Recursive Variational Autoencoders for 3D Blood Vessel Generative Modeling

2025-06-17

AI Summary

Existing rule-based vascular modeling methods struggle to capture the topological and geometric diversity of real anatomical tree structures. To address this, we propose Recursive Variational Neural Networks (RvNN), the first framework to introduce recursive variational modeling into vascular generation. RvNN jointly learns low-dimensional manifold representations of branch connectivity and geometric features via a Recursive Variational Autoencoder (RvVAE) integrated with a hierarchical graph neural network, enabling end-to-end joint topological–geometric modeling. The generated vascular trees faithfully reproduce key morphological metrics—including radius, length, and tortuosity—across multi-source clinical datasets, including aneurysm cases. The method ensures anatomical fidelity, controllability, and structural diversity, making it suitable for medical simulation and hemodynamic analysis.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Segmentation
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Web data generation and simulation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Anatomical trees play an important role in clinical diagnosis and treatment planning. Yet, accurately representing these structures poses significant challenges owing to their intricate and varied topology and geometry. Most existing methods to synthesize vasculature are rule based, and despite providing some degree of control and variation in the structures produced, they fail to capture the diversity and complexity of actual anatomical data. We developed a Recursive variational Neural Network (RvNN) that fully exploits the hierarchical organization of the vessel and learns a low-dimensional manifold encoding branch connectivity along with geometry features describing the target surface. After training, the RvNN latent space can be sampled to generate new vessel geometries. By leveraging the power of generative neural networks, we generate 3D models of blood vessels that are both accurate and diverse, which is crucial for medical and surgical training, hemodynamic simulations, and many other purposes. These results closely resemble real data, achieving high similarity in vessel radii, length, and tortuosity across various datasets, including those with aneurysms. To the best of our knowledge, this work is the first to utilize this technique for synthesizing blood vessels.

Problem

• Modeling complex 3D blood vessel anatomy accurately
• Capturing diversity in vessel topology and geometry
• Generating realistic vasculature for medical applications

Innovation

• Recursive variational Neural Network for 3D vessels
• Learns low-dimensional manifold encoding branches
• Generates diverse and accurate blood vessel models