A highly parallel simulation of patient-specific hepatic flows

Area of Science:

• Computational fluid dynamics
• Medical imaging
• Hepatology

Background:

• Computational hemodynamics aids liver disease diagnosis and treatment planning.
• Current methods face challenges with computational time and accuracy, especially with complex geometries like arterial plaque.
• Simulating full 3D patient-specific hepatic vasculature is computationally intensive.

Purpose of the Study:

• To develop and investigate a highly parallel method for simulating transient incompressible Navier-Stokes equations.
• To accurately model blood flow in patient-specific 3D hepatic artery, portal vein, and hepatic vein.
• To assess the robustness and scalability of the developed algorithm.

Main Methods:

• Implementation of a highly parallel algorithm for transient incompressible Navier-Stokes equations.
• Simulation of blood flow in full 3D patient-specific hepatic vasculature.
• Application to a patient with hepatectomy, including calculation of S (PPG).

Main Results:

• Achieved 83% parallel efficiency for a 7 million element problem on over 1000 cores.
• Demonstrated accurate simulation of blood flow in complex patient-specific liver vasculature.
• Enabled direct estimation of S (PPG) for portal hypertension assessment.

Conclusions:

• The developed parallel method offers an accurate and efficient approach for computational hemodynamics in liver diseases.
• Simulating all hepatic vessels provides a direct and reliable method for portal hypertension assessment.
• The algorithm exhibits excellent scalability and robustness for clinical applications.