Predicting the mechanical properties of biopolymer gels using neural networks trained on discrete fiber network data

Leng, Y.X.; Taç, Vahidullah; Calve, Sarah; Tepole, Adrián Buganza

doi:10.1016/j.cma.2021.114160

Cited by 22 publications

(9 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When trained on various experimental data, the NODE has no problem fitting the data, see Figs. (5)(6)(7)(8).…”

Section: Discussionmentioning

confidence: 99%

“…Liu et al [5] enforced convexity of strain energy with respect to the elements of the Green strain tensor by adding loss terms that ensure the Hessian matrix of the strain energy is positive semi-definite. Other studies [7,8] used the monotonicity of stress as an alternative criterion to enforce the same class of convexity. However, polyconvexity is the more physically relevant criterion.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Data-Driven Anisotropic Finite Viscoelasticity Using Neural Ordinary Differential Equations

Taç

Rausch

Costabal

et al. 2023

Preprint

View full text Add to dashboard Cite

“…When trained on various experimental data, the NODE has no problem fitting the data, see Figs. (5)(6)(7)(8).…”

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Data-Driven Anisotropic Finite Viscoelasticity Using Neural Ordinary Differential Equations

Taç

Rausch

Costabal

et al. 2023

Preprint

View full text Add to dashboard Cite

“…Recently, we and other research groups have developed constitutive models based on DNNs [45][46][47][48][49][50][51], which perform better than the expert-prescribed models in some applications. However, it can be difficult to implement a complex DNN as a user subroutine in the programming language that is compatible with commercial software packages, e.g., FORTRAN for Abaqus UMAT.…”

Section: Discussionmentioning

confidence: 99%

PyTorch-FEA: Autograd-enabled Finite Element Analysis Methods with Applications for Biomechanical Analysis of Human Aorta

Liang

Liu

Elefteriades

et al. 2023

Preprint

View full text Add to dashboard Cite

Motivation: Finite-element analysis (FEA) is widely used as a standard tool for stress and deformation analysis of solid structures, including human tissues and organs. For instance, FEA can be applied at a patient-specific level to assist in medical diagnosis and treatment planning, such as risk assessment of thoracic aortic aneurysm rupture/dissection. These FEA-based biomechanical assessments often involve both forward and inverse mechanics problems. Current commercial FEA software packages (e.g., Abaqus) and inverse methods exhibit performance issues in either accuracy or speed. Methods: In this study, we propose and develop a new library of FEA code and methods, named PyTorch-FEA, by taking advantage of autograd, an automatic differentiation mechanism in PyTorch. We develop a class of PyTorch-FEA functionalities to solve forward and inverse problems with improved loss functions, and we demonstrate the capability of PyTorch-FEA in a series of applications related to human aorta biomechanics. In one of the inverse methods, we combine PyTorch-FEA with deep neural networks (DNNs) to further improve performance. Results: We applied PyTorch-FEA in four fundamental applications for biomechanical analysis of human aorta. In the forward analysis, PyTorch-FEA achieved a significant reduction in computational time without compromising accuracy compared with Abaqus, a commercial FEA package. Compared to other inverse methods, inverse analysis with PyTorch-FEA achieves better performance in either accuracy or speed, or both if combined with DNNs.

show abstract

“…Various models have been proposed to model the multiscale mechanical behavior of 3D hydrogel networks. For example, discrete fiber networks (DFN) models with defined parameters for fiber length, diameter, orientation distribution, and volume fraction within a representative volume element (RVE) have been proposed [27][28][29]. DFN models can then be used to inform analytical or data-driven models representative of an average RVE response [29].…”

Section: Introductionmentioning

confidence: 99%

“…For example, discrete fiber networks (DFN) models with defined parameters for fiber length, diameter, orientation distribution, and volume fraction within a representative volume element (RVE) have been proposed [27][28][29]. DFN models can then be used to inform analytical or data-driven models representative of an average RVE response [29]. Alternatively, fitting analytical models at the macroscale that also incorporate mesoscale information such as fiber orientation distribution and volume fraction is also possible [30].…”

Section: Introductionmentioning

confidence: 99%

Multiscale Mechanical Characterization and Computational Modeling of Fibrin Gels

Jimenez¹,

Tuttle

Guo

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Fibrin is a naturally occurring protein network that forms a temporary structure to enable remodeling during wound healing. It is also a common tissue engineering scaffold because the structural properties can be controlled. However, to fully characterize the wound healing process and improve the design of regenerative scaffolds, understanding fibrin mechanics at multiple scales is necessary. Here, we present a strategy to quantify both the macroscale (1 – 10 mm) stress-strain response and the deformation of the mesoscale (10 – 1000 μm) network structure during unidirectional tensile tests. The experimental data is then used to inform a computational model to accurately capture the mechanical response of fibrin gels. Simultaneous mechanical testing and confocal microscopy imaging of fluorophore-conjugated fibrin gels revealed up to an 88% decrease in volume coupled with increase in volume fraction in deformed gels, and non-affine fiber alignment in the direction of deformation. Combination of the computational model with finite element analysis enabled us to predict the strain fields that were observed experimentally within heterogenous fibrin gels with spatial variations in material properties. These strategies can be expanded to characterize and predict the macroscale mechanics and mesoscale network organization of other heterogeneous biological tissues and matrices.

show abstract

Predicting the mechanical properties of biopolymer gels using neural networks trained on discrete fiber network data

Cited by 22 publications

References 51 publications

Data-Driven Anisotropic Finite Viscoelasticity Using Neural Ordinary Differential Equations

Data-Driven Anisotropic Finite Viscoelasticity Using Neural Ordinary Differential Equations

PyTorch-FEA: Autograd-enabled Finite Element Analysis Methods with Applications for Biomechanical Analysis of Human Aorta

Multiscale Mechanical Characterization and Computational Modeling of Fibrin Gels

Contact Info

Product

Resources

About