Development and characterization of a finite element model of lung motion

BACKGROUND: Finite element models of lung motion can aid in understanding mechanically driven lung deformation. Current finite element models consider each lung half as a continuum, lacking the ability to capture the displacement discontinuity at fissures caused by lobe sliding.;OBJECTIVE: The objective of this work was to develop and evaluate finite element models for simulating lung motion that incorporate the role of sliding at the lobe boundaries.;METHODS: Finite element models were developed from 4DCT of tidal breathing from five cancer subjects. To allow sliding, the lobes were modeled as independent bodies within a pleural cavity shell. Pleural cavity deformation was obtained from deformable image registration of the lung segmentations. Contact between the pleural cavity and lobes prevented penetration and allowed sliding at all interfaces. Lung parenchyma was modeled as a homogeneous, 2-parameter, Neo-Hookean finite elastic model. The parameters of the Neo-Hookean model, C1 and D1, were optimized by perturbation within realistic reported ranges; defined by the equivalent infinitesimal elasticity parameters: Young's modulus (from 0.7 kPa to 70 kPa) and v (from 0.2 to 0.49). The frictional coefficient at fissures was perturbed between 0 (free sliding) and 1.5 (no sliding). 1,960 finite element analyses were performed across the five subjects. The optimal parameter ranges were evaluated by average landmark error and percentage of converged solutions.;The developed finite element method, using optimized material and friction parameters, was further evaluated in a data set of six healthy subjects with image pairs spanning functional residual capacity (FRC) to total lung capacity (TLC). The finite element predicted displacement field for lobe sliding finite element models and continuum-based finite element models were compared using average landmark error and correlation with the lobe-by-lobe deformable image registration results.;RESULTS AND DISCUSSION: The optimal parameters for Young's modulus were 49 kPa to 70 kPa and Poisson's ratio were 0.2 to 0.4. Variation of inter-lobar frictional coefficients did change displacement field accuracy assessed by landmark error or correlation to lobe-by-lobe deformable image registration. Characteristics of sliding predicted by the lobe sliding finite element models were consistent with characteristics in sliding observed in deformable image registration results. Also, variations in regional ventilation, quantified at the lobe level, were predicted by the finite element models and were shown to be influenced by the amount of lobe sliding allowed by the models.