Invited Session III: Machine Learning and AI Approaches to Retinal Diagnostics: Using AI with small datasets of...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Training artificial intelligence (AI) models typically demands extensive datasets.
• This presents significant hurdles for applying AI to rare diseases or novel imaging techniques.
• Processing data from bespoke instruments also poses challenges for AI development.

Purpose of the Study:

• To demonstrate the application of AI methodologies in retinal imaging.
• To address the challenges of limited data availability in AI for ophthalmology.
• To advance the analysis of retinal structure and function using AI.

Main Methods:

• Development and application of AI algorithms tailored for retinal imaging.
• Strategies for handling and processing data acquired from non-standardized instruments.
• Focus on AI techniques suitable for conditions with less common data.

Main Results:

• Progress in applying AI methods to analyze retinal structure.
• Advancements in using AI for assessing retinal function.
• Demonstrated feasibility of AI in retinal imaging despite data limitations.

Conclusions:

• AI can be effectively applied to retinal imaging even with limited datasets.
• The presented methods offer a pathway for AI integration in ophthalmology for rare conditions.
• Further development holds promise for improved diagnosis and monitoring of retinal diseases.