Clinical Graph-Mediated Distillation for Unpaired MRI-to-CFI Hypertension Prediction

2026-03-23

AI Summary

This study addresses the challenge of leveraging high-cost MRI data to improve hypertension prediction from low-cost fundus images when paired multi-modal data are unavailable. To this end, the authors propose the Clinical Graph-Mediated Distillation (CGMD) framework, which introduces, for the first time in unpaired multi-modal medical learning, a kNN similarity graph constructed from clinical biomarkers. This graph facilitates cross-modal representation alignment and knowledge transfer by propagating representations over its structure and jointly optimizing both target-level and relation-level distillation objectives. Experiments on a newly curated unpaired MRI–fundus dataset demonstrate that CGMD significantly outperforms standard knowledge distillation and non-graph imputation baselines, confirming the efficacy of incorporating clinical graph structures to enhance fundus-based hypertension prediction.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Retinal fundus imaging enables low-cost and scalable hypertension (HTN) screening, but HTN-related retinal cues are subtle, yielding high-variance predictions. Brain MRI provides stronger vascular and small-vessel-disease markers of HTN, yet it is expensive and rarely acquired alongside fundus images, resulting in modality-siloed datasets with disjoint MRI and fundus cohorts. We study this unpaired MRI-fundus regime and introduce Clinical Graph-Mediated Distillation (CGMD), a framework that transfers MRI-derived HTN knowledge to a fundus model without paired multimodal data. CGMD leverages shared structured biomarkers as a bridge by constructing a clinical similarity kNN graph spanning both cohorts. We train an MRI teacher, propagate its representations over the graph, and impute brain-informed representation targets for fundus patients. A fundus student is then trained with a joint objective combining HTN supervision, target distillation, and relational distillation. Experiments on our newly collected unpaired MRI-fundus-biomarker dataset show that CGMD consistently improves fundus-based HTN prediction over standard distillation and non-graph imputation baselines, with ablations confirming the importance of clinically grounded graph connectivity. Code is available at https://github.com/DillanImans/CGMD-unpaired-distillation.

Problem

• hypertension prediction
• unpaired multimodal learning
• retinal fundus imaging
• brain MRI
• modality gap

Innovation

• unpaired multimodal distillation
• clinical graph
• knowledge transfer
• hypertension prediction
• retinal fundus imaging