Performance of OCT-Based Artificial Intelligence Models for Detecting and Predicting Glaucoma Progression: A...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Glaucoma is a leading cause of irreversible blindness worldwide.
• Early detection is crucial for vision preservation.
• Current standard methods like automated perimetry and OCT have limitations in detecting early glaucomatous changes.

Purpose of the Study:

• To systematically review the diagnostic and predictive performance of AI-driven OCT-based models for glaucoma.
• To compare AI model performance against standard automated perimetry and conventional OCT trend analysis.
• To evaluate AI models using deep learning, machine learning, and transformer architectures.

Main Methods:

• Comprehensive literature search of PubMed, Ovid, and Google Scholar.
• Screening of 13 relevant studies including observational, cohort, diagnostic accuracy, and clinical trials.
• Extraction and synthesis of key outcomes like sensitivity, specificity, and AUC.

Main Results:

• AI-driven OCT models demonstrated high diagnostic performance (sensitivity/specificity >80%, AUC >0.85).
• These AI models detected glaucoma progression up to 9 months earlier than standard methods.
• AI models identified subtle structural and functional changes indicative of early glaucoma.

Conclusions:

• AI-driven OCT models show potential to enhance early glaucoma detection and inform timely interventions.
• The majority of studies utilized internal validation, necessitating prospective multicenter studies with external validation.
• Further research is required before widespread clinical implementation in routine glaucoma care.