Synthetic Database of Aortic Morphometry and Hemodynamics: Overcoming Medical Imaging Data Availability

Area of Science:

• Cardiovascular research
• Medical imaging analysis
• Computational fluid dynamics

Background:

• Hemodynamics modeling and artificial intelligence (AI) show promise for clinical decision-making.
• Hemodynamics modeling is resource-intensive, and AI requires extensive data, often unavailable in clinical settings.

Purpose of the Study:

• To develop and evaluate a novel methodology for generating a large synthetic dataset of coarctation of the aorta (CoA) cases.
• To enable machine learning (ML) approaches for investigating aortic morphometric pathology and its hemodynamic impact.

Main Methods:

• Combined statistical shape modeling of aortic morphometry with aorta inlet flow fields and numerical flow simulations.
• Utilized hierarchical clustering and non-linear regression to analyze morphometry-hemodynamics relationships.
• Validated synthetic cohort credibility against a clinical cohort using MRI data from 154 patients and healthy subjects.

Main Results:

• Generated a database of 2652 synthetic cases with realistic anatomical and hemodynamic properties.
• Identified three distinct shape clusters with corresponding hemodynamic differences.
• The developed model accurately predicts the coarctation of the aorta pressure gradient with a root mean square error of 4.6 mmHg.

Conclusions:

• Synthetic data generation for anatomy and hemodynamics effectively addresses the scarcity of large clinical datasets.
• This novel approach provides a robust foundation for applying ML to uncover new insights into cardiovascular diseases like CoA.