PRETI: Patient-Aware Retinal Foundation Model via Metadata-Guided Representation Learning

2025-05-18

AI Summary

To address the scarcity of clinical annotations and poor generalizability in retinal image analysis, this paper proposes a patient-aware retinal foundation model. Methodologically, it introduces two key innovations: (1) Learnable Metadata Embeddings (LME), which incorporate readily available clinical metadata—such as age and sex—into self-supervised representation learning; and (2) Retina-Aware Adaptive Masking (RAAM), a region-sensitive, anatomy-informed dynamic masking strategy. The model is jointly optimized via metadata-guided contrastive learning and masked modeling. Evaluated on ophthalmic disease diagnosis and biomarker prediction tasks, it achieves state-of-the-art performance while demonstrating significantly improved cross-center generalizability. The source code and pre-trained models are publicly released.

Technology Category

• Computer Vision: Other Foundations of Computer Vision
• Machine Learning: Other Foundations of Machine Learning
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Retinal foundation models have significantly advanced retinal image analysis by leveraging self-supervised learning to reduce dependence on labeled data while achieving strong generalization. Many recent approaches enhance retinal image understanding using report supervision, but obtaining clinical reports is often costly and challenging. In contrast, metadata (e.g., age, gender) is widely available and serves as a valuable resource for analyzing disease progression. To effectively incorporate patient-specific information, we propose PRETI, a retinal foundation model that integrates metadata-aware learning with robust self-supervised representation learning. We introduce Learnable Metadata Embedding (LME), which dynamically refines metadata representations. Additionally, we construct patient-level data pairs, associating images from the same individual to improve robustness against non-clinical variations. To further optimize retinal image representation, we propose Retina-Aware Adaptive Masking (RAAM), a strategy that selectively applies masking within the retinal region and dynamically adjusts the masking ratio during training. PRETI captures both global structures and fine-grained pathological details, resulting in superior diagnostic performance. Extensive experiments demonstrate that PRETI achieves state-of-the-art results across diverse diseases and biomarker predictions using in-house and public data, indicating the importance of metadata-guided foundation models in retinal disease analysis. Our code and pretrained model are available at https://github.com/MICV-yonsei/PRETI

Problem

• Reduces reliance on costly clinical reports for retinal analysis
• Integrates patient metadata to enhance disease progression understanding
• Improves retinal image representation via adaptive masking strategies

Innovation

• Uses metadata-aware learning for retinal analysis
• Introduces Learnable Metadata Embedding (LME)
• Proposes Retina-Aware Adaptive Masking (RAAM)