# Early Diagnosis of Glaucoma By Optical Coherence Tomography: A Systematic Review and Network Meta-analysis

## TL;DR

This study compares OCT parameters for early glaucoma detection and finds combining them with AI could improve accuracy.

## Contribution

A network meta-analysis comparing OCT-derived parameters for early glaucoma diagnosis, suggesting combined use with AI for better accuracy.

## Key findings

- Average and inferior RNFL parameters achieved the highest diagnostic accuracy (0.77) for glaucoma detection.
- Combining OCT parameters with artificial intelligence could significantly enhance early glaucoma detection accuracy.
- Diagnostic accuracy was comparable across peripapillary RNFL, macular metrics, and optic nerve head parameters.

## Abstract

Optical coherence tomography (OCT) diagnostic technology is increasingly being integrated into clinical practice and is evolving rapidly. Performing a meta-analysis of these advancements can provide valuable insights to clinicians and healthcare decision-makers and help identify gaps that need further research and development. We conducted a systematic review and network meta-analysis (NMA) comparing the diagnostic accuracy of three key OCT-derived parameters in identifying early glaucoma: peripapillary retinal nerve fiber layer (RNFL) thickness, macular metrics, and optic nerve head (ONH) characteristics. We systematically searched PubMed, Embase, Web of Science, and Scopus for studies published from January 2004 through March 2024, and identified 47 eligible studies. These studies comprised 12,723 eyes from 8177 participants, including patients with mild primary open-angle glaucoma, pre-perimetric glaucoma (PPG), and ocular hypertension, as well as healthy controls. Our NMA summarized the existing evidence on the diagnostic performance of these parameters. The highest diagnostic accuracies for glaucoma detection were observed for average and inferior RNFL parameters, both achieving an accuracy of 0.77 and macular GLV parameter with an accuracy of 0.76. These were followed by inferotemporal GCIPL and two ONH parameters—rim area and cup-to-disc ratio—each with an accuracy of 0.75. In conclusion, the overall diagnostic accuracy was comparable across all three categories of OCT parameters. The results also suggest the potential benefits of combining two or three of these parameters in a single report, along with the application of artificial intelligence. This approach could significantly enhance the diagnostic accuracy of OCT in detecting glaucoma at its earliest stages, ultimately improving patient outcomes through earlier intervention.