Stochastic finite element analysis of biological systems: comparison of a simple intervertebral disc model with experimental results

Statistical methods allow the effects of uncertainty to be incorporated into finite element models. This has potential benefits for the analysis of biological systems where natural variability can give rise to substantial uncertainty in both material and geometrical properties. In this study, a simple model of the intervertebral disc under compression was created and analysed as both a deterministic and a stochastic system. Factorial analysis was used to determine the important parameters to be included in the stochastic analysis. The predictions from the model were compared to experimental results from 21 sheep discs. The size and shape of the distribution of the axial deformations predicted by the model was consistent with the experimental results given that the number of model solutions far exceeded the number of experimental results. Stochastic models could be valuable in determining the range and most likely value of stress in a tissue or implant.     

Biological response of the intervertebral disc to dynamic loading

Disc degeneration is a chronic remodeling process that results in alterations of matrix composition and decreased cellularity. This study tested the hypothesis that dynamic mechanical forces are important regulators in vivo of disc cellularity and matrix synthesis. A murine model of dynamic loading was developed that used an external loading device to cyclically compress a single disc in the tail. Loads alternated at a 50% duty cycle between 0MPa and one of two peak stresses (0.9 or 1.3MPa) at one of two frequencies (0.1 or 0.01Hz) for 6h per day for 7 days. An additional group received static compression at 1.3MPa for 3h/day for 7 days. A control group wore the device with no loading. Sections of treated discs were analyzed for morphology, proteoglycan content, apoptosis, cell areal density, and aggrecan and collagen II gene expression. Dynamic loading induced differential effects that depended on frequency and stress. No significant changes to morphology, proteoglycan content or cell death were found after loading at 0.9MPa, 0.1Hz. Loading at lower frequency and/or higher stress increased proteoglycan content, matrix gene expression and cell death. The results have implications in the prevention of intervertebral disc degeneration, suggesting that loading conditions may be optimized to promote maintenance of normal structure and function.     

The internal mechanics of the intervertebral disc under cyclic loading

The mechanics of the intervertebral disc (IVD) under cyclic loading are investigated via a one-dimensional poroelastic model and experiment. The poroelastic model, based on that of Biot (J. Appl. Phys. 12 (1941) 155; J. Appl. Mech. 23 (1956) 91), includes a power-law relation between porosity and permeability, and a linear relation between the osmotic potential and solidity. The model was fitted to experimental data of the unconfined IVD undergoing 5 cyclic loads of 20 min compression by an applied stress of 1MPa, followed by 40 min expansion. To obtain a good agreement between experiment and theory, the initial elastic deformation of the IVD, possibly associated with the bulging of the IVD into the vertebral bodies or laterally, was removed from the experimental data. Many combinations of the permeability-porosity relationship with the initial osmotic potential (pi(i)) were investigated, and the best-fit parameters for the aggregate modulus (H(A)) and initial permeability (k(i)) were determined. The values of H(A) and k(i) were compared to literature values, and agreed well especially in the context of the adopted high-stress testing regime, and the strain related permeability in the model.     

The relation between intervertebral disc bulging and annular fiber associated strains for simple and complex loading

Mechanical failure of the annulus fibrosus is mostly indicated by tears, fissures, protrusions or disc prolapses. Some of these annulus failures can be caused by a high intradiscal pressure. This has an effect on disc bulging. However, it is not fully understood how disc bulging is related to disc loading. Therefore, the aim of this study was to investigate the annular fiber strains and disc bulging under simple and complex spinal loads. A novel laser scanner was used to image surfaces of six L2-3 segments. Specimens were loaded with 500 N or 7.5 Nm in a spine tester while acquiring surface maps. Loading was applied in the three principal main directions and four combined directions. Disc bulging and tissue surface strains in annulus collagen fiber directions were computed. Two conditions were measured; intact and defect (vertebral body-disc-body units). Axial compression resulted in 2.7% fiber associated strains in intact segments and the defect increased strains up to 6.7%. Disc bulging increased from 0.7 mm to 0.87 mm. Flexion produced 7.2% fiber associated strains and 1.63 mm bulge going up to 17.5% and 2.21 mm after the defect. Highest fiber associated strains were found for the combination of axial rotation plus lateral bending with 24.6% and with a maximal bulging of 1.14 mm. It was found that there is no tight relationship between fiber associated strains and disc bulging. This was especially seen for the load combinations. Highest fiber associated strains were found to be located in small posterolateral regions. Fiber associated strains had a much higher magnitude than previously reported fiber associated strains. The results showed that combined loading is most likely to produce higher associated fiber strains compared to single axis loading.     

The micromechanical environment of intervertebral disc cells determined by a finite deformation,anisotropic, and biphasic finite element model

Cellular response to mechanical loading varies between the anatomic zones of the intervertebral disc. This difference may be related to differences in the structure and mechanics of both cells and extracellular matrix, which are expected to cause differences in the physical stimuli (such as pressure, stress, and strain) in the cellular micromechanical environment. In this study, a finite element model was developed that was capable of describing the cell micromechanical environment in the intervertebral disc. The model was capable of describing a number of important mechanical phenomena: flow-dependent viscoelasticity using the biphasic theory for soft tissues; finite deformation effects using a hyperelastic constitutive law for the solid phase; and material anisotropy by including a fiber-reinforced continuum law in the hyperelastic strain energy function. To construct accurate finite element meshes, the in situ geometry of IVD cells were measured experimentally using laser scanning confocal microscopy and three-dimensional reconstruction techniques. The model predicted that the cellular micromechanical environment varies dramatically between the anatomic zones, with larger cellular strains predicted in the anisotropic anulus fibrosus and transition zone compared to the isotropic nucleus pulposus. These results suggest that deformation related stimuli may dominate for anulus fibrosus and transition zone cells, while hydrostatic pressurization may dominate in the nucleus pulposus. Furthermore, the model predicted that micromechanical environment is strongly influenced by cell geometry, suggesting that the geometry of IVD cells in situ may be an adaptation to reduce cellular strains during tissue loading.     

Initiation and progression of mechanical damage in the intervertebral disc under cyclic loading using continuum damage mechanics methodology: A finite element study

It is difficult to study the breakdown of disc tissue over several years of exposure to bending and lifting by experimental methods. There is also no finite element model that elucidates the failure mechanism due to repetitive loading of the lumbar motion segment. The aim of this study was to refine an already validated poro-elastic finite element model of lumbar motion segment to investigate the initiation and progression of mechanical damage in the disc under simple and complex cyclic loading conditions. Continuum damage mechanics methodology was incorporated into the finite element model to track the damage accumulation in the annulus in response to the repetitive loading. The analyses showed that the damage initiated at the posterior inner annulus adjacent to the endplates and propagated outwards towards its periphery under all loading conditions simulated. The damage accumulated preferentially in the posterior region of the annulus. The analyses also showed that the disc failure is unlikely to happen with repetitive bending in the absence of compressive load. Compressive cyclic loading with low peak load magnitude also did not create the failure of the disc. The finite element model results were consistent with the experimental and clinical observations in terms of the region of failure, magnitude of applied loads and the number of load cycles survived.     

An accurate finite element model of the cervical spine under quasi-static loading

Cervical disc injury due to impact has been observed in clinical and biomechanical investigations; however, there is a lack of data that helps to elucidate the mechanisms of disc injury during these collisions. Therefore, it is necessary to understand the behavior of the cervical spine under different types of loading situations. A three dimensional finite element (FE) model for the multi-level cervical spine segment (C0-C7) was developed using computed tomography (CT) data and applied to study the internal stresses and strains of the intervertebral discs under quasi-static loading conditions. The intervertebral discs were treated as nonlinear, anisotropic and incompressible subjected to large deformations. The model accuracy was validated by comparing it with previously published experimental and numerical results for different movements. It was shown that the use of a fiber reinforced model to describe the behavior of the annulus of the discs would predict higher maximum shear strains than an isotropic one, being therefore important the use of complex constitutive models in order to be able to detect the appearance of injured zones, since those strains and stresses are supposed to be related with damage to soft tissues. Several movements were analyzed: flexion, extension and axial rotation, obtaining that the maximum shear stresses in the disc were higher for a flexo-extension movement.     

A finite element model technique to determine the mechanical response of a lumbar spine segment under complex loads

This study presents a CT-based finite element model of the lumbar spine taking into account all function-related boundary conditions, such as anisotropy of mechanical properties, ligaments, contact elements, mesh size, etc. Through advanced mesh generation and employment of compound elements, the developed model is capable of assessing the mechanical response of the examined spine segment for complex loading conditions, thus providing valuable insight on stress development within the model and allowing the prediction of critical loading scenarios. The model was validated through a comparison of the calculated force-induced inclination/deformation and a correlation of these data to experimental values. The mechanical response of the examined functional spine segment was evaluated, and the effect of the loading scenario determined for both vertebral bodies as well as the connecting intervertebral disc.     

Mechanical damage to the intervertebral disc annulus fibrosus subjected to tensile loading

Damage of the annulus fibrosus is implicated in common spinal pathologies. The objective of this study was to obtain a quantitative relationship between both the number of cycles and the magnitude of tensile strain resulting in damage to the annulus fibrosus. Four rectangular tensile specimens oriented in the circumferential direction were harvested from the outer annulus of 8 bovine caudal discs (n = 32) and subjected to one of four tensile testing protocols: (i) ultimate tensile strain (UTS) test; (ii) baseline cyclic test with 4 series of 400 cycles of baseline cyclic loading (peak strain = 20% UTS); (iii & iv) acute and fatigue damage cyclic tests consisting of 4 x 400 cycles of baseline cyclic loading with intermittent loading to 1 and 100 cycles, respectively, with peak tensile strain of 40%, 60%, and 80% UTS. Normalized peak stress for all mechanically loaded specimens was reduced from 0.89 to 0.11 of the baseline control levels, and depended on the magnitude of damaging strain and number of cycles at that damaging strain. Baseline, acute, and fatigue protocols resulted in permanent deformation of 3.5%, 6.7% and 9.6% elongation, respectively. Damage to the laminate structure of the annulus in the absence of biochemical activity in this study was assessed using histology, transmission electron microscopy, and biochemical measurements and was most likely a result of separation of annulus layers (i.e., delamination). Permanent elongation and stress reduction in the annulus may manifest in the motion segment as sub-catastrophic damage including increased neutral zone, disc bulging, and loss of nucleus pulposus pressure. The preparation of rectangular tensile strip specimens required cutting of collagen fibers and may influence absolute values of results, however, it is not expected to affect the comparisons between loading groups or dose-response reported.     

Statistical factorial analysis approach for parameter calibration on material nonlinearity of intervertebral disc finite element model

In the biomechanics field, material parameters calibration is significant for finite element (FE) model to ensure a legit estimation of biomechanical response. Determining an appropriate combination of calibration factors is challenging as each constitutive component responds differently. This study proposes a statistical factorial analysis approach using L₁₆(4⁵) orthogonal array to evaluate material nonlinearity and applicable calibration factor of the intervertebral disc FE model in pure moment. The calibrated model exhibits improved agreement to the experimental findings for all directions. Appropriate combination of calibration parameter reduces the estimation gap to the experimental findings, ensuring agreeable biomechanical responses.     

Nonlinear numerical analysis of the structural response of the intervertebral disc to impact loading

The analysis of intervertebral disc dynamics under impact loading, using computational simulation, is scarcely reported. In this study, the contribution of the characteristic structure of the disc to its dynamic response has been evaluated. The influence of several model features on the dynamic response was investigated. A hyperelastic large deformation formulation was used to describe the nonlinear behaviour of the soft tissues. The material parameters were determined by the fitting of experimental data from the literature. The model demonstrated pressure wave propagation and reflection through the disc, with a periodic oscillation of the system in response to a single impulse load, and highlighted a potential primary role played by the collagen fibre reinforcement. Their tensioning contributes to changing the stress propagation and oscillation, with a faster reduction in the internal pressure peak. The natural frequency of the disc was predicted to be approximately 9.8 Hz for the vertical oscillation.     

A finite element model predicts the mechanotransduction response of tendon cells to cyclic tensile loading

The importance of fluid-flow-induced shear stress and matrix-induced cell deformation in transmitting the global tendon load into a cellular mechanotransduction response is yet to be determined. A multiscale computational tendon model composed of both matrix and fluid phases was created to examine how global tendon loading may affect fluid-flow-induced shear stresses and membrane strains at the cellular level. The model was then used to develop a quantitative experiment to help understand the roles of membrane strains and fluid-induced shear stresses on the biological response of individual cells. The model was able to predict the global response of tendon to applied strain (stress, fluid exudation), as well as the associated cellular response of increased fluid-flow-induced shear stress with strain rate and matrix-induced cell deformation with strain amplitude. The model analysis, combined with the experimental results, demonstrated that both strain rate and strain amplitude are able to independently alter rat interstitial collagenase gene expression through increases in fluid-flow-induced shear stress and matrix-induced cell deformation, respectively.     

Poroelastic finite element analysis of a bone specimen under cyclic loading

It had been suggested that the fluid embodied in bone lacunar-canalicular porosity may play an important role in bone remodelling [Weinbaum et al., 1994. Journal of Biomechanics 27, 339-360]. In this paper a finite element model of a poroelastic prismatic solid of rectangular cross-section is considered to simulate bone behaviour, precisely as in the previous work by Zhang and Cowin [Zhang and Cowin, 1994. Journal of Mechanical Physics of Solids 42, 1575-1599]. This solid is subject to combined cyclic axial and bending loads at its end. The objectives of the study are: (1) to verify the accuracy of the simplifying hypotheses underlying the analytical solutions established by the above authors; (2) to provide further insight into the behaviour of that solid; (3) to test the advantages in generality and versatility and the computing costs of general-purpose finite element codes in poroelastic analysis. The study is parametric with respect to the fluid leakage coefficient, to the ratio of the bending moment and axial load, and to the ratio of the characteristic relaxation time of the pore pressure over the excitation period. Results show that, for all the cases considered, the pore pressure distribution along the section height of the poroelastic beam exhibits a very good matching with previous analytical results. Stresses transversal with respect to the beam axis (assumed as constant or zero in previous analytical solutions) are evaluated. The analysis pointed out that: (1) the effects due to end-loads with zero resultants practically extinguish within a distance from the beam end almost equal to a typical length of the cross-section; (2) cross-sections remain plane above that distance; (3) the transversal total stresses are three orders of magnitude lower than axial stress.     

A 2D finite element model of the interventricular septum under normal and abnormal loading

The interventricular septum is the structure that separates the left and right ventricles of the heart. Under normal loading conditions, it is concave to the left ventricle, but under abnormal loading the septum flattens and occasionally inverts. In the past, the septum has frequently been modelled as integral to the left ventricle with the effects of pressure from the right ventricle being ignored. Under abnormal loading, the septum has been described as behaving equivalent to a "flapping sail". There has been no consideration of structural behaviour under these conditions. A 2-D plane stress FE model of the septum was used to investigate the difference in structural behaviour of the septum during diastole between normal and abnormal loading. The biaxial stress patterns that develop are distinctively disparate. Under normal loading, the septum behaves much like a thick-walled cylinder subject to internal and external pressure, with the resulting stresses being circumferential tension and radial compression, both varying with radius. These stresses are very low throughout most of diastole. However, under abnormal loading, the septum behaves in an arch-like fashion, with high compressive stresses almost circumferential in direction, combined with radial compression. We conclude that right ventricular pressures cause bending effects in the wall of the heart, and that under abnormal loading, the compressive stresses that develop in the septum may lead to an understanding of certain, previously unexplained, pathological conditions.     

Nonlinear finite element analysis of anular lesions in the L4/5 intervertebral disc

Degenerate intervertebral discs exhibit both material and structural changes. Structural defects (lesions) develop in the anulus fibrosus with age. While degeneration has been simulated in numerous previous studies, the effects of structural lesions on disc mechanics are not well known. In this study, a finite element model (FEM) of the L4/5 intervertebral disc was developed in order to study the effects of anular lesions and loss of hydrostatic pressure in the nucleus pulposus on the disc mechanics. Models were developed to simulate both healthy and degenerate discs. Degeneration was simulated with either rim, radial or circumferential anular lesions and by equating nucleus pressure to zero. The anulus fibrosus ground substance was represented as a nonlinear incompressible material using a second-order polynomial, hyperelastic strain energy equation. Hyperelastic material parameters were derived from experimentation on sheep discs. Endplates were assumed to be rigid, and annulus lamellae were assumed to be vertical in the unloaded state. Loading conditions corresponding to physiological ranges of rotational motion were applied to the models and peak rotation moments compared between models. Loss of nucleus pulposus pressure had a much greater effect on the disc mechanics than the presence of anular lesions. This indicated that the development of anular lesions alone (prior to degeneration of the nucleus) has minimal effect on disc mechanics, but that disc stiffness is significantly reduced by the loss of hydrostatic pressure in the nucleus. With the degeneration of the nucleus, the outer innervated anulus or surrounding osteo-ligamentous anatomy may therefore experience increased strains.     

Prediction of the structural response of the femoral shaft under dynamic loading using subject-specific finite element models

The goal of this study was to predict the structural response of the femoral shaft under dynamic loading conditions using subject-specific finite element (SS-FE) models and to evaluate the prediction accuracy of the models in relation to the model complexity. In total, SS-FE models of 31 femur specimens were developed. Using those models, dynamic three-point bending and combined loading tests (bending with four different levels of axial compression) of bare femurs were simulated, and the prediction capabilities of five different levels of model complexity were evaluated based on the impact force time histories: baseline, mass-based scaled, structure-based scaled, geometric SS-FE, and heterogenized SS-FE models. Among the five levels of model complexity, the geometric SS-FE and the heterogenized SS-FE models showed statistically significant improvement on response prediction capability compared to the other model formulations whereas the difference between two SS-FE models was negligible. This result indicated the geometric SS-FE models, containing detailed geometric information from CT images with homogeneous linear isotropic elastic material properties, would be an optimal model complexity for prediction of structural response of the femoral shafts under the dynamic loading conditions. The average and the standard deviation of the RMS errors of the geometric SS-FE models for all the 31 cases was 0.46 kN and 0.66 kN, respectively. This study highlights the contribution of geometric variability on the structural response variation of the femoral shafts subjected to dynamic loading condition and the potential of geometric SS-FE models to capture the structural response variation of the femoral shafts.     

A simple model for the function of proteoglycans and collagen in the response to compression of the intervertebral disc.

The nucleus pulposus of the intervertebral disc exerts a pressure which enables it to support axial compression when contained by the annulus fibrosus. The disc was modelled as a thick-walled cylindrical pressure vessel in which the nucleus was contained radially by the annulus. As a result, the stress in the annulus had radial (compressive) as well as tangential (tensile) components. The radial stress at a given point in the annulus was considered to be balanced by the internal pressure which is expected to arise from the attraction of water by proteoglycans. There was a reasonable agreement between the calculated radial stress distribution and published results on the distribution of water within the annulus. As the internal pressure is expected to be isotropic, the annulus was expected to contribute to the axial resistance to compression of the disc; this contribution would be equal, in magnitude, to the radial stress. Predictions of the pressure distribution within the annulus were similar to published experimental measurements made in the radial and axial directions. The tangential stress within the annulus was considered to arise from the restoring stress in its strained collagen fibrils.     

Material constants for a finite element model of the intervertebral disk with a fiber composite annulus

A simple axisymmetric finite element model of a human spine segment containing two adjacent vertebrae and the intervening intervertebral disk was constructed. The model incorporated four substructures: one to represent each of the vertebral bodies, the annulus fibrosus, and the nucleus pulposus. A semi-analytic technique was used to maintain the computational economies of a two-dimensional analysis when nonaxisymmetric loads were imposed on the model. The annulus material was represented as a layered fiber-reinforced composite. This paper describes the selection of material constants to represent the anisotropic layers of the annulus. It shows that a single set of material constants can be chosen so that model predictions of gross disk behavior under compression, torsion, shear, and moment loading are in reasonable agreement with the mean and range of experimentally measured disk behaviors. It also examines the effects of varying annular material properties.