Detecting Macular Disease Based on Optical Coherence Tomography Using a Deep Convolutional Network.

Area of Science:

• Ophthalmology
• Artificial Intelligence
• Medical Imaging

Background:

• Neovascular age-related macular degeneration (nAMD) and central serous chorioretinopathy (CSC) are leading causes of vision impairment.
• Accurate classification of nAMD and CSC subtypes is crucial for effective treatment.

Purpose of the Study:

• To develop and validate a deep learning model for classifying nAMD and CSC subtypes using spectral-domain optical coherence tomography (SD-OCT) images.
• To compare the model's diagnostic performance against ophthalmologists.

Main Methods:

• A convolutional neural network (CNN) model was trained on 6063 SD-OCT images.
• Utilized VGG-16, VGG-19, and ResNet architectures with transfer learning and data augmentation.
• External validation and comparison with eight ophthalmologists were performed.

Main Results:

• The model achieved 99.7% accuracy for nAMD classification and 91.1% for CSC classification.
• Demonstrated superior classification accuracy compared to human experts in external testing.
• Gradient-weighted class activation mapping confirmed clinically relevant feature identification.

Conclusions:

• The proposed CNN model offers a highly accurate and reliable tool for classifying nAMD and CSC subtypes from SD-OCT images.
• This AI-driven approach has the potential to aid clinicians in diagnosing and managing macular diseases.
• The model's performance suggests its clinical utility in ophthalmological practice.