Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Dillan Imans, Phuoc-Nguyen Bui, Duc-Tai Le, Hyunseung Choo

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.

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

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.
Paper Structure (9 sections, 9 equations, 2 figures, 2 tables)

This paper contains 9 sections, 9 equations, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. MRI Teacher and Smoothed Priors
  4. Cross-cohort Prior Imputation
  5. Student training with prior and relational distillation
  6. Experiments
  7. Performance evaluation
  8. Conclusion

Figures (2)

  • Figure 1: Overview of the proposed CGMD. MRI teacher embeddings are smoothed on a clinical $k$NN graph to impute fundus priors, and the fundus student is trained with supervised, prior, and relational distillation (MRI-free inference).
  • Figure 2: Impact of (a) data availability and (b) prior construction strategies on the final HTN classification performance.