Enhancement of Optical Coherence Tomography Images Using Adversarial Neural Networks: Impacts on Ophthalmic Practice

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Artificial intelligence (AI) integration in ophthalmology is rapidly advancing.
• Optical coherence tomography (OCT) is crucial for visualizing retinal layers.
• AI shows potential to improve OCT image resolution and diagnostic interpretation.

Purpose of the Study:

• To evaluate an AI-driven filter for enhancing OCT images.
• To assess the impact on image clarity and diagnostic accuracy.
• To investigate AI's role in ophthalmological diagnostics.

Main Methods:

• Used generative adversarial networks (GANs), specifically Real-ESRGAN, for OCT image enhancement.
• 147 ophthalmologists assessed anonymized OCT images before and after AI enhancement.
• Statistical analysis included descriptive statistics and chi-squared tests.

Main Results:

• 69.4% of ophthalmologists reported improved clarity of retinal layers with AI-enhanced images.
• 78.9% acknowledged AI's benefit in highlighting biomarkers.
• Opinions varied on AI's utility for diagnosis and case management.

Conclusions:

• AI-processed OCT images (using GANs) improve clarity and detail, potentially aiding diagnostics.
• Variability in perceived clinical utility necessitates cautious AI integration.
• Further research with larger samples is needed to define AI's clinical impact and cost-benefit ratio.