Isfahan Artificial Intelligence Event 2023: Macular Pathology Detection Competition

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Computer-aided diagnosis (CAD) systems, particularly those utilizing artificial intelligence (AI), are increasingly vital for diagnosing macular diseases.
• AI-based CADs offer enhanced speed, objectivity, and thoroughness, serving as valuable assistance tools for clinicians.
• Retinal AI-CADs specifically aid ophthalmologists in analyzing optical coherence tomography (OCT) images, improving diagnostic accuracy and accessibility.

Purpose of the Study:

• To evaluate current AI-based classification methods for macular pathology detection through the Isfahan AI-2023 challenge.
• To provide objective, formal evaluations of alternative AI tools for retinal diagnostics.
• To identify the most successful algorithms for analyzing OCT images in macular disease diagnosis.

Main Methods:

• A curated dataset of OCT images from normal subjects and patients with macular disorders (diabetic macular edema, etc.) was provided.
• The dataset included labeled training and unlabeled test sets, accessible to challenge participants.
• Algorithms were developed and tested by researchers aiming to maximize performance on the test set labels.

Main Results:

• Multiple submissions demonstrated significant interest in AI-CAD technology for macular pathology.
• Deep learning-based methods proved capable of learning critical features from pathological OCT images.
• The competition facilitated the comparison and evaluation of various AI classification algorithms.

Conclusions:

• The Isfahan AI-2023 challenge successfully evaluated AI-based classification methods for macular diseases.
• Deep learning approaches show promise but require careful model selection for imbalanced datasets.
• Further development in AI-CAD technology is crucial for advancing eye care, especially for underserved populations.