• 4d flow
• aorta
• cardiovascular
• cfd
• deep learning
• mri
• neural network

Cardiovascular diseases remain a leading cause of mortality in developed countries. While modern imaging techniques have improved anatomical visualization, they often lack detailed hemodynamic information, which is crucial for accurate diagnosis and prognosis. 4D flow MRI is an innovative technology that provides both anatomical and functional data, including the 3D velocity field of blood flow. However, its limitations, such as low spatial resolution and complex processing, hinder its clinical adoption.

To address these challenges, data-driven techniques and Computational Fluid Dynamics (CFD) have emerged as key tools for improving segmentation and flow analysis. However, medical data collection remains difficult, making synthetic data generation a promising alternative to enhance model reliability.

This thesis develops a workflow for 4D flow MRI processing, integrating CFD-based synthetic data, generative adversarial networks (cGANs), neural representations for arterial wall motion, and a Python tool for feature extraction. These advancements enhance segmentation accuracy, data availability, and dynamic modeling, with future work focusing on clinical integration and complex geometries.