Clinical validation of RIA-G, an automated optic nerve head analysis software

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence in Healthcare

Background:

• Glaucoma diagnosis relies on expert interpretation of optic nerve head parameters.
• Automated diagnostic tools can improve efficiency and accessibility in glaucoma screening.

Purpose of the Study:

• To clinically validate the accuracy and reliability of the RIA-G automated software for glaucoma diagnosis.

Main Methods:

• A double-blinded study compared RIA-G's analysis of 229 fundus images against three expert ophthalmologists.
• Key parameters quantified included cup-disc ratio (CDR) and neuroretinal rim (NRR) changes.
• Software performance was assessed for Disc Damage Likelihood Scale (DDLS) staging and glaucoma presence.

Main Results:

• RIA-G demonstrated strong agreement with experts for vertical CDR (ICC 0.79) and moderate agreement for horizontal CDR (ICC 0.48).
• High accuracy was observed in detecting glaucomatous neuropathy (90.3%) with sensitivity (82.3%) and specificity (91.8%).
• Overall agreement for glaucoma diagnosis was good (Cohen's kappa = 0.62).

Conclusions:

• The RIA-G software is a reliable tool for detecting glaucomatous optic neuropathy, showing good concordance with expert evaluations.
• Its ability to quantify optic nerve head parameters from fundus photographs is valuable for large-scale glaucoma screening programs.