In silico simulation of hepatic arteries: An open-source algorithm for efficient synthetic data generation.

Area of Science:

• Medical Imaging
• Computational Anatomy
• Biomedical Engineering

Background:

• In silico testing of novel imaging algorithms requires realistic arterial models.
• Data synthesis for deep learning necessitates efficient and random arterial tree generation.

Purpose of the Study:

• To present a method for generating hepatic arterial trees that are anatomically and physiologically motivated.
• To ensure the generated trees are computationally efficient, random, and high-resolution.

Main Methods:

• Utilized a constrained constructive optimization approach with volume minimization.
• Incorporated Couinaud liver classification for segment-specific feeding arteries.
• Ensured non-intersecting vasculature and smooth vessel curvature using cubic polynomial fits.

Main Results:

• Generated a 40,000-branch hepatic arterial tree in 11 seconds.
• Achieved realistic morphological features, including branching angles and radii, with non-intersecting vessels.
• Demonstrated randomness (variability = 0.98 ± 0.01) and assured main feeding arteries to each Couinaud segment.

Conclusions:

• Facilitates the creation of large datasets of high-resolution hepatic angiograms.
• Supports training of deep learning algorithms for medical imaging.
• Enables initial testing of novel 3D reconstruction and quantitative algorithms for interventional imaging.