Professional Documents
Culture Documents
A Review of Validation Approaches and Modelling Challenges
A Review of Validation Approaches and Modelling Challenges
a r t i c l e i n f o a b s t r a c t
Article history: The knee joint is a complex mechanical system, and computational modelling can provide vital infor-
Received 25 October 2018 mation for the prediction of disease progression and of the potential for therapeutic interventions. This
Revised 5 August 2019
review provides an overview of the challenges involved in developing finite element models of the
Accepted 21 August 2019
tibiofemoral joint, including the representation of appropriate geometry and material properties, loads
and motions, and establishing pertinent outputs. The importance of validation for computational mod-
Keywords: els in biomechanics has been highlighted by a number of papers, and finite element models of the
Finite element tibiofemoral joint are a particular area in which validation can be challenging, due to the complex na-
Computational model ture of the knee joint, its geometry and its constituent tissue properties. A variety of study designs have
Knee
emerged to tackle these challenges, and these can be categorised into several different types. The role of
Tibiofemoral
validation, and the strategies adopted by these different study types, are discussed. Models representing
Osteoarthritis
Meniscus trends and sensitivities often utilise generic representations of the knee and provide conclusions with
Cartilage relevance to general populations, usually without explicit validation. Models representing in vitro speci-
Validation mens or in vivo subjects can, to varying extents, be more explicitly validated, and their conclusions are
Ligaments more subject-specific. The potential for these approaches to examine the effects of patient variation is
Review explored, which could lead to future applications in defining how treatments may be stratified for sub-
groups of patients.
© 2019 The Authors. Published by Elsevier Ltd on behalf of IPEM.
This is an open access article under the CC BY license. (http://creativecommons.org/licenses/by/4.0/)
https://doi.org/10.1016/j.medengphy.2019.08.002
1350-4533/© 2019 The Authors. Published by Elsevier Ltd on behalf of IPEM. This is an open access article under the CC BY license.
(http://creativecommons.org/licenses/by/4.0/)
2 R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12
particularly in the context of disease scenarios where interventions (1) The capture and representation of appropriate geometry and
are becoming increasingly common. These investigations may oc- material properties.
cur on highly subject-specific models or more generic representa- (2) The representation of appropriate motions, loads and
tions of the knee. constraints.
Kazemi et al. [6] wrote an extensive review on advances in (3) The establishment of relevant outputs and their levels of
computational mechanics in the human knee in 2013, and a more uncertainty.
general review on knee biomechanics was provided by Madeti et
al. [14] in 2015, but there has since been significant further work 2.1. Inclusion of geometry and material properties
produced, especially in the area of model validation. The purpose
of the present review is to provide an overview of the main pro- 2.1.1. Geometry
cesses and current challenges in knee modelling, and then to focus The level of detail required for the representation of the struc-
on examining validation strategies and the circumstances in which tures in the knee depends on the particular application of the
validation can be omitted. Three particular categories of study are model. For example, in a study focussed primarily on cartilage
identified: those representing trends and sensitivities, often using response, it may be appropriate to use basic spring elements to
generic models; and two types of more subject-specific models, represent ligaments, or even omit them completely for computa-
representing in vitro specimens and representing in vivo subjects. tional efficiency in which case their effect on the primary tissue of
This review is not intended to provide a comprehensive list of interest should be taken into account using kinematic constraints
papers utilising FE models to investigate the knee, but to focus [7]. Thus there exists variation among models in the manner
on the key challenges and the state of the art for validation when of implementation and detail in the representation of different
these different study types are utilised. This includes highlight- structures in the knee. There are several options for including
ing the importance of model reuse, verification, calibration and constituent tissues (e.g. ligaments, meniscus), in knee models:
context of use, as well as discussing good practices and potential
• The tissues can be explicitly modelled, and output measures
areas for future development.
taken (with detail driven by output precision/accuracy).
• The tissues can be explicitly modelled, but perform a support-
2. Processes and challenges for knee models
ing role in the model (with detail driven by precision/accuracy
of their effect on the outputs of interest).
The knee is a highly complex physical system and comprehen-
• The effect of the tissues can be wrapped into boundary condi-
sive models remain elusive due to the sparsity of precise data on
tions and loads, with no geometry included (applies primarily
knee tissue properties and limited understanding of the interac-
to ligaments).
tions between them, as well as how these factors vary among dif-
ferent subjects. Therefore models of the knee may be generated by For tissues that are included with explicit geometric repre-
considering a subset of the system, pragmatically chosen based on sentation, geometry is usually incorporated by utilising medical
a focused question that the model is designed to address. The fo- imaging (CT or MRI) of cadaveric specimens, volunteers or patients
cus of studies modelling the tibiofemoral joint in particular is often [15–17]. Dependent on image resolution, this can provide an
on the behaviour of the cartilage, meniscus or ligaments (Fig. 1). approximation of the native joint geometry, although it does not
In addition to validation, which will be covered in detail in provide a true representation since there will be errors due to
Section 3, there are several key challenges to address in order image resolution, imaging artefacts and simplifications inherent
to develop computational models of the tibiofemoral joint. These in the segmentation process [18], as well as smoothing applied
include: to specimen-specific models to ensure robustness of contact algo-
rithm solutions [19]. In some cases, multiple users may perform
segmentation to minimise variability [15], with variation between
paired images required to be below a specified threshold [20].
Pena and colleagues provided some earlier instances of models
using CT and MR imaging to include all of the main structures
of the knee, producing models featuring cartilage layers, menisci
and ligaments, as well as rigid bone representations [21,22]. These
models were used to analyse the effects of meniscectomy [22], and
later to investigate the combined role of menisci and ligaments in
load transmission and knee joint stability [21]. Although only basic
validation was provided in terms of comparisons to the kinematics
and stresses reported by other studies using different subjects,
and an idealised model was necessary to test mesh convergence,
these studies nevertheless demonstrated the potential for subject-
specific FE models to predict complex stress and strain patterns
and kinematics occurring in knee joints.
As an alternative to segmentation-based models, the geometry
for knee models can also be described mathematically [23–25],
reducing computation and analysis time [26–28]. It has been
shown recently that trends predicted by idealised parametric
models of joints based on mathematical geometric descriptors can
be similar to those seen in models based on image segmentation,
in both the knee and hip joints [10,29,30]. Thus simplified models
can provide reliable qualitative predictions of expected trends,
with particular potential to identify the aspects on which to
Fig. 1. Diagram of some of the typical components commonly featured in compu- focus in more sophisticated models. However, quantitative data
tational models of the tibiofemoral joint. from idealised models may not match well with experimental
R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12 3
predictions [10,31]. When a fully parametric geometric approach modifying the mechanical environment. Both computationally and
is taken, levels of geometric complexity can be added depending experimentally, motions may be applied directly or they may re-
on the intended application. For example, the menisci play an sult from applied loads when translational or rotational freedoms
important role in knee stability and they are essential for both are applied to the femur or tibia. The application of only loads
load transmission and joint lubrication [32]. Meniscus injury is without any constraints on motion allows the model the most
associated with osteoarthritis [33], and meniscal pathology pro- freedom but could result in physiologically inaccurate movements.
vides a key biomarker for osteoarthritis progression. Modelling the Different modelling studies therefore approximate knee function
behaviour of the menisci can therefore potentially provide new in different ways. Loads may be controlled and adjusted for sen-
insight into disease progression and a parametric approach can be sitivity purposes [11], or specific loads and boundary conditions
taken to investigate the effects of different meniscal geometries may be chosen to ensure motions replicate experiments [47].
[11]. On the other hand, a study focused on forces occurring in Specific in vivo motions can be difficult to derive; Stentz et al.
osteochondral grafts during cartilage-on-cartilage contact may [48] reported using CT bone models combined with a dynamic ra-
include a more basic meniscal representation [8]. diostereometric analysis system to achieve non-invasive measure-
ments of joint kinematics. This approach to measuring knee move-
2.1.2. Material properties ment was validated against a gold standard skin marker method,
In terms of material properties, a large amount is known and has potential clinical applications to prosthesis migration.
about the internal structure and properties of the bone, articular Deriving in vivo knee forces for use in quasi-static FE models may
cartilage, meniscus, and ligaments within the knee. Equally, there also be achieved using multibody models [49,50]. Fully dynamic
exist theoretical models and numerical techniques that include FE models may be necessary if the effects of inertia were thought
collagen fibre alignment, hyperelastic behaviour, fluid contribution to be crucial, for example in a study of knee replacements [49].
and multi-layered aspects of the knee as a mechanical system Ligaments play a large role in both knee kinematics and
[12,15,34,35]. However, obtaining sufficiently detailed experimental biomechanical load bearing [51] and their injury is associated
data to calculate the many property values required for such with increases in pain, osteoarthritis and knee joint instability
sophisticated representations is very challenging. Thus the mate- [33,52]. Ligaments are therefore a key factor which influence the
rial property models within computational knee simulations are relationship between applied loads (or motions) and resulting
commonly set up based on a sensible choice of physics informed motions (or loads) in the knee. One of the principal clinical drivers
by literature and not explicitly validated. There is great variability for the detailed modelling of ligaments in the knee is the analysis
in the properties used for each tissue, which may be taken from of their repair after injury, particularly the anterior and posterior
existing literature or be derived experimentally. Imaging can also cruciate ligaments (ACL and PCL), and FE models are increasingly
be used to obtain subject-specific material properties; for example being utilised to investigate ligament rupture and reconstruction
location-specific bone density is commonly derived using CT imag- [53–55]. Ali et al. [16] demonstrated that ACL resection can pro-
ing [36–38], and one study [39] used sodium MRI to determine duce altered knee mechanics and motion by testing cadaveric knee
fixed charge density distributions in the tibial cartilage, although specimens in an electro-hydraulic knee simulator with motor-
this technology is limited by imaging resolution. actuated quadriceps and loads applied at the hip and ankle, in
Bone is often assumed to be rigid in knee models for compar- each case first with the ligament intact and then resected. FE mod-
ative studies where loading effects on cartilaginous soft tissues or els developed to simulate these scenarios revealed that changes
ligaments are of particular interest [10,16]. A more complex bone resulting from ACL resection can manifest differently among dif-
representation may be important for making subject-specific pre- ferent specimens; one specimen exhibited altered anterior tibial
dictions of regions at risk of joint failure for particular specimens, translation, whilst the other exhibited elevated joint loads. Since
as bone stiffness can affect tibial cartilage stresses [36]. individual differences exhibited most clearly when calibrated
For representing cartilage, a linear elastic material model is ligament properties were used, this suggests a subject-specific
commonly used, due to the equivalence between short-time bipha- ligament modelling approach would be beneficial for a larger
sic and incompressible elastic material responses demonstrated study. This is supported by findings of Beidokhti et al. [15], who
by Ateshian et al. [40]. Depth-dependent material properties, found that including subject-specific derived ligament properties
inhomogeneity of the cartilage and the biphasic response may in continuum modelled ligaments improved predictions of exper-
also be relevant if the intended application is to better understand imental kinematics and contact pressures. Earlier models [19] also
longer-term cartilage mechanics [12,34,41]. Osteochondral defects, found that tuning ligament properties so that model kinematics
consisting of damage to both articular cartilage and the underlying matched those found in a cadaveric specimen aided the validation
subchondral bone, present a large clinical burden by altering of model derived joint contact forces. More recently, it has been
the local biomechanics and biotribology of the knee joint and suggested that additional peripheral soft tissues including knee
causing joint pain [2]. Whilst imaging approaches can be used to capsules as well as ligaments may alter predicted knee mechanics
assess cartilage deformation [42], such methods do not provide [56].
an effective means to analyse the contact force distribution on Another major challenge related to modelling knee motion is
the articulating surfaces. Inclusion of detailed cartilage layers understanding the role of the meniscus and analysing meniscal
within FE models of the knee is therefore crucial for progressing movement in response to loading. This may be crucial for mod-
understanding of their degeneration. This may include multiscale elling the potential for damage progression in the knee. Meniscal
modelling of cartilage [43–45]. Freutel et al. [46] discussed the translation has previously been captured using MR imaging [57],
challenges of material models for soft tissues in greater detail, so and Halonen et al. [47] created a subject-specific FE model using
this will not be covered further here. MR images of a volunteer’s knee specifically to investigate menis-
cus movements and cartilage strains. One particular issue with ac-
2.2. Inclusion of motions and loads curately modelling the meniscus is incorporating meniscal attach-
ments. This aspect can be difficult to accurately capture within
This section is concerned with the challenge of ensuring knee models due to the challenge of achieving precise segmentation
models produce motions corresponding to an experimental sit- of attachment site geometry and establishing material models for
uation of interest. This requires an understanding of the moving their behaviour. The attachment sites can be particularly difficult
parts within the knee and how they react to different motions to identify in imaging, especially without prior ligament removal if
4 R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12
using a cadaveric specimen, and may be virtually impossible to dis- 2.4. Model sharing
tinguish from other soft tissues if using CT imaging. Thus generic
spring elements with estimated properties and locations have been As the prominence of computational modelling in biomechanics
used to represent meniscal attachments, which along with friction increases, uncertainty about modelling results can be decreased
serve to limit meniscal movement. Freutel et al. [58] segmented through increased sharing of models and data [64,65]. It can be a
medial meniscus geometry from MR images of porcine knee joints, challenge to fully describe the methods used in FE models of the
with meniscal displacements having previously been determined knee within research papers, but this is important for understand-
experimentally. From this data, optimisation was used to deter- ing simplifications that may affect the results. Sharing models and
mine subject-specific material properties of the meniscus and its data of sensitivity studies in particular can help clarify the effects
attachments, allowing the time-dependent behaviour of the menis- of different levels of input precision on likely model outcomes.
cus and its attachments to be investigated. Sharing of scripts and protocols in addition to data can help mit-
igate potential issues with software and version compatibility and
ensure repeatability. In addition to the OAI project [63] mentioned
2.3. Input precision and establishing relevant outputs previously, other researchers are beginning to make knee models
freely available online [16,66,67]. This can have significant impact;
There are many different motivations for developing computa- models made available through the Open Knee Project [66] have
tional models of the knee, producing many distinctive approaches supported many new publications, including [10–12,68]. Sharing
to doing so. It is therefore important to consider the relevance of of models also provides the means for improved understanding
model outputs assessed to the original aim of the study. In partic- of what aspects of the model were validated. Transparently re-
ular, when the ultimate aim is to use the models to predict disease porting numerical quantification of validation evidence is also
risk or assess the suitability of treatments in vivo, it is necessary essential, because it is plausible that a given set of experimental
to consider the clinical relevance of outputs reported [59]. This and computational results would be described as similar by one
may include outputs to indicate risk of damage progression or to group where another would conclude that the model has failed to
assess intervention suitability. For example, meniscus movement precisely replicate the experiment. Knee model validation is the
might be a crucial metric in a study of the progression of meniscal focus of the next section.
tears [58], whilst in a study of femoral osteochondral grafts it
may be pertinent to analyse tibial cartilage contact patterns to 3. Knee model validation and calibration strategies
understand the effects of graft recession or extrusion [8,60].
In the case of FE models used to examine mechanical re- 3.1. Validation, verification and calibration
sponse in the knee, model outputs can be highly sensitive to the
chosen representation and condition of included tissues such as Having considered some of the ways in which methodological
the menisci and cartilage. Ambiguity in input values can result challenges in knee modelling are being addressed, the challenge
in a wide range of reported values for outputs of interest. One of providing validation for computational models of the knee is
study [61] found that output uncertainty can be reduced when now examined. To validate a model is to provide evidence that
specimen-specific data for certain input parameters is known, model generated results correspond to the outcomes of the real
including joint geometry such as meniscal insertion site positions, world scenario simulated [69]. Several guidelines exist with con-
kinematics and BMI to inform loading. Thus some uncertainty can siderations for reporting FE validation studies in biomechanics and
be reduced for specimen-specific models, although many inputs including sufficient detail for repeatability [70–72]. In particular,
may be difficult or impossible to obtain clinically. Furthermore, in Pathmanathan et al. [73] recently proposed a framework for the
certain situations some parameters may be impossible to control applicability analysis of validation evidence in computational mod-
experimentally, and in this case they may be used as tuning els for biomedical applications, and this provides a resource for
parameters for each knee-specific model. For example, an earlier evaluating validation quality. The framework recommends the
study by the same group [62] used the varus-valgus angle as systematic assessment of the relevance of the validation evidence
a tuning parameter and found similar regions of contact stress for the proposed context of use, which encompasses the purpose
between models and experimental work (in this case quantified of the model and what factors its results are used to inform.
by normalised cross correlation values within 69 to 85%). For the purposes of this paper, direct validation is used to refer
Even when the uncertainty for output measures is minimised, to situations in which a comparison between a model prediction
it remains crucial to establish exactly which outputs are of in- and an experimental test result is made after a model has been de-
terest for the particular focus of the model so that they can be veloped to match the corresponding experiment as closely as pos-
used to predict intervention response or disease progression. sible [69]. Indirect validation is used when the model prediction is
The Osteoarthritis Initiative (OAI) [63] provides a database on the compared to a physical case where it not known whether the con-
natural history of osteoarthritis by making publically available ditions are the same. Confidence can also be built by performing
clinical evaluation data and imaging (X-ray and MRI) for nearly several related validation checks, for example by comparing:
50 0 0 subjects. One study [7] considering subjects from the OAI • Several different outputs (e.g. displacement, stress)
database defined outputs specific to disease progression by split- • Under a variety of conditions (e.g. loading cases, restraint cases)
ting subjects into two groups based on osteoarthritis risk and BMI. • Against data sets from a variety of sources (e.g. multiple
FE models of one representative subject from each group were
specimens)
generated and collagen fibril damage was defined to occur when
tensile stresses exceeded a threshold limit during gait loading, Prior to validation of a model, it is necessary to assess the
with control of degeneration based on the duration of loading in ability of the model to provide accurate numerical approximations
different regions over successive iterations. In this way an algo- to its underlying equations, including the testing of mesh conver-
rithm was presented to predict knee cartilage degeneration based gence. This is known as model verification, which has been widely
on accumulated excessive stresses in the medial tibiofemoral discussed [69,70] and will not be covered in detail here. However
compartment. Approaches like this may become more common it is important to highlight that solving computational contact
as modelling complexity increases, allowing outputs relevant to problems in biomechanics generally remains very challenging,
specific scenarios to be analysed. and image processing and smoothing steps may be necessary to
R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12 5
Table 1
Summary of the main types of model used in finite element studies of the knee. Here subject-specificity is used to refer to the geometric representation employed in the
model, but calibration of material properties may mean that the materials could also be described as subject-specific.
Representing trends and Generic or subject-specific Not included May report literature Independent demonstration of the effect of
sensitivities comparisons inputs and interactions within a complex
physical system. May also be used
complementary to experiments to reveal
internal stresses and strains or provide
sensitivity data.
Representing in vitro subjects Subject-specific Frequently included Matching experiment Testing a device, procedure or disease in the
context of particular in vitro specimens.
Representing in vivo subjects Subject-specific Challenging to include Matching in vivo data Testing a device, procedure or disease in in
the context of particular in vivo subjects.
achieve model convergence. Contact algorithms in finite element 3.2. Approaches to validation in different study designs
codes may not be sufficiently robust to handle meshes on complex
specimen-specific geometries such as those found in the knee, Validation is often more challenging when models are devel-
with small changes to geometry resulting in models that do not oped to represent in vivo subjects. On the other hand, generic
converge. Through the FEBio project [74], open-source code has models aiming for broad conclusions may not require detailed
been specifically designed for such biomechanical applications and validation strategies. Calibration and validation strategies used in
is addressing some of these challenges. different knee modelling studies therefore vary according to the
One aspect of verification for which the modeller is responsible study purpose and the types of model utilised; this is summarised
is demonstrating the suitability of the chosen mesh. Hexahedral in Table 1.
elements are generally preferred for modelling contact, but present Further discussion of each of these identified approaches is
a particular challenge when meshing complex geometries such as provided in subsequent parts of this section. Some of the key
femoral cartilage and the menisci. Quadratic tetrahedral elements studies discussed are outlined in Table 2 to demonstrate examples
provide a possible alternative to alleviate this issue; they are more of each study type from the literature.
straightforward to implement and have previously been seen to
perform well in models of foot biomechanics [75] and articular 3.3. Models representing trends and sensitivities
contact in the hip [76,77]. Recently some authors have also used
quadratic tetrahedral elements in modelling knee joints [15]. Computational models can be used for investigating features of
Computational models can be developed contemporaneously the knee as a complex physical system, with the aim of evaluating
with, and validated against, in vitro or in vivo experiments. This sensitivity of general outcomes to input parameters. In these
may require some calibration of model parameters so that model scenarios, models may incorporate generic or previously measured
results align with experimental results [78]. Calibration involves inputs, which have inherent uncertainty associated, into an as-
tuning input parameters based on model results. If these tuned sumed physics framework. Generic representations of the knee
parameters are not specific to each specimen, this generally are particularly well suited to parametric testing to demonstrate
means minimising the model-experiment error across a set of trends and highlight which uncertainties are most critical. In this
specimens. A gold standard for validation is thus to test whether case the conclusions are population-based rather than specimen-
model results continue to correspond well with experimental data or group-specific.
when independent specimens are tested. However, because several The Open Knee model [66] is commonly used for investigating
factors affect their outputs, it is important to avoid erroneously generic trends in knee mechanics. For example, its geometry has
concluding a model is validated based on its calibration. At the been used to investigate the time dependent behaviour of cartilage
study design phase, researchers should carefully consider what and fluid pressure at the cartilage-meniscus interface [12], and for
their models aim to elucidate and plan validation steps accord- investigating the effects of meniscal tears and full meniscectomy
ingly. Several different combinations of model parameters may [11]. In studies like these, there is generally no direct validation
lead to similar results, and consequently it may be possible to included due to the lack of experimental counterpart for the de-
erroneously ‘validate’ a model by chance. Parameters that initially veloped models, but the results are often compared with literature
appear unimportant could cause crucial differences in model out- findings. In one study [12] results were compared with other pub-
put following the addition of further parameters. For example, in a lished models and experimental predictions of outputs including
model of knee contact mechanics, calibrating the meniscus prop- contact areas and femoral displacements under static loads to
erties may produce the cartilage contact pressures that align well provide confidence in the modelling approach. In another study
with experimental data, but the meniscus properties themselves [11] the authors explicitly stated that physical validation could not
may actually be incorrect. These incorrect properties, coupled with be included as an abnormally flat meniscus geometry was used to
incorrect properties of meniscal attachments and the coefficient of investigate meniscal extrusion in the presence of meniscal tears.
friction between the meniscus and cartilage layers could lead to Their findings are likely relevant to certain real cases with similar
inaccurate conclusions about meniscal pathology even if cartilage morphological characteristics, and would be more difficult to
pressures were observed to be correct. Experimental data also has achieve using a complex specimen-specific model, where conver-
limitations (for example in resolution and accuracy of sensors and gence issues may arise due to more complex articular geometry.
detection of environmental noise), so researchers should take into Generic or specimen-specific models can also be developed
consideration that model outputs may need to be compared to alongside experimental tests to generate additional information
suboptimal experimental data. There is substantial literature on regarding internal stresses or strains, or to provide sensitivity data
sensitivity testing in knee models [10,61] and researchers should allowing for fewer test runs. In these cases the response of the
be encouraged to report these findings to provide the community tibiofemoral joint to different loading scenarios can be investigated
with an improved basis for output interpretation and understand- without requiring detailed direct validation to support the study
ing of the circumstances in which conclusions remain valid. conclusions. For example, one study [60] used a model based on
6 R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12
Table 2
Applications of a selection of recent finite element modelling studies of the human tibiofemoral joint, arranged into the three identified study types to highlight the validation
strategies used. Model parameters such as material properties are commonly taken from literature, so instances of calibration are also highlighted.
a single subject to parametrically investigate different alignments, recommendations, but rather to contribute to the body of evidence
geometries and properties for osteochondral grafts. This study on the factors which are important in cartilage behaviour. Thus
mentioned basic validation in terms of comparisons to experi- models from studies such as these fit into the category of models
mental studies of different cadaveric specimens, and whilst this used to better understand the trends and sensitivities in the knee
means the results could not provide subject-specific information as a mechanical system.
on approaches to alleviate chondral lesions, the data do provide
guidance on the types of specimen on which to focus in future
studies using more complex models. Thus the specimen-specific 3.4. Balancing output relevance and validation in specimen-specific
geometry here was largely incidental, and the study findings sug- models
gest generic trends, such as indicating proud placement of grafts in
particular increase the stress they experience. These trends can be Whilst generic models are well suited to parametric testing
used to inform future studies and provide useful sensitivity data. to establish trends, the development of models with greater
When highly complex material representations are used in subject-specific detail can provide biomechanical insights allowing
knee models, particular outputs from specimen-specific cases for subject-specific predictions. This is important since variations
may be impossible to directly validate. Gu et al. [82] generated in anatomy and tissue properties between patients may lead to
a model of a single volunteer to investigate effects of collagen differences in treatment outcomes. Furthermore, specimen-specific
fibres on fluid pressurisation in cartilage, and their findings models provide the means to set up one-to-one matches with
were not directly validated since collagen fibre directions were experiments to allow for direct validation. Geometrically specimen-
assumed to follow a generic split-line direction and zonal differ- specific models may also include generic or previously measured
ence was not considered. Similarly, Mononen et al. [41] used a inputs and the associated uncertainty this brings, but incorporate
poro-viscoelastic model based on a single subject to examine the some subject-specific inputs relevant to the scenario they are
importance of collagen fibril organisation for the optimal function used to examine. Such complex, specimen-specific knee models
of articular cartilage, which again could not be directly validated. are becoming increasingly prevalent [15,16], but remain in their
Although subject-specific geometries from in-vivo subjects were infancy, warranting a more in depth discussion regarding their
utilised for these studies, they were not seeking to make clinical validation and extracting useful information from their outputs.
R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12 7
Fig. 2. There are multiple possible sources of error when comparing computational models of the knee against experimental tests in order to assess their capability to
predict real world scenarios. Potential areas for alignment or discrepancy are displayed within dashed boxes. The double lined arrows indicate that there must be a trade-off
between validation of computational models against experimental data with controlled conditions, whilst replicating uncertain real world data as closely as possible.
Whether simulating in vitro or in vivo subjects, a stronger re- validation study and later use a modified version of the model
lationship between a model and an experiment can often only be for a different purpose [73]. In these cases, it is important to
achieved at the expense of the relevance of the experimental data understand whether the validation evidence supports the trust-
to the real world scenario, such as the progression of osteoarthritis worthiness of the model’s predictions in the new context. The
under different mechanical conditions. Here there must be a risks of uncertainty over model trustworthiness depend on the
trade-off; experiments can be closely aligned with a model and model’s application and its decision consequence. In the future
potentially lose certainty about the relevance of motions and loads for example, validation may be performed on software intended
to in vivo scenarios, or the model could be compared directly to to provide medical advice, in which case the outcome of erro-
an in vivo scenario, potentially losing the capacity for calibration neous validation is likely to be severe. More typically in current
of specimen-specific properties (Fig. 2). In both cases, the outputs knee modelling, validation is performed to provide credibility to
that can be practically compared in validation tests may be proxy simulation results as scientific evidence.
measures for the true output(s) of interest. For example, elevated
contact pressures may be used to predict regions with increased
risk of osteoarthritis in knee models, as this can be measured both 3.5. Models representing in vitro subjects
experimentally and computationally, but a direct observation of
cartilage quality might be preferred in a cadaveric joint. In the Subject-specific inputs for models representing in vitro subjects
case of models based on a cadaveric specimen, it is possible to may be derived from imaging of the in vitro specimens or through
obtain specimen-specific measures of some material properties calibration with experimental measures on these specimens.
(for example ligament parameters can be derived through laxity Calibration may be repeated for the model of each subject to
testing [15]), and experiments can be set up in controlled condi- yield a set of validated subject-specific models (models where the
tions that can be replicated in FE models using matching boundary response is known to be correct for each subject), or calibrated
conditions [15,16]. However it is difficult to be certain that applied average properties may be used with potential for testing across a
loads match the conditions that the specimen experienced in population subset.
vivo, and removal of supporting tissues such as knee ligaments Piezoresistive thin-film sensors are commonly used for mea-
during experimental set up may alter constraints experienced by suring experimental contact mechanics in experiments using
cartilaginous soft tissues of interest. In the case of a model based cadaveric specimens [19]. In one of the earlier demonstrations of
on an in vivo subject, it is possible to set up in vivo experiments this technique, it was shown that a transversely isotropic linearly
to capture cartilage deformation under in vivo conditions, for elastic material model for meniscal tissue could provide good
example by using an MR imaging compatible loading device [83]. computational estimations of experimental measurements of con-
However, non-invasively deriving specimen-specific material prop- tact pressure, with discrepancies between mean contact pressures
erties to include in the computational model may not be possible, in the region of 10-15% [84]. Fundamental validation studies like
so literature-based material properties are often required [39]. this can provide confidence when models are later used for more
Regardless of the validation approach taken, it is important specific scenarios, in this case the investigation of meniscectomy
to consider that there are always a number of different aspects under axial loading [85]. However, since physical experiments
which can influence how effectively experimental and computa- contain inherent uncertainty, an important part of performing
tional data can be compared, as well as the relevance of the data effective validation of in vitro models is judging when to stop
to the real life scenario of interest. Potential sources of error are calibrating to exactly match laboratory test results. In particular,
illustrated in Fig. 2. thin-film sensors have some limitations; measurements of contact
Whilst validation evidence for knee models should ideally be pressure and area are subject to experimental errors in detection
obtained from an experimental scenario as relevant as possible to and signal processing, particularly on highly curved or uneven sur-
its proposed context of use, it is not unusual to report an initial faces, and they cannot record shear components of stress [15,86].
8 R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12
Furthermore, the insertion of sensors may lead alteration of the by ligament laxity tests to allow calibration of tibiofemoral soft
surrounding tissue, interfering with the natural joint mechanics. tissue material properties and attachment locations for models of
When describing subject-specific knee models as validated, cadaveric human knees [80]. Following this, subject-specific knee
it is important to be clear about the scope of the model. One mechanics were modelled in silico for intact and ACL-deficient
study [8] describes a knee model as validated, but this was based conditions under several loads to simulate experimentally mod-
primarily on literature comparisons rather than direct validation, elled dynamic activity [16]. This complex study design with
and the model used in the study was a development of a much several stages resulted in models where validation evidence was
earlier knee model [87] with the mesh modified to refine the provided under both healthy and pathological conditions through
cartilage and menisci. Although the conclusions were within a comparison of experiments and models. Validation data were
scope that could be reasonably determined using more generic generated in terms of both kinematics in the tibiofemoral and
models (indicating that bone damage and cartilage splits can alter patellofemoral joints and in terms of forces experienced by the
the magnitude and pattern of cartilage pressure and strain), to knee. Model predicted kinematics were seen to largely agree
make more specific conclusions would require further supporting with experimental data in both trends and magnitudes, assessed
validation work. Contrastingly, another study, which assessed rela- using root mean squared differences, and as expected, the largest
tionships between contact pressure and osteochondral defect size differences occurred at flexion angles greater than those for which
[38] was clear about the extent of validation achieved. Although a ligament laxity calibration had been performed. Furthermore,
non-specimen-specific linearly elastic material representation was quadriceps forces seen in the models were comparable to quadri-
used for the cartilage, predictions of peak pressure in the lateral ceps forces seen following loading in the experimental simulator
and medial condyles closely matched results from cadaveric exper- tests, although these forces actually changed negligibly following
imental models for defect sizes up to 20 mm, with the exception ACL resection, possibly missing potential effects of any adaptive
of the smallest size tested (5 mm). Thus a range was provided for behaviour that could manifest in vivo.
the predictive capabilities of the presented model, with poorer A key aspect of Ali et al.’s study [16] was the commendable use
agreement for smaller defects possibly due the mesh density. of validation under very different loading scenarios: knees with
Freutel et al. [58] were similarly clear about the circumstances in intact and resected ACLs. However, in this study the ligaments
which their model was valid. On the basis of low prediction errors were modelled as spring elements for calibration purposes, and
for experimental meniscal displacements, the authors analysed the meniscus was not included when these ligament calibrations
stress distributions in modelled porcine menisci, and stress mag- were performed [16]. Considering that Beidokhti et al. [15] found
nitudes were seen to correspond well with previous studies of the a continuum model for ligaments more closely reproduced exper-
meniscus [22]. This level of validation suggests the methods could imental results, a logical next step would be to use constitutive
be utilised for further investigations, for example into the effects material models for the ligaments. Furthermore, it would be
of meniscal tears, but in order to be applied to human specimens, beneficial to include the meniscus when these calibrations are
corresponding experimental data with human tissue would be carried out, since the meniscus may provide additional stability,
required. and damage from dissection to facilitate its removal could cause
Comparison of models against in vitro experiments can also altered responses. Since multiple experiments were run on the
be used as a method to illuminate the necessary complexity of same samples, it is also important to consider whether this may
a model in different testing scenarios. Beidokhti et al. [15] for induce tissue damage in the specimens beyond that which is
example demonstrated that models featuring subject-specific likely to occur in vivo. Further, it may be possible to consider any
material parameters for ligaments (derived from experimental detrimental effects of bone fixation processes and the order in
tensile tests) produced kinematics more aligned with experimental which the experiments were conducted. All these factors would
data than those with literature-based properties. Literature-based however need to be offset against increased costs in terms of
spring models for ligaments produced high errors in contact additional experimental and computational resources in studies
pressure but acceptable kinematics, suggesting scenarios where which are already complex and challenging.
this approach may be sufficient. As restricting knee movement is a
major function of ligaments, validating the kinematic output was 3.6. Models representing in vivo subjects
important for the application of these models. Precisely matching
contact pressures for this scenario was arguably less crucial, but Developing subject-specific models of in vivo subjects, as op-
this aspect was nevertheless also reported in detail in the paper. posed to in vitro cadaveric specimens, presents its own unique set
It is important to recall that experimental tests themselves of challenges. Obtaining data for calibration is likely to be much
may not precisely emulate the in vivo situation of interest, al- more challenging in an in vivo scenario, and similarly outputs that
though demonstration that a model can replicate a controlled can be practically compared for validation are more limited. For
environment is generally the first step in assessing its potential example, there are both ethical and practical limitations to the
to simulate more complex in vivo scenarios. Loads applied during direct measurement of joint contact forces in vivo, so it may be
the validation experiments in Beidokhti et al.’s study [15] were necessary to partially validate an in vivo model against experimen-
reduced due to structural limitations of the testing apparatus, and tal data of other specimens [81]. On the other hand, modelling in
were not intended to represent in vivo quadriceps loads. They vivo subjects allows patient-specific multibody dynamics models
were however selected based on the intended application of the to be developed based on in vivo movements to aid the applica-
models, the analysis of ACL reconstructions. Technical issues can tion of potentially more clinically relevant joint forces to an FE
also reduce the relevance of an in vitro scenario to an in vivo model [88].
situation; for one of the study’s specimens, the collateral ligaments One group approached the validation of an FE model based
required excision in order to permit sensor insertion. To mitigate on an in vivo subject by using an MR compatible compression
this issue, this condition was replicated in the corresponding device to load the knees of an asymptomatic subject [39]. The
model before contact pressure and area assessments. cartilage deformation magnitudes were found to correspond well
In highly complex models of in vitro specimens, several cal- at equivalent loads, but there was a discrepancy in load distri-
ibration stages are often necessary to align computational and bution between the medial and lateral plateaus, which might
experimental results and achieve validation. One group initially be partially explained by free varus-valgus rotation allowed in
performed calibration of patellofemoral mechanics [79], followed the model. This validation was sufficient to support comparative
R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12 9
studies into the effect of local variations of fixed charged densities teoarthritis, consistent with experimental data from obese subjects
in the cartilage [17,39]. However to derive conclusions specific to in the OAI database. These comparisons provided confidence in
the knee mechanics of the particular subject used for the model, the predicative capability of the models, but areas of uncertainty,
it would be beneficial to seek an improved match between the such as the threshold limit for determining degeneration initia-
computational and experimental conditions. tion, prevented more subject-specific conclusions. Future studies
Another group also developed models based on MR images of that employ a range of at-risk subjects and use subject-specific
healthy volunteers, initially observing the knee joint becoming rather than representative loading data, could be used to predict
stiffer as a result of elevated fluid pressure following meniscec- osteoarthritis onset in a range of knee subsets, suggesting in vivo
tomy [89]. Although there was no way to validate these findings, models have strong potential for future clinical applications.
the group provided discussion of validation for follow up models
used to investigate load transfer from the cartilage to the meniscus
4. Discussion
[35]. In this study, indirect validation was discussed on two levels.
Firstly, the material model used for cartilage was compared to
4.1. Summary
stress relaxation and creep data for bovine cartilage explants [90].
Secondly, results from the overall knee model were compared to
This paper has described some of the challenges in developing
older experimental data [91]. Whilst these steps do not constitute
computational models of the tibiofemoral joint, describing the
direct validation, they provide a means to evaluate whether the
difficulties in ascertaining representative geometries, material
model results seem sensible. The region of maximum contact
properties and loading conditions, as well as challenges in de-
pressure was observed in the medial compartment, consistent
ciding appropriate outputs to measure. Although the focus has
with previous data, although pressure values were relatively low.
been on the tibiofemoral joint, many of the concepts discussed
It is challenging to derive specimen-specific material properties
are also relevant to the modelling of other joints. In particular, the
in vivo, so bovine cartilage was used for material calibration,
sharing of computational models and data in biomechanics has
implying a general representation of cartilage was used. Thus
the potential to move the field forward more rapidly.
the discrepancy in pressures values may have been due to the
Thorough validation strategies remain an important aspect for
non-specimen-specific nature of the employed material model
inclusion in knee modelling studies. As has been indicated by
and potentially the shape of the specific knee used in the study.
others [70–73], a model can only be described as validated for the
Another study [47] also used material properties originally derived
scenarios and outputs that have been tested against corresponding
from bovine tissue in a knee model to assess cartilage strains
experimental data, and validated models can only provide strong
and meniscal motions. This model was able to capture patterns of
conclusions when varied within the scope of their original pur-
meniscal motion with only slight differences from those observed
pose. However, to maximise the returns of developing models,
in vivo using CT with contrast media. In this case, the non-specific
further scenarios or specimens are generally investigated. Thus it
material properties along with other uncertainties meant the
is crucial to consider whether validation evidence is appropriate
authors stated that they did not seek to calibrate the model for
for the context of use of a given model. Equally, if every aspect
further validation against the experimental data. However these
of every model requires validation, there is little purpose in
comparisons again provide a means to assess whether the results
producing models of the knee joint at all. Eventually, unvalidated
may be sensible without the authors claiming to have developed a
cases must be run in order to take advantage of the capability
fully validated model.
that computational models provide. In this case, researchers must
In order to derive conclusions relevant to larger populations,
decide what constitutes a reasonable step away from a validated
development of models with subject-specific geometry will be
case, where modelled outcomes remain trustworthy. In particular,
required for greater numbers of in vivo subjects. This could allow
it may be desirable to use models to bypass further experiments
predictions to be made on the efficacy of potential treatments
and predict in vivo knee biomechanics or kinematics. For example
for the knee in different patient subgroups. This will require the
Ali et al. [16] analysed additional measurements beyond those
development of models that can be validated for multiple uses
directly validated, such as ligament AP shear forces with respect
(such as under different loading scenarios) and confirmation that
to the tibia. In future, it would be useful for study authors to
model development procedures are reproducible. It would also
include their rationale for why additional outputs are thought to
be beneficial to understand where the greatest variation lies in
be appropriate. Parametric tests, such as material changes from
populations, in order to concentrate modelling efforts on particular
a validated baseline are also in a sense unvalidated, but initial
knee subgroups. A recent study [20] used measurements of the
validated cases provide a degree of confidence, for example in
menisci from both knees of subjects from the OAI database to
work by Räsänen et al. [17,39].
generate active appearance models. Meniscal damage locations
In some cases detailed validation may not be required at all;
were consistently seen primarily in the posterior medial region,
this can be true when generic knee models are generated with the
and meniscal thickness and tibial coverage were identified as risk
aim of investigating trends in the complex joint system without
factors for osteoarthritis progression, which may be an important
making more specific conclusions related to individual specimens
consideration for future modelling studies. Mononen et al. [7] also
or sub-populations. Computational modelling has been used to
used in vivo subjects from the OAI database, in this case seeking to
great effect to provide detailed information about the trends and
validate FE models based on these subjects. The complexity of the
uncertainties in the knee joint using generic and idealised models
model scenario, in which an algorithm was presented to predict
[10,26], although it is not yet clear in which circumstances unvali-
cartilage degeneration, meant that direct validation would be
dated cases can be trusted and it is useful to compare conclusions
very challenging and would need to occur over a prolonged time
of these studies with contemporary literature findings.
period, because a model would need to be developed to match
the experimentally observed conditions seen as osteoarthritis
progressed in a particular patient. Follow-up Kellgren–Lawrence 4.2. Outlook
grades on the OAI database subjects did however provide a basis
for comparison, and correlations were derived between model data The long term ambitious aim for the modelling community as a
and clinical observations. Progress of collagen fibril degeneration whole could be to continue to increase understanding of the knee
in the models was seen to occur mostly in the initial stages of os- joint using complex computational models. Checking surprising or
10 R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12
[11] Łuczkiewicz P, Daszkiewicz K, Witkowski W, Chróścielewski J, Zarzycki W. [36] Venäläinen MS, Mononen ME, Väänänen SP, Jurvelin JS, Töyräs J, Virén T, et al.
Influence of meniscus shape in the cross sectional plane on the knee Effect of bone inhomogeneity on tibiofemoral contact mechanics during phys-
contact mechanics. J Biomech 2015;48:1356–63. doi:10.1016/j.jbiomech.2015. iological loading. J Biomech 2016;49:1111–20. doi:10.1016/j.jbiomech.2016.02.
03.002. 033.
[12] Meng Q, Jin Z, Wilcox R, Fisher J. Computational investigation of the time- [37] McErlain DD, Milner JS, Ivanov TG, Jencikova-Celerin L, Pollmann SI,
dependent contact behaviour of the human tibiofemoral joint under body Holdsworth DW. Subchondral cysts create increased intra-osseous stress in
weight. Proc Inst Mech Eng Part H J Eng Med 2014;228:1193–207. doi:10.1177/ early knee OA: a finite element analysis using simulated lesions. Bone
0954411914559737. 2011;48:639–46. doi:10.1016/j.bone.2010.11.010.
[13] Farrokhi S, Keyak JH, Powers CM. Individuals with patellofemoral pain exhibit [38] Papaioannou G, Demetropoulos CK, King YH. Predicting the effects of knee fo-
greater patellofemoral joint stress: a finite element analysis study. Osteoarthr cal articular surface injury with a patient-specific finite element model. Knee
Cartil 2011;19:287–94. doi:10.1016/j.joca.2010.12.001. 2010;17:61–8. doi:10.1016/j.jbiomech.2009.07.022.
[14] Madeti BK, Chalamalasetti SR, Bolla Pragada SK Sundara siva rao. Biome- [39] Räsänen LP, Tanska P, Mononen ME, Lammentausta E, Zbýň Š, Venäläinen MS,
chanics of knee joint—a review. Front Mech Eng 2015;10:176–86. doi:10.1007/ et al. Spatial variation of fixed charge density in knee joint cartilage from
s11465-014-0306-x. sodium MRI–implication on knee joint mechanics under static loading. J
[15] Beidokhti HN, Janssen D, van de Groes S, Hazrati J, den Boogaard T, Verdon- Biomech 2016;49:3387–96. doi:10.1016/j.jbiomech.2016.09.011.
schot N. The influence of ligament modelling strategies on the predictive capa- [40] Ateshian GA, Ellis BJ, Weiss JA. Equivalence between short-time biphasic and
bility of finite element models of the human knee joint. J Biomech 2017;65:1– incompressible elastic material responses. J Biomech Eng 2007;129:405–12.
11. doi:10.1016/j.jbiomech.2017.08.030. doi:10.1115/1.2720918.
[16] Ali AA, Harris MD, Shalhoub S, Maletsky LP, Rullkoetter PJ, Shelburne KB. Com- [41] Mononen ME, Mikkola MT, Julkunen P, Ojala R, Nieminen MT, Jurvelin JS,
bined measurement and modeling of specimen-specific knee mechanics for et al. Effect of superficial collagen patterns and fibrillation of femoral articu-
healthy and ACL-deficient conditions. J Biomech 2017;57:117–24. doi:10.1016/j. lar cartilage on knee joint mechanics—a 3D finite element analysis. J Biomech
jbiomech.2017.04.008. 2012;45:579–87. doi:10.1016/j.jbiomech.2011.11.003.
[17] Räsänen LP, Tanska P, Zbýň Š, van Donkelaar CC, Trattnig S, Nieminen MT, et al. [42] Wang H, Koff MF, Potter HG, Warren RF, Rodeo SA, Maher SA. An MRI-
The effect of fixed charge density and cartilage swelling on mechanics of knee compatible loading device to assess knee joint cartilage deformation: effect
joint cartilage during simulated gait. J Biomech 2017;61:34–44. doi:10.1016/j. of preloading and inter-test repeatability. J Biomech 2015;48:2934–40. doi:10.
jbiomech.2017.06.041. 1016/j.jbiomech.2015.08.006.
[18] den Broeck J, Vereecke E, Wirix-Speetjens R, Vander Sloten J. Segmentation ac- [43] Tanska P, Mononen ME, Korhonen RK. A multi-scale finite element model for
curacy of long bones. Med Eng Phys 2014;36:949–53. doi:10.1016/j.medengphy. investigation of chondrocyte mechanics in normal and medial meniscectomy
2014.03.016. human knee joint during walking. J Biomech 2015;48:1397–406. doi:10.1016/j.
[19] Mootanah R, Imhauser CW, Reisse F, Carpanen D, Walker RW, Koff MF, et al. jbiomech.2015.02.043.
Development and validation of a computational model of the knee joint for the [44] Halloran JP, Sibole S, Van Donkelaar CC, Van Turnhout MC, Oomens CWJ,
evaluation of surgical treatments for osteoarthritis. Comput Methods Biomech Weiss JA, et al. Multiscale mechanics of articular cartilage: potentials and chal-
Biomed Eng 2014;17:1502–17. doi:10.1080/10255842.2014.899588. lenges of coupling musculoskeletal, joint, and microscale computational mod-
[20] Dube B, Bowes MA, Kingsbury SR, Hensor EMA, Muzumdar S, Conaghan PG. els. Ann Biomed Eng 2012;40:2456–74. doi:10.1007/s10439- 012- 0598- 0.
Where does meniscal damage progress most rapidly? An analysis using three- [45] Shim VB, Hunter PJ, Pivonka P, Fernandez JW. A multiscale framework based
dimensional shape models on data from the Osteoarthritis Initiative. Os- on the physiome markup languages for exploring the initiation of osteoarthri-
teoarthr Cartil 2018;26:62–71. doi:10.1016/j.joca.2017.10.012. tis at the bone–cartilage interface. IEEE Trans Biomed Eng 2011;58:3532–6.
[21] Pena E, Calvo B, Martinez MA, Doblare M. A three-dimensional finite element doi:10.1109/TBME.2011.2165955.
analysis of the combined behavior of ligaments and menisci in the healthy [46] Freutel M, Schmidt H, Dürselen L, Ignatius A, Galbusera F. Finite element mod-
human knee joint. J Biomech 2006;39:1686–701. doi:10.1016/j.jbiomech.2005. eling of soft tissues: material models, tissue interaction and challenges. Clin
04.030. Biomech 2014;29:363–72. doi:10.1016/j.clinbiomech.2014.01.006.
[22] Pena E, Calvo B, Martinez MA, Palanca D, Doblaré M. Finite element analysis of [47] Halonen KS, Mononen ME, Jurvelin JS, Töyräs J, Salo J, Korhonen RK.
the effect of meniscal tears and meniscectomies on human knee biomechanics. Deformation of articular cartilage during static loading of a knee joint–
Clin Biomech 2005;20:498–507. doi:10.1016/j.clinbiomech.2005.01.009. experimental and finite element analysis. J Biomech 2014;47:2467–74. doi:10.
[23] Dhaher YY, Delp SL, Rymer WZ. The use of basis functions in modelling joint 1016/j.jbiomech.2014.04.013.
articular surfaces: application to the knee joint. J Biomech 20 0 0;33:901–7. [48] Stentz-Olesen K, Nielsen ET, De Raedt S, Jørgensen PB, Sørensen OG,
doi:10.1016/S0 021-9290(0 0)0 0 024-5. Kaptein BL, et al. Validation of static and dynamic radiostereometric analysis
[24] Ateshian GA, Soslowsky LJ, Mow VC. Quantitation of articular surface topog- of the knee joint using bone models from CT data. Bone Jt Res 2017;6:376–84.
raphy and cartilage thickness in knee joints using stereophotogrammetry. J doi:10.1302/2046- 3758.66.BJR- 2016- 0113.R3.
Biomech 1991;24:761–76. doi:10.1016/0021- 9290(91)90340- S. [49] Shu L, Yamamoto K, Yao J, Saraswat P, Liu Y, Mitsuishi M, et al. A subject-
[25] Wismans JAC, Veldpaus F, Janssen J, Huson A, Struben P. A three-dimensional specific finite element musculoskeletal framework for mechanics analysis of
mathematical model of the knee-joint. J Biomech 1980;13:677–85. doi:10.1016/ a total knee replacement. J Biomech 2018;77:146–54. doi:10.1016/j.jbiomech.
0021- 9290(80)90354- 1. 2018.07.008.
[26] Vaziri A, Nayeb-Hashemi H, Singh A, Tafti BA. Influence of meniscectomy and [50] Halonen KS, Dzialo CM, Mannisi M, Venäläinen MS, de Zee M, Andersen MS.
meniscus replacement on the stress distribution in human knee joint. Ann Workflow assessing the effect of gait alterations on stresses in the medial tib-
Biomed Eng 2008;36:1335–44. doi:10.1007/s10439- 008- 9515- y. ial cartilage-combined musculoskeletal modelling and finite element analysis.
[27] Meakin JR, Shrive NG, Frank CB, Hart DA. Finite element analysis of the menis- Sci Rep 2017;7:17396. doi:10.1038/s41598- 017- 17228- x.
cus: the influence of geometry and material properties on its behaviour. Knee [51] Galbusera F, Freutel M, Dürselen L, D’Aiuto M, Croce D, Villa T, et al. Mate-
2003;10:33–41. doi:10.1016/S0968-0160(02)00106-0. rial models and properties in the finite element analysis of knee ligaments:
[28] Wilson W, Van Rietbergen B, Van Donkelaar CC, Huiskes R. Pathways of a literature review. Front Bioeng Biotechnol 2014;2:54. doi:10.3389/fbioe.2014.
load-induced cartilage damage causing cartilage degeneration in the knee 0 0 054.
after meniscectomy. J Biomech 2003;36:845–51. doi:10.1016/S0021-9290(03) [52] Lohmander LS, Östenberg A, Englund M, Roos H. High prevalence of knee
0 0 0 04-6. osteoarthritis, pain, and functional limitations in female soccer players
[29] Cooper RJ, Williams S, Mengoni M, Jones AC. Patient-specific parameterised twelve years after anterior cruciate ligament injury. Arthritis Rheumatol
cam geometry in finite element models of femoroacetabular impingement of 2004;50:3145–52. doi:10.1002/art.20589.
the hip. Clin Biomech 2018;54:62–70. doi:10.1016/j.clinbiomech.2018.03.007. [53] Halonen KS, Mononen ME, Töyräs J, Kröger H, Joukainen A, Korhonen RK. Opti-
[30] Hua X, Li J, Wilcox RK, Fisher J, Jones AC. Geometric parameterisation of mal graft stiffness and pre-strain restore normal joint motion and cartilage re-
pelvic bone and cartilage in contact analysis of the natural hip: An ini- sponses in ACL reconstructed knee. J Biomech 2016;49:2566–76. doi:10.1016/j.
tial study. Proc Inst Mech Eng Part H J Eng Med 2015:570–80. doi:10.1177/ jbiomech.2016.05.002.
0954411915592656. [54] Marouane H, Shirazi-Adl A, Adouni M, Hashemi J. Steeper posterior tibial slope
[31] Anderson AE, Ellis BJ, Maas SA, Weiss JA. Effects of idealized joint geometry markedly increases ACL force in both active gait and passive knee joint under
on finite element predictions of cartilage contact stresses in the hip. J Biomech compression. J Biomech 2014;47:1353–9. doi:10.1016/j.jbiomech.2014.01.055.
2010;43:1351–7. doi:10.1016/j.jbiomech.2010.01.010. [55] Mesfar W, Shirazi-Adl A. Biomechanics of changes in ACL and PCL material
[32] Masouros SD, McDermott ID, Amis AA, Bull AMJ. Biomechanics of the properties or prestrains in flexion under muscle force-implications in ligament
meniscus-meniscal ligament construct of the knee. Knee Surg Sport Trauma- reconstruction. Comput Methods Biomech Biomed Eng 2006;9:201–9. doi:10.
tol Arthrosc 2008;16:1121–32. doi:10.10 07/s0 0167-0 08-0616-9. 1080/10255840600795959.
[33] Lohmander LS, Englund PM, Dahl LL, Roos EM. The long-term consequence of [56] Beidokhti HN, Janssen D, van de Groes S, Verdonschot N. The peripheral soft
anterior cruciate ligament and meniscus injuries: osteoarthritis. Am J Sports tissues should not be ignored in the finite element models of the human knee
Med 2007;35:1756–69. doi:10.1177/0363546507307396. joint. Med Biol Eng Comput 2018;56:1189–99. doi:10.1007/s11517-017-1757-0.
[34] Meng Q, An S, Damion RA, Jin Z, Wilcox R, Fisher J, et al. The effect of collagen [57] Yao J, Lancianese SL, Hovinga KR, Lee J, Lerner AL. Magnetic resonance im-
fibril orientation on the biphasic mechanics of articular cartilage. J Mech Behav age analysis of meniscal translation and tibio-menisco-femoral contact in deep
Biomed Mater 2017;65:439–53. doi:10.1016/j.jmbbm.2016.09.001. knee flexion. J Orthop Res 2008;26:673–84. doi:10.1002/jor.20553.
[35] Kazemi M, Li LP. A viscoelastic poromechanical model of the knee joint in [58] Freutel M, Galbusera F, Ignatius A, Dürselen L. Material properties of individual
large compression. Med Eng Phys 2014;36:998–1006. doi:10.1016/j.medengphy. menisci and their attachments obtained through inverse FE-analysis. J Biomech
2014.04.004. 2015;48:1343–9. doi:10.1016/j.jbiomech.2015.03.014.
12 R.J. Cooper, R.K. Wilcox and A.C. Jones / Medical Engineering and Physics 74 (2019) 1–12
[59] Viceconti M, Olsen S, Nolte L-P, Burton K. Extracting clinically relevant data [77] Jorge JP, Simões FMF, Pires EB, Rego PA, Tavares DG, Lopes DS, et al. Finite
from finite element simulations. Clin Biomech 2005;20:451–4. doi:10.1016/j. element simulations of a hip joint with femoroacetabular impingement. Com-
clinbiomech.2005.01.010. put Methods Biomech Biomed Eng 2014:1275–85. doi:10.1080/10255842.2012.
[60] D’Lima DD, Chen PC, Colwell CW Jr. Osteochondral grafting: effect of 744398.
graft alignment, material properties, and articular geometry. Open Orthop J [78] Mengoni M, Kayode O, Sikora SNF, Zapata-Cornelio FY, Gregory DE,
2009;3:61. doi:10.2174/1874325000903010061. Wilcox RK. Annulus fibrosus functional extrafibrillar and fibrous mechanical
[61] Guo H, Santner TJ, Lerner AL, Maher SA. Reducing uncertainty when using behaviour: experimental and computational characterisation. R Soc Open Sci
knee-specific finite element models by assessing the effect of input parame- 2017;4:170807. doi:10.1098/rsos.170807.
ters. J Orthop Res 2017;35:2233–42. doi:10.1002/jor.23513. [79] Ali AA, Shalhoub SS, Cyr AJ, Fitzpatrick CK, Maletsky LP, Rullkoetter PJ, et al.
[62] Guo H, Santner TJ, Chen T, Wang H, Brial C, Gilbert SL, et al. A Validation of predicted patellofemoral mechanics in a finite element model of
statistically-augmented computational platform for evaluating meniscal func- the healthy and cruciate-deficient knee. J Biomech 2016;49:302–9. doi:10.1016/
tion. J Biomech 2015;48:1444–53. doi:10.1016/j.jbiomech.2015.02.031. j.jbiomech.2015.12.020.
[63] Peterfy CG, Schneider E, Nevitt M. The osteoarthritis initiative: report on the [80] Harris MD, Cyr AJ, Ali AA, Fitzpatrick CK, Rullkoetter PJ, Maletsky LP, et al. A
design rationale for the magnetic resonance imaging protocol for the knee. Os- combined experimental and computational approach to subject-specific analy-
teoarthr Cartil 2008;16:1433–41. doi:10.1016/j.joca.2008.06.016. sis of knee joint laxity. J Biomech Eng 2016;138:81004. doi:10.1115/1.4033882.
[64] Erdemir A, Hunter PJ, Holzapfel GA, Loew LM, Middleton J, Jacobs CR, et al. Per- [81] Kiapour A, Kiapour AM, Kaul V, Quatman CE, Wordeman SC, Hewett TE, et al.
spectives on Sharing Models and Related Resources in Computational Biome- Finite element model of the knee for investigation of injury mechanisms:
chanics Research. J Biomech Eng 2018;140:24701. doi:10.1115/1.4038768. development and validation. J Biomech Eng 2014;136:11002. doi:10.1115/1.
[65] Erdemir A, Besier TF, Halloran JP, Imhauser C, Laz PJ, Morrison T, et al. Deci- 4025692.
phering the “Art” in modeling and simulation of the knee joint. J Biomech Eng [82] Gu KB, Li LP. A human knee joint model considering fluid pressure and fiber
2019. doi:10.1115/1.4043346. orientation in cartilages and menisci. Med Eng Phys 2011;33:497–503. doi:10.
[66] Erdemir A. Open knee: open source modeling and simulation in knee biome- 1016/j.medengphy.2010.12.001.
chanics. J Knee Surg 2016;29:107–16. doi:10.1055/s- 0035- 1564600. [83] Schoenbauer E, Szomolanyi P, Shiomi T, Juras V, Zbýň Š, Zak L, et al. Carti-
[67] Rodriguez-Vila B, Sánchez-González P, Oropesa I, Gomez EJ, Pierce DM. Au- lage evaluation with biochemical MR imaging using in vivo Knee compression
tomated hexahedral meshing of knee cartilage structures–application to data at 3 T-comparison of patients after cartilage repair with healthy volunteers. J
from the osteoarthritis initiative. Comput Methods Biomech Biomed Eng Biomech 2015;48:3349–55.
2017;20:1543–53. doi:10.1080/10255842.2017.1383984. [84] Donahue TLH, Hull ML, Rashid MM, Jacobs CR. How the stiffness of menis-
[68] Khoshgoftar M, Vrancken ACT, van Tienen TG, Buma P, Janssen D, Verdon- cal attachments and meniscal material properties affect tibio-femoral contact
schot N. The sensitivity of cartilage contact pressures in the knee joint to pressure computed using a validated finite element model of the human knee
the size and shape of an anatomically shaped meniscal implant. J Biomech joint. J Biomech 2003;36:19–34. doi:10.1016/S0021- 9290(02)00305- 6.
2015;48:1427–35. doi:10.1016/j.jbiomech.2015.02.034. [85] Zielinska B, Donahue TLH. 3D finite element model of meniscectomy: changes
[69] Jones AC, Wilcox RK. Finite element analysis of the spine: towards a in joint contact behavior. J Biomech Eng 2006;128:115–23. doi:10.1115/1.
framework of verification, validation and sensitivity analysis. Med Eng Phys 2132370.
2008;30:1287–304. doi:10.1016/j.medengphy.2008.09.006. [86] Drewniak EI, Crisco JJ, Spenciner DB, Fleming BC. Accuracy of circular contact
[70] Erdemir A, Guess TM, Halloran J, Tadepalli SC, Morrison TM. Considera- area measurements with thin-film pressure sensors. J Biomech 2007;40:2569–
tions for reporting finite element analysis studies in biomechanics. J Biomech 72. doi:10.1016/j.jbiomech.20 06.12.0 02.
2012;45:625–33. doi:10.1016/j.jbiomech.2011.11.038. [87] Bendjaballah MZ, Shirazi-Adl A, Zukor DJ. Biomechanics of the human knee
[71] Henninger HB, Reese SP, Anderson AE, Weiss JA. Validation of computational joint in compression: reconstruction, mesh generation and finite element anal-
models in biomechanics. Proc Inst Mech Eng Part H J Eng Med 2010;224:801– ysis. Knee 1995;2:69–79. doi:10.1016/0968-0160(95)0 0 018-K.
12. doi:10.1243/09544119JEIM649. [88] Zhang J, Chen Z, Wang L, Li D, Jin Z. A patient-specific wear prediction frame-
[72] Anderson AE, Ellis BJ, Weiss JA. Verification, validation and sensitivity stud- work for an artificial knee joint with coupled musculoskeletal multibody-
ies in computational biomechanics. Comput Methods Biomech Biomed Eng dynamics and finite element analysis. Tribol Int 2017;109:382–9. doi:10.1016/j.
2007;10:171–84. doi:10.1080/10255840601160484. triboint.2016.10.050.
[73] Pathmanathan P, Gray RA, Romero VJ, Morrison TM. Applicability analysis of [89] Kazemi M, Li LP, Buschmann MD, Savard P. Partial meniscectomy changes
validation evidence for biomedical computational models. J Verif Valid Uncer- fluid pressurization in articular cartilage in human knees. J Biomech Eng
tain Quantif 2017;2:21005. doi:10.1115/1.4037671. 2012;134:21001. doi:10.1115/1.4005764.
[74] Maas SA, Ellis BJ, Ateshian GA, Weiss JA. FEBio: finite elements for biomechan- [90] Li LP, Korhonen RK, Iivarinen J, Jurvelin JS, Herzog W. Fluid pressure driven
ics. J Biomech Eng 2012;134:11005. doi:10.1115/1.4005694. fibril reinforcement in creep and relaxation tests of articular cartilage. Med
[75] Tadepalli SC, Erdemir A, Cavanagh PR. Comparison of hexahedral and tetrahe- Eng Phys 2008;30:182–9. doi:10.1016/j.medengphy.20 07.03.0 01.
dral elements in finite element analysis of the foot and footwear. J Biomech [91] Shrive NG, O’connor JJ, Goodfellow JW. Load-bearing in the knee joint. Clin
2011;44:2337–43. doi:10.1016/j.jbiomech.2011.05.006. Orthop Relat Res 1978;Mar–Apr;131:279–87.
[76] Maas SA, Ellis BJ, Rawlins DS, Weiss JA. Finite element simulation of ar-
ticular contact mechanics with quadratic tetrahedral elements. J Biomech
2016;49:659–67. doi:10.1016/j.jbiomech.2016.01.024.