Related Experiment Video
Updated: Aug 5, 2026
A Three-Dimensional Digital Model for Early Diagnosis of Hepatic Fibrosis Based on Magnetic Resonance Elastography
Published on: July 21, 2023
Uncertainty quantification in FEM simulation of human liver: sensitivity analysis, Gradient-Enhanced Kriging, and
Navina Waschinsky1, Carmen van Meegen2, Lena Lambers1
1Faculty of Aerospace Engineering and Geodesy, Institute of Structural Mechanics and Dynamics in Aerospace Engineering (ISD), University of Stuttgart, Pfaffenwaldring 27, 70191, Stuttgart, Germany.
This study enhances liver perfusion-function models by incorporating material variability using sensitivity analysis and surrogate modeling. This approach provides probabilistic assessments for complex simulations, improving accuracy and interpretability.
Area of Science:
- Computational biology
- Biomedical engineering
- Mathematical modeling
Background:
- Numerical simulations of liver metabolic processes face challenges in accuracy, physiological representation, and data uncertainty.
- High computational costs limit current models to deterministic scenarios, neglecting biological variability.
- Existing models struggle to capture the complex, time-dependent behavior of soft tissues like the liver.
Purpose of the Study:
- To extend a deterministic liver perfusion-function model (Finite Element Method) to incorporate material variability.
- To develop a workflow integrating sensitivity analysis, surrogate modeling (Gradient-Enhanced Kriging), and uncertainty quantification.
- To enable identification of critical parameters, enhance model interpretability, and provide probabilistic outcome assessments.
Main Methods:
- A multiscale, poroelastic framework representing liver tissue on the lobule scale with coupled function-perfusion dynamics.
- Integration of sensitivity analysis to identify critical parameters for a Gradient-Enhanced Kriging (GEK) metamodel.
- Application of Monte Carlo (MC) methods on the surrogate model for uncertainty propagation and probabilistic outcome assessment.
Main Results:
- A workflow was presented to systematically incorporate material variability into liver perfusion-function models.
- Sensitivity analysis successfully identified critical parameters for GEK metamodel development.
- The surrogate model enabled efficient uncertainty quantification and probabilistic outcome prediction for FEM simulations.
Conclusions:
- The developed approach provides a systematic method to explore and interpret liver model behavior under uncertainty, moving beyond deterministic assumptions.
- Probabilistic outcome assessments enhance the understanding of liver function and disease simulation, such as tumor growth in metabolic associated fatty liver disease (MAFLD).
- This framework improves the reliability and interpretability of computational models for human liver physiology and pathology.
