MeshGrow: Integrated framework for simulation-ready cardiac and vascular mesh construction from medical imaging

Area of Science:

• Cardiovascular research
• Medical imaging
• Computational modeling

Background:

• Patient-specific cardiovascular simulations are crucial for research and clinical practice.
• Automating the construction of simulation-ready models from medical images is challenging.
• Existing methods are often limited to either cardiac or vascular models, not both.

Purpose of the Study:

• To introduce MeshGrow, a novel framework for automated construction of combined cardiac and vascular models.
• To enable patient-specific cardiovascular hemodynamic simulations by streamlining model generation.
• To address limitations of existing methods by integrating cardiac and vascular modeling.

Main Methods:

• MeshGrow employs a two-stage approach combining two machine learning techniques.
• Stage (a) focuses on meshing cardiac structures.
• Stage (b) involves growing the vasculature from the cardiac mesh, including the aorta and its branches.

Main Results:

• MeshGrow successfully reconstructed cardiac chambers, aorta, and major branches from CT data.
• The framework generated simulation-suitable meshes with defined surfaces for boundary conditions.
• On five CT datasets, MeshGrow outperformed state-of-the-art benchmark methods in metric scores.
• Three-dimensional computational fluid dynamics simulations were successfully run on two test cases.

Conclusions:

• MeshGrow offers an automated solution for creating integrated cardiac and vascular models.
• This advancement facilitates patient-specific cardiovascular hemodynamic simulations.
• The method has the potential to accelerate large cohort studies and clinical applications of cardiovascular modeling.