TY - JOUR
T1 - Predictive constitutive modelling of arteries by deep learning
AU - Holzapfel, Gerhard A.
AU - Linka, Kevin
AU - Sherifova, Selda
AU - Cyron, Christian J.
N1 - Funding Information:
Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—257981274. C.J.C. greatfully acknowledges financial support from TU Hamburg within the I3-Lab ‘Modellgestütztes maschinelles Lernen für die Weichgewebsmodellierung in der Medizin’. The work of G.A.H. and S.S. was partially supported by the Lead Project on ‘Mechanics, Modeling and Simulation of Aortic Dissection’ granted by Graz University of Technology, Austria.
Publisher Copyright:
© 2021 The Authors.
PY - 2021/9/8
Y1 - 2021/9/8
N2 - The constitutive modelling of soft biological tissues has rapidly gained attention over the last 20 years. Current constitutive models can describe the mechanical properties of arterial tissue. Predicting these properties from microstructural information, however, remains an elusive goal. To address this challenge, we are introducing a novel hybrid modelling framework that combines advanced theoretical concepts with deep learning. It uses data from mechanical tests, histological analysis and images from second-harmonic generation. In this first proof of concept study, our hybrid modelling framework is trained with data from 27 tissue samples only. Even such a small amount of data is sufficient to be able to predict the stress-stretch curves of tissue samples with a median coefficient of determination of R 2 = 0.97 from microstructural information, as long as one limits the scope to tissue samples whose mechanical properties remain in the range commonly encountered. This finding suggests that deep learning could have a transformative impact on the way we model the constitutive properties of soft biological tissues.
AB - The constitutive modelling of soft biological tissues has rapidly gained attention over the last 20 years. Current constitutive models can describe the mechanical properties of arterial tissue. Predicting these properties from microstructural information, however, remains an elusive goal. To address this challenge, we are introducing a novel hybrid modelling framework that combines advanced theoretical concepts with deep learning. It uses data from mechanical tests, histological analysis and images from second-harmonic generation. In this first proof of concept study, our hybrid modelling framework is trained with data from 27 tissue samples only. Even such a small amount of data is sufficient to be able to predict the stress-stretch curves of tissue samples with a median coefficient of determination of R 2 = 0.97 from microstructural information, as long as one limits the scope to tissue samples whose mechanical properties remain in the range commonly encountered. This finding suggests that deep learning could have a transformative impact on the way we model the constitutive properties of soft biological tissues.
KW - constitutive modelling
KW - data-driven modelling
KW - deep learning
KW - hybrid modelling
KW - soft biological tissues
UR - http://www.scopus.com/inward/record.url?scp=85115982101&partnerID=8YFLogxK
U2 - 10.1098/rsif.2021.0411
DO - 10.1098/rsif.2021.0411
M3 - Article
C2 - 34493095
AN - SCOPUS:85115982101
SN - 1742-5689
VL - 18
JO - Journal of the Royal Society Interface
JF - Journal of the Royal Society Interface
IS - 182
M1 - 20210411
ER -