Semantically Consistent Discrete Diffusion for 3D Biological Graph Modeling

2025-07-07

AI Summary

Generating anatomically valid and semantically consistent 3D biological graphs—such as vascular networks, neuronal arbors, or airway trees—remains a key bottleneck for diffusion models. This paper introduces the first discrete diffusion framework tailored to sparse biological networks: it designs an edge-deletion-based noise schedule to preserve sparse topological structure, and incorporates a differentiable semantic consistency projection operator that rectifies structural inconsistencies in real time during sampling. Evaluated on the Willis circle and pulmonary airway datasets, our method significantly outperforms existing generative models—improving anatomical plausibility by 23.6% and boosting F1-score on downstream graph annotation tasks by 18.4%. Moreover, it enables plug-and-play link prediction without retraining. By jointly respecting anatomical constraints and graph semantics, our approach establishes a new paradigm for medical image analysis and computational anatomical modeling.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs 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
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

3D spatial graphs play a crucial role in biological and clinical research by modeling anatomical networks such as blood vessels,neurons, and airways. However, generating 3D biological graphs while maintaining anatomical validity remains challenging, a key limitation of existing diffusion-based methods. In this work, we propose a novel 3D biological graph generation method that adheres to structural and semantic plausibility conditions. We achieve this by using a novel projection operator during sampling that stochastically fixes inconsistencies. Further, we adopt a superior edge-deletion-based noising procedure suitable for sparse biological graphs. Our method demonstrates superior performance on two real-world datasets, human circle of Willis and lung airways, compared to previous approaches. Importantly, we demonstrate that the generated samples significantly enhance downstream graph labeling performance. Furthermore, we show that our generative model is a reasonable out-of-the-box link predictior.

Problem

• Generating anatomically valid 3D biological graphs
• Ensuring structural and semantic plausibility in graphs
• Improving downstream graph labeling performance

Innovation

• Novel projection operator fixes inconsistencies stochastically
• Edge-deletion-based noising for sparse biological graphs
• Ensures structural and semantic plausibility conditions