Enhancing Fundus Image-based Glaucoma Screening via Dynamic Global-Local Feature Integration

2025-04-01

AI Summary

To address the challenges of high variability in fundus image quality, poor cross-device and cross-ethnic generalizability, and ambiguous cup-to-disc boundary delineation—leading to insufficient classification robustness in glaucoma diagnosis—this paper proposes a dynamic global-local feature fusion framework. Our method introduces two key innovations: (1) an adaptive attention window mechanism that automatically identifies optimal anatomical regions for optic cup and disc localization; and (2) a multi-head attention-driven linear readout strategy that explicitly models structural consistency (global) and local discriminability (local). Evaluated on a large-scale, real-world clinical dataset encompassing multiple centers, imaging devices, and ethnic groups, our approach achieves significant improvements in classification accuracy and cross-domain generalization, consistently outperforming state-of-the-art methods in robustness and reliability.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Data transparency and provenance

Abstract

With the advancements in medical artificial intelligence (AI), fundus image classifiers are increasingly being applied to assist in ophthalmic diagnosis. While existing classification models have achieved high accuracy on specific fundus datasets, they struggle to address real-world challenges such as variations in image quality across different imaging devices, discrepancies between training and testing images across different racial groups, and the uncertain boundaries due to the characteristics of glaucomatous cases. In this study, we aim to address the above challenges posed by image variations by highlighting the importance of incorporating comprehensive fundus image information, including the optic cup (OC) and optic disc (OD) regions, and other key image patches. Specifically, we propose a self-adaptive attention window that autonomously determines optimal boundaries for enhanced feature extraction. Additionally, we introduce a multi-head attention mechanism to effectively fuse global and local features via feature linear readout, improving the model's discriminative capability. Experimental results demonstrate that our method achieves superior accuracy and robustness in glaucoma classification.

Problem

• Addressing image quality variations across different devices
• Resolving discrepancies in training and testing images among races
• Improving uncertain boundaries in glaucomatous case detection

Innovation

• Self-adaptive attention window for optimal boundaries
• Multi-head attention mechanism for feature fusion
• Dynamic global-local feature integration for robustness