Automatic detection and prediction of nAMD activity change in retinal OCT using Siamese networks and Wasserstein Distance for ordinality

2025-01-24

AI Summary

To address the clinical challenge of detecting disease activity and predicting progression in neovascular age-related macular degeneration (nAMD), this work proposes an automated longitudinal analysis method leveraging retinal optical coherence tomography (OCT) volumes. Methodologically, we introduce Wasserstein distance into the loss function to explicitly encode the ordinal nature of nAMD severity transitions, and design a Vision Transformer–Siamese (ViT-Siamese) architecture to quantify semantic differences between OCT scans acquired at different time points. Evaluated on the MICCAI 2024 MARIO Challenge preliminary phase, our approach achieves top-tier performance in both disease activity detection and prediction of severity change at the three-month horizon. These results demonstrate its efficacy and clinical potential for supporting personalized therapeutic decision-making in nAMD management.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Neovascular age-related macular degeneration (nAMD) is a leading cause of vision loss among older adults, where disease activity detection and progression prediction are critical for nAMD management in terms of timely drug administration and improving patient outcomes. Recent advancements in deep learning offer a promising solution for predicting changes in AMD from optical coherence tomography (OCT) retinal volumes. In this work, we proposed deep learning models for the two tasks of the public MARIO Challenge at MICCAI 2024, designed to detect and forecast changes in nAMD severity with longitudinal retinal OCT. For the first task, we employ a Vision Transformer (ViT) based Siamese Network to detect changes in AMD severity by comparing scan embeddings of a patient from different time points. To train a model to forecast the change after 3 months, we exploit, for the first time, an Earth Mover (Wasserstein) Distance-based loss to harness the ordinal relation within the severity change classes. Both models ranked high on the preliminary leaderboard, demonstrating that their predictive capabilities could facilitate nAMD treatment management.

Problem

• Automatic Detection
• Age-related Macular Degeneration
• Optical Coherence Tomography

Innovation

• Siamese Network
• Wasserstein Distance
• Automated AMD Progression Prediction