122 healthy controls.
• Provides concrete diagnostic accuracy metrics, such as an AUC of 0.903 for the combined AI approach.
• Links AI predictions to ground truth from swept-source optical coherence tomography and visual field testing.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect temporal or institutional bias inherent to the single retrospective study design.

Provenance

Source

figshare, authored by Gilvan Vilarinho da Silva Filho.

Collection Method

Retrospective cross-sectional study; data gathered from clinical evaluations.

Freshness

Last updated 2026-05-28 17:34:34; freshness should be verified.