Influences of brain tissue poroelastic constants on intracranial pressure (ICP) during constant-rate infusion

Abstract

A 3D finite element (FE) model has been developed to study the mean intracranial pressure (ICP) response during constant-rate infusion using linear poroelasticity. Due to the uncertainties in the poroelastic constants for brain tissue, the influence of each of the main parameters on the transient ICP infusion curve was studied. As a prerequisite for transient analysis, steady-state simulations were performed first. The simulated steady-state pressure distribution in the brain tissue for a normal cerebrospinal fluid (CSF) circulation system showed good correlation with experiments from the literature. Furthermore, steady-state ICP closely followed the infusion experiments at different infusion rates. The verified steady-state models then served as a baseline for the subsequent transient models. For transient analysis, the simulated ICP shows a similar tendency to that found in the experiments, however, different values of the poroelastic constants have a significant effect on the infusion curve. The influence of the main poroelastic parameters including the Biot coefficient , Skempton coefficient B, drained Young's modulus E, Poisson's ratio , permeability , CSF absorption conductance and external venous pressure was studied to investigate the influence on the pressure response. It was found that the value of the specific storage term is the dominant factor that influences the infusion curve, and the drained Young's modulus E was identified as the dominant parameter second to . Based on the simulated infusion curves from the FE model, artificial neural network (ANN) was used to find an optimised parameter set that best fit the experimental curve. The infusion curves from both the FE simulation and using ANN confirmed the limitation of linear poroelasticity in modelling the transient constant-rate infusion.

Keywords::

• finite element model
• CSF dynamics
• parametric study
• specific storage term
• artificial neural network

Acknowledgements

We are grateful for the technical support from COMSOL support on the implementation of poroelasticity in the software. We thank the anonymous reviewers for their valuable suggestions for revision. We would like to acknowledge financial support provided by the Swedish Research Council D.nr. 621-2008-3400, the Swedish Governmental Agency for Innovation Systems (VINNOVA) and the Chinese Council Scholarship (CSC) for the first author.

Notes

^1. Email: [email protected]

^2. Email: [email protected]

^3. Here, the time-infusion curves at were discretized at 16 time points.

^4. Note that the curves with pressure jumping were excluded since the pressure jumping is physically unrealistic (see Section 4).