Vessel-Aware Deep Learning for OCTA-Based Detection of AMD

2026-03-06

AI Summary

This study addresses the limitation of existing deep learning approaches for detecting age-related macular degeneration (AMD) from optical coherence tomography angiography (OCTA), which typically rely on global features while overlooking clinically relevant vascular biomarkers. To overcome this, the authors propose an external multiplicative attention framework that, for the first time, integrates multiscale-smoothed, vessel-specific biomarker maps—including tortuosity and perfusion deficit maps of arteries, veins, and capillaries—into a deep classification network to guide attention toward AMD-related pathological regions. Experimental results demonstrate that arterial tortuosity is the most discriminative biomarker, while capillary perfusion deficit maps under large-scale smoothing achieve optimal performance among density-based metrics. The proposed method significantly enhances both the accuracy and interpretability of AMD detection, aligning closely with established pathophysiological mechanisms.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Multiagent Systems: Other Foundations of Multi Agent Systems

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Age-related macular degeneration (AMD) is characterized by early micro-vascular alterations that can be captured non-invasively using optical coherence tomography angiography (OCTA), yet most deep learning (DL) models rely on global features and fail to exploit clinically meaningful vascular biomarkers. We introduce an external multiplicative attention framework that incorporates vessel-specific tortuosity maps and vasculature dropout maps derived from arteries, veins, and capillaries. These biomarker maps are generated from vessel segmentations and smoothed across multiple spatial scales to highlight coherent patterns of vascular remodeling and capillary rarefaction. Tortuosity reflects abnormalities in vessel geometry linked to impaired auto-regulation, while dropout maps capture localized perfusion deficits that precede structural retinal damage. The maps are fused with the OCTA projection to guide a deep classifier toward physiologically relevant regions. Arterial tortuosity provided the most consistent discriminative value, while capillary dropout maps performed best among density-based variants, especially at larger smoothing scales. Our proposed method offers interpretable insights aligned with known AMD pathophysiology.

Problem

• age-related macular degeneration
• optical coherence tomography angiography
• vascular biomarkers
• deep learning
• micro-vascular alterations

Innovation

• vessel-aware attention
• tortuosity maps
• vasculature dropout maps
• OCTA
• interpretable deep learning