Computer Methods in Biomechanics and Biomedical

Engineering

ISSN: 1025-5842 (Print) 1476-8259 (Online) Journal homepage: https://www.tandfonline.com/loi/gcmb20

A finite element model of the human left

ventricular systole

F. Dorri , P. F. Niederer & P. P. Lunkenheimer

To cite this article: F. Dorri , P. F. Niederer & P. P. Lunkenheimer (2006) A finite element model
of the human left ventricular systole, Computer Methods in Biomechanics and Biomedical
Engineering, 9:5, 319-341, DOI: 10.1080/10255840600960546

To link to this article: https://doi.org/10.1080/10255840600960546

Published online: 25 Jan 2007.

Submit your article to this journal

Article views: 2065

View related articles

Citing articles: 2 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=gcmb20
Computer Methods in Biomechanics and Biomedical Engineering,
Vol. 9, No. 5, October 2006, 319–341

A finite element model of the human left ventricular systole

F. DORRI†*, P. F. NIEDERER† and P. P. LUNKENHEIMER‡

†Institute for Biomedical Engineering, University and ETH Zurich, Gloriastrasse 35, CH-8092 Zurich, Switzerland
‡Experimental Thoracovascular and Cardiac Surgery, University of Munster, Domagkstrasse 11, D-48149 Munster, Germany

(Received 20 July 2005; in final form 8 August 2006)

Local wall stress is the pivotal determinant of the heart muscle’s systolic function. Under in vivo
conditions, however, such stresses cannot be measured systematically and quantitatively. In contrast,
imaging techniques based on magnetic resonance (MR) allow the determination of the deformation
pattern of the left ventricle (LV) in vivo with high accuracy. The question arises to what extent
deformation measurements are significant and might provide a possibility for future diagnostic
purposes.
The contractile forces cause deformation of LV myocardial tissue in terms of wall thickening,
longitudinal shortening, twisting rotation and radial constriction. The myocardium is thereby
understood to act as a densely interlaced mesh. Yet, whole cycle image sequences display a distribution
of wall strains as function of space and time heralding a significant amount of inhomogeneity even
under healthy conditions. We made similar observations previously by direct measurement of local
contractile activity. The major reasons for these inhomogeneities derive from regional deviations of the
ventricular walls from an ideal spheroidal shape along with marked disparities in focal fibre orientation.
In response to a lack of diagnostic tools able to measure wall stress in clinical routine, this
communication is aimed at an analysis and functional interpretation of the deformation pattern of an
exemplary human heart at end-systole. To this end, the finite element (FE) method was used to simulate
the three-dimensional deformations of the left ventricular myocardium due to contractile fibre forces at
end-systole. The anisotropy associated with the fibre structure of the myocardial tissue was included in
the form of a fibre orientation vector field which was reconstructed from the measured fibre trajectories
in a post mortem human heart. Contraction was modelled by an additive second Piola –Kirchhoff active
stress tensor.
As a first conclusion, it became evident that longitudinal fibre forces, cross-fibre forces and shear
along with systolic fibre rearrangement have to be taken into account for a useful modelling of systolic
deformation. Second, a realistic geometry and fibre architecture lead to typical and substantially
inhomogeneous deformation patterns as they are recorded in real hearts. We therefore, expect that the
measurement of systolic deformation might provide useful diagnostic information.

Keywords: Heart; Myocardium; Contraction; Systole

1. Introduction It is the task of mathematical modelling to bridge the

Recent developments in cardiac imaging, primarily by
way of magnetic resonance (MR), but also with ultrasound
and X-ray allow for an advanced analysis and documen-
tation of ventricular wall motion under in vivo conditions.
The spatial and temporal resolution of these imaging
methods has in particular been improved to an extent that
the assessment of regional wall dynamics has become a
realistic option in clinical routine. These developments
provide a basis for a renewed and far more detailed
interpretation of the driving structures that keep the
ventricular wall moving than has been possible hitherto.
gap between ventricular morphology and mechanical
function. To this end, a number of theoretical approaches
and numerical models have been put forward to assess
cardiodynamics theoretically and to calculate the stress
and strain distribution in the ventricular wall (Chadwick
1982, Arts et al. 1982, Nielsen et al. 1991, Hunter et al.
1992, Huyghe et al. 1992, Bovendeerd et al. 1992, 1994,
1996, Nash and Hunter 2000, Shoucri 2000, Guccione
et al. 2001a,b). These models have generally been based
on idealised, geometrical representations of animal and
human hearts; likewise, the anisotropic structure associ-
ated with the myocardial fibre architecture has been

*Corresponding author. Email: dorri@biomed.ee.ethz.ch

Computer Methods in Biomechanics and Biomedical Engineering

ISSN 1025-5842 print/ISSN 1476-8259 online q 2006 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/10255840600960546
320 F. Dorri et al.
smoothed and modelled in an idealised fashion,
disregarding local details.
A scrutiny of the human myocardium reveals that it
represents an anisotropic structure whereby the myocytes
are multi-branched and organised in bundles which are
interconnected by a complex hierarchy of connective
tissue (Feneis 1943, Hort 1960, Streeter and Basset 1966,
Streeter et al. 1969, Spotnitz et al. 1974, LeGrice et al.
1995, Hanley et al. 1999). In search of a functional
understanding, various tertiary structural concepts
regarding the ventricular architecture have been postu-
lated in the past, thereby in particular, tractus (Mall
1911), bands, ropes (Torrent-Guasp et al. 1997, 2001),
helices (Ingels 1997) or lamellae (LeGrice et al. 1995,
Young et al. 1998) although corresponding histological
evidence is only partially conclusive. Two further, often
tacitly made assumptions associated with such concepts
are that the wall structure is largely the same throughout
the ventricle and that the myocardial fibres are aligned
almost completely in a tangential direction with respect
to the epi-cardial surface. Although the main fibre
arrangement is found to exhibit a typical gross
architecture, which is similar in all mammalian hearts
examined so far, there are marked local and individual
variations throughout the ventricular wall. Likewise,
short axis sections of a healthy left ventricle (LV) are
mostly round but with numerous local irregularities.
There is furthermore evidence that countless, very short
cleavage clefts are interspersed throughout the myocar-
dium which facilitate some relative movement between
fibre strands, in particular fibre rearrangement during
systolic contraction (LeGrice et al. 1997, Costa et al.
1999, Usyk et al. 2000). The geometry, size and
orientation of these cleavage planes are highly variable;
they are found to exhibit a wide range of variability in
spatial alignment.
From a geometrical point of view, the main features of
systolic ventricular contraction are wall thickening,
longitudinal shortening, rotational wringing and radial
constriction of the myocardium. These patterns are,
however, found to be subjected to appreciable local
variations; in particular, wall thickening is quite
inhomogeneous throughout the ventricle (Chadwick et al.
1989). Likewise, measured wall stress exhibits pro-
nounced and characteristic variations (Lunkenheimer et al.
2004). Mathematical models which have been presented
to date, however, have only partially been aimed at the
description and analysis of these features.
It is conceivable that stress distributions within the
myocardium can be calculated from an inverse
mathematical analysis (Moustakidis et al. 1999). A
prerequisite for such an analysis is a detailed knowledge
of the constitutive properties of myocardium including
the systolic contraction behaviour. MR techniques allow
furthermore the determination of local fibre shortening
during systole (Rademakers et al. 1994, MacGowan et al.
1997). The question, therefore, arises to what extent
a motion and deformation analysis based on MR
measurements along with a mathematical heart model
can be utilised for diagnostic purposes, in particular with
respect to the ventricular fibre pattern and wall stress
distribution.
This work is devoted to an analysis of the deformation
pattern of an exemplary human heart under the influence
of systolic contraction forces. To this end, the geometry of
the epi- and endo-cardial surfaces along with the fibre
direction field was derived from a post mortem human
heart. A continuum mechanics’ approach, i.e. a finite
element (FE) formulation was chosen for the calculation
of systolic contraction. As constitutive model, an
extension of the formulation of Lin and Yin (1998), for
transversely isotropic materials was used, and the systolic
stimulation was modelled by the addition of an active
component to the second Piola – Kirchhoff stress tensor.
As representative for the process the end-systolic state was
considered and compared with measurements within the
limits given by the fact that we analysed an individual case
associated with individual aberrations.
The chosen constitutive model did not allow to include
the effect of systolic fibre rearrangement, i.e. the effect of
the cleavage clefts as well as the influence of the
hierarchical organisation of the connective tissue or the
“accordion-like” contraction of the myocardial tissue
mentioned recently by Harrington et al. (2005), explicitly.
Nevertheless, by adjusting the compressibility of the
material, this shortcoming could partially be compensated.
Since the model was derived from one human heart, it
is exemplary, represents a particular case and not a
population average. An extrapolation to other hearts is not
a priori possible, however, we consider the chosen heart
as representative with respect to the extent of the local
variations which are to be expected in the systolic
contraction pattern due to the individual characteristics of
a healthy human heart.
2. Material behaviour
Myocardial tissue is anisotropic and weakly compressible
because of intra- and extra-vascular fluid displacements
occurring during systole (Yin et al. 1996). The material
behaviour of myocardium depends furthermore on time
and it changes during a heart cycle. Anisotropy, which is
due to the fibrous structure of the tissue, causes moreover
directional dependency of the material behaviour on the
applied loads. Transversely isotropic material models of
various functional forms and material parameters have
been widely used for the mathematical modelling of the
LV. The assumption of transverse isotropy for myocardial
tissue, i.e. the existence of a unique preferred direction
in the material, enables the deduction of a strain energy
function directly from experimental data (Humphrey et al.
1990a,b, Novak et al. 1994). Recently, also orthotropic
material models for myocardium were proposed (Nash
and Hunter 2000, Usyk et al. 2000), but experimental data
are hardly sufficient to deduce the functional form of such
 FE model of the human left ventricular systole
constitutive equations and to determine the associated
material constants without further assumptions.
Transversely isotropic composites can be modelled as
being composed of fibre bundles which are embedded in a
homogeneous matrix (Spencer 1972, 1984) whereby the
fibres are able to interact with the matrix. With C denoting
the right Cauchy – Green deformation tensor and N a unit
vector which determines the anisotropy direction, a
general form of the strain energy function in case of
incompressibility reads (Rivlin and Saunders 1951,
Humphrey et al. 1990)
WðI 1 ; I 2 ; I 4 Þ ¼ W matrix ðI 1 ; I 2 Þ þ W fibres ðI 4 Þ
þ W interaction ðI 1 ; I 2 ; I 4 Þ
I 1 ðCÞ ¼ tr C
1
 where J ¼ (det C)1/2. On the basis of the assumption that
I 2 ðCÞ ¼ ðtr CÞ2 2 tr C 2
2
I3 ¼ 1
I 4 ðC; NÞ ¼ N
C
N
I1, . . . I4 are the invariant measures of deformations,
whereby I3 ¼ 1 implies incompressibility.
In case of myocardial tissue, strain energy functions
have mostly been based on two approaches (Weiss1994).
In the first approach, the symmetry properties of the
structure are considered and the strain energy function is
formulated in the local reference system reflecting the
symmetries (Chuong and Fung 1986, Guccione et al.
1991). Second, a strain energy function is introduced in
the global reference system which depends explicitly on
the fibre vector field N (identical with the anisotropy
direction mentioned above). This field represents the
preferred direction of the material in each point of the
body (Humphrey et al. 1990a,b, Novak et al. 1994).
We modelled the ventricular myocardium as a
transversely isotropic material, composed of a weakly
compressible matrix and fibres (Spencer 1972, 1984),
whereby, the final constitutive behaviour consisted of a
passive as well as of an active component.
The passive constitutive behaviour was formulated by
choosing an appropriate form for Wtotal which was
consisted of a deviatoric (incompressible) part in the form
of equation (1) and a volumetric contribution describing
the compressibility. For the deviatoric part, we used the
functional form proposed by Lin and Yin (1998) for
mammalian left ventricular myocardium under steady-
state barium stimulation for the simulation of systolic
excitation, viz.,
WðI 1 ; I 4 Þ ¼C 0 þ C4 ðI 1 2 3Þ þ C2 ðI 1 2 3Þ2
þ C5 ðI 4 2 1Þ þ C3 ðI 4 2 1Þ2
þ C1 ðI 1 2 3ÞðI 4 2 1Þ
321
The following values for the material parameters
ðg=cm2 Þ were chosen
C 1 ¼ 27:89; C2 ¼ 66:20; C 3 ¼ 51:12;
ð7Þ
C 4 ¼ 40:12; C5 ¼ 0:0032
These parameters correspond to the stiffest state under
steady contraction according to the biaxial tests of Lin and
Yin (1998) and were used as baseline. A number of runs
were also made with other material parameters in order to
determine their influence.
Systolic intra- and extra-vascular fluid displacements
give rise to a compressibility of the tissue. This is taken
into account by adding a volumetric part of the strain
ð1Þ energy function
9K 1=3
ð2Þ W Volumetric ¼ ðJ 2 1Þ2 ð8Þ
2
ð3Þ the relative change in the volume of the ventricular wall
due to contraction is about 4% (Yin et al. 1996), a value
ð4Þ K ¼ 600 kPa ð9Þ
is appropriate. In order to take into account systolic fibre
ð5Þ
arrangement, in turn, an effective compressibility
K ¼ 70 kPa ð10Þ
was chosen. Accordingly, the total passive strain energy
function was written as
W total ¼ W þ W Volumetric ð11Þ
During the systolic phase, the active contraction forces
produced by the myocardial fibres increase gradually.
These forces were modelled in the form of additive
components of the second Piola – Kirchhoff stress tensor
which were added to the passive components in a stepwise
fashion described further down.
3. A representative fibre field N and the ventricular
geometry
A healthy human heart in rigor mortis weighing about
500 g was chosen as a representative example. It was
perfused via the coronary arteries for 24 h with saline at a
pressure of 120 mm Hg. The perfusate was subsequently
replaced by a 10% formaldehyde solution and the
perfusion was continued for another 24 h. The atria were
trimmed down to the ventricular base and the heart was
submerged for two weeks in a 10% formaldehyde
solution. The ventricles were then filled with Technovit
(Haereus-Kulzer, Germany) such that mouldings of the
ventricular cavities were obtained. Together with the two-
component resin a wooden rod was axially anchored in the
left ventricular cavity, which served to fix the heart in a jig
on top of an electromagnetic digitising tablet. After having
removed the epi-cardium together with the perivascular
fat, the geometries of the epi- and endo-cardial surfaces,
ð6Þ respectively, were determined. Subsequently, the SPOT
322
(fibre Strand Peel-Off Technique, see 7, Discussion)
method was applied to prepare both ventricles including
the septum strand by strand, from base to apex and from
epi-cardium to endo-cardium. During the entire process, a
stepwise digitisation was performed by using a manual
stylus (figure 1). The data were processed to produce a
three-dimensional unit fibre vector field, N, which
described the fibre architecture for the entire LV. The
procedure is described in detail by Lunkenheimer et al.
(1997), Cryer et al. (1997).
Figure 1. Fibre trajectories of the LV approximated as cubic splines from the digitised points outlining the myocardial fibre strands obtained from a post
mortem human heart in rigor mortis.
F. Dorri et al.
A geometrical model of the LV was produced from the
data sets containing about 2000 points representing the
endo-cardial and some 4500 points representing the epi-
cardial surface (figure 2). For this purpose, Nonuniform
Rational B-Splines (NURBs) (Rogers 2001) were used to
connect the points in a smooth fashion (Raindrop
Geomagicw). A mesh of nine-node FE was created
whereby eight nodes were used to define hexahedral
elements while the ninth node was reserved for the
pressure. More than 46,700 elements were necessary in
 FE model of the human left ventricular systole 323

Figure 2. Geometry of the LV constructed from the data sets containing typically 2000 points representing the endo-cardial and some 4500 points
representing the epi-cardial surface.

order to include the desired level of anatomical details
(MSC-Marc Mentatw).
The measured fibre trajectories obtained from SPOT
were then introduced into the geometrical model such that
the mesh covered the fibres entirely. For each element, the
average fibre direction was determined as described by
Dorri (2004).
4. Boundary conditions
In the natural state, the myocardium is surrounded by the
pericardial sac. At the basis, the LV is restrained by the
aorta, the atria and it is furthermore connected to the right
ventricle. Quantitative information on these boundary
conditions is not available, yet, in order to avoid undesired
rigid body displacements, appropriate boundary con-
ditions have to be chosen in the FE formulation. In our
model, the motion of the basal elements was suppressed.
A further boundary condition derives from the
intracavital blood pressure acting on the endo-cardial
surface. The blood pressure in a real ventricle depends on
time and location. In order to determine the spatial
distribution of the pressure as a function of time, the fluid
dynamics of the blood in the ventricle would have to be
considered. However, since the pressure gradients in the
324 F. Dorri et al.
LV during ejection are mostly small (typically
, 10 mm Hg) in comparison with the absolute pressure,
we discarded local variations and assumed a uniform
dependence of the pressure on time. The curve
PLV ¼ 2944t 2 þ 245t 0 # t # 0:2 s ð12Þ
represents a parabola which was considered to approxi-
mate a representative systolic pressure curve with a
maximum of
PMax ¼ 16:0 kPa tMax ¼ 0:13 s ð13Þ
which corresponds to 120 mm Hg, the normal systolic
blood pressure for a healthy heart.
This pressure was uniformly applied to the entire endo-
cardial surface.
The potential impact of the right ventricular pressure on
septal deformation was modelled by the curve
PRV ¼ 2237t 2 þ 62t 0 # t # 0:2 s ð14Þ
with a maximum of
PMax ¼ 4:0 kPa tMax ¼ 0:13 s ð15Þ
which corresponds to 30 mm Hg. This pressure was
applied in the septal region.
In prescribing the intracavital pressures as boundary
condition, an indirect simulation is performed: In a real
heart, the contractile forces create the intracavital
pressure, while in our simulation the forces equilibrated
the pressure. This procedure is justified insofar as inertial
forces associated with wall motion can be neglected.
5. Modelling of active contraction
A global coordinate system (unit vectors i, j, k, associated
coordinates x, y, z) was chosen such that the k-axis is
oriented along the long axis of the LV. In our geometry,
which is characterised by pronounced local deviations
from a symmetric form this axis cannot be defined exactly,
yet, this is of secondary importance only since the global
coordinate system can be chosen arbitrarily. The j-axis is
perpendicular to the k-axis and is oriented approximately
in the anterior – inferior direction. The i-axis, finally, is
perpendicular to both axes defined before (figure 2).
A local coordinate system, xF, yS, zN for each element
was defined as follows. First, the fibre orientation was
determined in the middle point of each element as an
average vector of the measured neighbour fibre orien-
tations (Dorri 2004), which we denoted by e01 . We
identified this direction with positive values of xF, which
therefore represented the axis along which the fibres were
oriented in the middle point of the considered element. A
second vector e02 , defining yS, was given as the vector
product of e01 and the vector 2k
e02 ¼ 2e01 £ k ð16Þ
As the long axis of the LV was chosen along the k-axis
of the global reference system ði; j; kÞ, e02 was directed
inward and perpendicular to the e01 and k axes. The third
direction, e03 , defining the coordinate zN followed from
e03 ¼ e01 £ e02 ð17Þ
The relation between the unit vectors e0m of the local
coordinate system (xF, yS, zN) and the unit vectors e n of the
global reference system (x, y, z) was given by
transformation matrix Q
0 1
q11 q12 q13
B C
Q¼B @ 21
q q22 q23 C
A ð18Þ
q31 q32 q33
with components
qmn ¼ e0m
e n ð19Þ
as follows
{e0m } ¼ Q{e n } m; n ¼ 1; 2; 3 ð20Þ
Systolic contraction was modelled by defining the total
second Piola –Kirchhoff stress tensor S as the sum of the
passive stress tensor S passive derived from the strain energy
function (equation 11) and an additional active component
S active :
S ¼ S passive þ S active ð21Þ
While for the FE calculation the tensor, S, had to be
known in the global coordinate system, the active
component of the second Piola –Kirchhoff stress tensor
ðS0active Þ was first determined in the local coordinate
system, because it depended on the local fibre direction, N.
To this end, the initial state corresponding to end-diastole
was assumed as stress-free (eventual end-diastolic initial
stresses were assumed to be small in comparison with the
active stresses resulting at end-systole and were therefore
ignored in our simulations). A fixed time step, Dt, was
chosen and the intracavital pressure curve was subdivided
accordingly (since the simulation was quasistatic, the time
step had no further influence). The components of the
active stress tensor were at each integration step subjected
to an incremental increase ðS0 inc active Þ, which was chosen
such that the intracavital pressure was overcome and could
be equilibrated at the end of each step, the criterion being
quasistatic contraction (i.e. a sequence of incremental
equilibrium states). The assumption was thereby
implicitly made that all fibres exhibit the same contraction
behaviour, i.e. all fibres were treated alike. In case of our
thick-walled, anisotropic and irregular structure, this
integration scheme was however associated with a certain
 FE model of the human left ventricular systole
amount of uncertainty which was basically inherent in the
chosen procedure (prescribed intracavital pressure and
wall stresses derived thereof instead of the physiologically
true opposite). Yet, parametric variation studies allowed
us to ascertain that this uncertainty was within 10% of
the increment and as such substantially less than the
inhomogeneity due to the fibre architecture.
In the development of a specific form of the active stress
tensor, a number of experimental facts were observed.
First, from imaging of the heart (primarily by MRI and X-
ray ventriculography) can be concluded (Stuber et al.
1999), that there are four gross features regarding the
deformation process during the contraction of a healthy
LV. These features include wall thickening, longitudinal
shortening, twisting rotation and radial constriction.
Second, the amount of cross-fibre stresses has recently
been studied experimentally (Lin and Yin 1998). These
measurements along with other observations document
that fibre, cross-fibre and shear stresses develop
simultaneously in the myocardium as is to be expected
due to the syncytial nature of the tissue (Mazhari and
McCulloch 1999, Usyk et al. 2000).
Accordingly, we conducted a consecutive analysis of
the influence of the various components of the active stress
tensor in view of these effects. Thereby, there was no
influence of sarcomere length or the velocity of sarcomere
shortening as the simulation was essentially quasistatic.
First calculations were performed with an active stress
tensor which contained only one component S0active ðF; FÞ,
i.e. the component in the fibre direction was incrementally
increased ðS0 inc
active Þ along with the prescribed intracavital
pressure as described previously. As expected, we found
that the component S0active ðF; FÞ alone could not produce a
realistic deformation pattern. In fact, a simulation of the
contraction with S0active ðF; FÞ only caused longitudinal
elongation of the LV instead of shortening, as was already
reported in previous FE models (Huyghe et al. 1992,
Bovendeerd et al. 1994).
To simulate the longitudinal shortening, in a second
step, a normal component S0active ðN; NÞ, i.e. an active stress
in cross-fibre direction zN was added. The relation between
S0active ðF; FÞ and S0active ðN; NÞ is not straightforward,
however. Nevertheless, the measurements performed by
Lin and Yin (1998) on barium-contracted rabbit
myocardium had shown that S0active ðN; NÞ could lie
between 20 and 62% of the S0active ðF; FÞ. Accordingly,
the cross-fibre stress was assumed to be proportional to the
longitudinal fibre stress. Upon application of an active
stress tensor with fibre and cross-fibre components in the
form
0 1
S0active ðF; FÞ 0 0 components of the strain tensor depend on the local
B C
S0active ¼ B
@
0 0 0 C
A ð22Þ
0 0 S0active ðN; NÞ
the longitudinal shortening and wall thickening could be
simulated but it was associated with some unphysiological
325
swelling of the geometry, moreover the twisting rotation
was not simulated correctly.
The dense branching of the myocytes suggested that a
shear component in the active stress tensor should be
included; therefore, in a third step, an active stress tensor
S0active in the following form
0 1
S0active ðF; FÞ 0 0
B C
S0active ¼ B 0 0 S0active ðS; NÞ C
@ A
0 S0active ðN; SÞ S0active ðN; NÞ
ð23Þ
was used. Again, no quantitative information was
available that would allow to choose the amount of the
shear. We concluded from the simulations that the shear
component S0active ðS; NÞ should be between 3 and 7% of the
fibre component S0active ðF; FÞ.
We set the incremental values of the second Piola –
Kirchhoff active stress tensor S0 inc
active as follows
0 1
3 0 0
inc B C
S0 active ¼ B
@ 0 0 0:1 A
C ð24Þ
0 0:1 1:8
Here we assumed that the value of S0active ðN; NÞ is about
60% of S0active ðF; FÞ, and the value of S0active ðS; NÞ is about
3% of S0active ðF; FÞ. All values of the stress components in
(equation 24) are given in kPa.
In the global coordinate system, S active could finally be
determined from
S active ¼ Q T S0active Q ð25Þ
whereby the transformation matrix Q had to be
determined for each element.
6. Results
6.1 Evaluation procedure
In our geometrical model of the human LV which exhibits
no symmetry, all calculations were made in a global
Cartesian coordinate system. To assess theoretically,
estimated values of strains with available experimental
data we, therefore, calculated the components of the strain
tensor in the fibre-specific coordinate system xF, yS, zN.
Thereby, it has to be noted that the directions of the xF, yS,
zN axes change during the deformation process and had to
be updated at the end of each increment. As the
orientation of the coordinate system, care has to be
exercised when estimated values of strain components are
compared with reported experimental data from high
resolution MR imaging (Moore et al. 2000, Sinusas et al.
2001). The values given in the following represent
extrapolations to coordinate systems which are oriented
326
similarly as those usually applied when symmetric
approximations are used.
In contrast to principal and normal strains, shear strains
cannot be measured with sufficient precision to allow for a
useful comparison. Accordingly, we restricted our analysis
to principal and normal strains. To make a regional
comparison of the model with the results of tagged MRI
measurements in the normal human LV, we estimated
average values of strain components in the basal,
equatorial and apical areas of the LV, respectively.
6.2 Model with effective compressibility
In the model with an increased compressibility ðK ¼
70 kPaÞ to take into account systolic fibre rearrangement
using (equation 24), the volume of the LV decreased
during systolic contraction from the initial value of
118 ml to the final value of 47 ml, which corresponds to
a stroke volume of 71 ml, i.e. the ejection fraction was
about 0.6. An overall assessment of our simulations
showed that the final value of the (F, F) component of
the active stress tensor (after completion of 45
increments in this simulation) was largely in agreement
with theoretical estimations (maximum value of about
135 kPa) made by other investigators (McCulloch et al.
1992, Guccione and McCulloch 1993, Hunter et al.
1998).
The von Mises stresses were found to vary in most
regions around 100 kPa within a range of about 50 kPa.
Some extreme values of stress were recorded which
however can be attributed to singular irregularities
associated with surface elements and which are not of
significance because they are mostly due to numerical
effects associated with the modelling.
Due to the compressibility, wall thickening was
somewhat less (around 20% in most areas) in comparison
with reported values in vivo. Longitudinal shortening, in
turn, measured from epi- and endo-cardial surfaces to the
base were equal and about 20%, this value is comparable
with experimental measurements. In a healthy adult heart
longitudinal shortening from endo-cardial surface is more
than epi-cardial surface to the base. For the LV it is about
15% measured from the epi-cardial surface (Stuber 1997).
Table 1. Estimated average of principal strains (EMax, EInt, EMin) ðK ¼ 70 kPaÞ.
Strain Septal
EMax
Basal 0.27 ^ 0.07
Equatorial 0.33 ^ 0.07
Apical 0.33 ^ 0.07
EInt
Basal 20.13 ^ 0.07
Equatorial 20.20 ^ 0.07
Apical 20.20 ^ 0.07
EMin
Basal 20.33 ^ 0.07
Equatorial 20.27 ^ 0.07
Apical 20.27 ^ 0.07
F. Dorri et al.
Twisting rotation is counterclockwise when viewed from
the apex but irregularly distributed from the base to the
apex on the endo-cardial and epi-cardial surfaces.
Experimental data (Moore et al. 2000), have shown that
this angle increases between basal and apical levels from
the epi-cardium (about 10 degrees) to the endo-cardium
(about 14 degrees). In this simulation twisting rotation was
about 10 degrees on the anterior wall.
The average radial displacement of the myocardial
midwall material points is directed inward throughout the
LV but with significant spatial variations. According to
recent experimental reports (see table 4 of Moore et al.
(2000)), it is smaller in the septum (about 2 3 ^ 2 mm)
than in the lateral and inferior walls (about 2 5 ^ 2 mm).
It was found to be largest at the apical inferior wall
(2 6.3 ^ 1.3 mm) and lowest at the apical anterior wall
(2 2.4 ^ 1.1 mm). The radial displacement of the midwall
material points in our model from base to apex and around
the LV showed also an irregular distribution; it was about
2 2.5 ^ 1 mm and in some locations near the base
practically negligible. This could partially be attributed to
the boundary condition applied at the base.
The results of the estimated average values of the
principal strains in basal, equatorial and apical areas of the
LV are summarised in table 1. In most regions the principal
strain EMax was found to vary around 0.20, EInt around
2 0.20 and EMin around 2 0.27 within a range of about
0.07. Figure 3 shows colour-coded maps of calculated
principal and normal strains in a septal –lateral long-axis
section at end-systole, components of the strain tensor are
referenced to the end-diastole. Estimated average values
of strain along fibre ðEff Þ and cross-fibre directions (Ess ,
Enn ) are summarised in table 2. Figure 3 shows in addition
that in most regions Eff varied around 2 0.20, while Ess
varied around 0.20 and Enn around 2 0.27 within a range
of about 0.07. In this simulation calculated principal
strains, fibre and cross-fibre strains are generally smaller
than experimental measurements (see table 3 of Moore
et al. (2000), Tseng et al. (2000)). Figures 4 –6, show
transmural variation of radial (ERR), circumferential (ECC)
and longitudinal (ELL) strains in basal, equatorial and
apical areas of the LV from epi-cardium to endo-cardium.
Estimated average values of radial, circumferential and
Anterior Lateral Inferior
0.20 ^ 0.07 0.27 ^ 0.07 0.13 ^ 0.07
0.20 ^ 0.07 0.27 ^ 0.07 0.20 ^ 0.07
0.07 ^ 0.07 0.13 ^ 0.07 0.20 ^ 0.07
20.20 ^ 0.07 20.20 ^ 0.07 20.07 ^ 0.07
20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07
20.13 ^ 0.07 20.27 ^ 0.07 20.20 ^ 0.07
20.27 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07
20.33 ^ 0.07 20.27 ^ 0.07 20.33 ^ 0.07
20.33 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07
 FE model of the human left ventricular systole 327

Figure 3. Colour-coded maps of strain pattern in a septal–lateral long-axis section at end-systole ðK ¼ 70 kPaÞ. These figures show distribution of
strain along: (A) fibre direction F ðEff Þ; (B) cross-fibre direction S ðEss Þ; (C) cross-fibre direction N ðEnn Þ; (D –F) show principal strains ðEMax ; EInt ; EMin Þ.
Components of the strain tensor are referenced to the end-diastole.
328 F. Dorri et al.

Table 2. Estimated average of normal strains along fibre (Eff) and cross-fibre(Ess, Enn) directions ðK ¼ 70 kPaÞ.

Strain Septal Anterior Lateral Inferior

Eff
Basal 20.13 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07 20.13 ^ 0.07
Equatorial 20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07
Apical 20.13 ^ 0.07 20.20 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07
Ess
Basal 0.20 ^ 0.07 0.13 ^ 0.07 0.20 ^ 0.07 0.13 ^ 0.07
Equatorial 0.20 ^ 0.07 0.13 ^ 0.07 0.20 ^ 0.07 0.13 ^ 0.07
Apical 0.20 ^ 0.07 0.07 ^ 0.07 0.13 ^ 0.07 0.13 ^ 0.07
Enn
Basal 20.27 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07 20.20 ^ 0.07
Equatorial 20.27 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07
Apical 20.27 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07 20.20 ^ 0.07

longitudinal strains are summarised in table 3. In most
regions the ERR was found to vary around 0.20, ECC
around 2 0.20 and ELL around 2 0.27 within a range of
about 0.07. As such, these results were smaller in
comparison to the model with a weekly compressible
material discussed in the next paragraph. Here, ECC is in
agreement with measured values but ERR and ELL are
smaller than experimental measurements (see table 2 of
Moore et al. (2000), Tseng et al. (2000)).
6.3 Model with weakly compressible material
In the model with weekly compressible material
ðK ¼ 600 kPaÞ, the results of the simulation using the

Figure 4. Variation of radial ðERR Þ, circumferential ðECC Þ and longitudinal ðELL Þ components of elastic strain (^0.07) in septal, anterior, lateral and
inferior of basal area from epi-cardium to endo-cardium ðK ¼ 70 kPaÞ.
 FE model of the human left ventricular systole
Figure 5. Variation of radial ðERR Þ, circumferential ðECC Þ and longitudinal ðELL Þ components of elastic strain (^0.07) in septal, anterior, lateral and
inferior of equatorial area from epi-cardium to endo-cardium ðK ¼ 70 kPaÞ.
tensor according to (equation 24) are shown in (figure 7).
In this particular case, the volume of the LV decreased
from the initial value of 118 ml to the final value of 79 ml,
which corresponds to a stroke volume of 39 ml i.e. ejection
fraction was about 0.33. This low value for the ejection
fraction will be commented later on in paragraph 7.2. All
values of the von Mises stress in the pictures (figure 7) are
given in kPa. These values are seen to vary in most regions
around 100 kPa within a range of about 50 kPa. Extreme
values of stress for a number of elements were again
confined to certain regions close to the surface.
Areas exhibiting high stresses are particularly well
apparent on the endo-cardial surface. As expected, wall
thickening was inhomogeneously distributed; it is
remarkable that in some locations maximal wall
thickening was more than 40% as is measured in vivo.
Longitudinal shortening, in turn, measured from epi-
cardial and endo-cardial surfaces to the base were equal
and about 5%, which is too small in comparison with
experimental measurements. In this model, the maximal
amount of the epi-cardial torsion was again about 10
329
degrees which is clearly represented on anterior wall
(figure 7).
The radial displacement of the midwall material points
in this model from base to apex and around the LV showed
also an irregular distribution; but it was generally (too)
small (about 2 1.5 ^ 1 mm) and in some locations near
the base practically negligible. Particularly the radial
inward motion of the apical wall was very small in
comparison with experimental measurements.
The results of estimated average values of principal
strains, summarised in table 4, are in good agreement with
recent tagged MRI measurements of principal strains (see
table 3 of Moore et al. (2000), Tseng et al. (2000)). It can
be seen that in most regions the principal strain EMax was
found to vary around 0.40, EInt around 2 0.13 and EMin
around 2 0.20 within a range of about 0.07. Figure 8
shows colour-coded maps of calculated principal and
normal strains in a septal –lateral long-axis section at end-
systole. As expected, the strain pattern represented a
heterogeneous distribution but with a limited range of
fluctuations. Estimated average values of strain along fibre
330 F. Dorri et al.

Figure 6. Variation of radial ðERR Þ, circumferential ðECC Þ and longitudinal ðELL Þ components of elastic strain (^0.07) in septal, anterior, lateral and
inferior of apical area from epi-cardium to endo-cardium ðK ¼ 70 kPaÞ.

ðEff Þ and cross-fibre directions (Ess ,Enn ) are summarised
in table 5. Figure 8 shows that in most regions Eff which
characterizes fibre shortening varied around 2 0.13, while
Ess varied around 0.40 and Enn around 2 0.13 within a
range of about 0.07. These results are in agreement with
MRI measurements of myocardial fibre shortening in
human using diffusion-sensitive MR imaging which
showed that fibre shortening was more uniform than
cross-fibre or principal strains with a mean value of about
2 0.12 ^ 0.01 (Tseng et al. 2000).
Figures 9 – 11 exhibit furthermore colour-coded maps of
strain distributions in basal, equatorial and apical short-
axis sections at end-systole. The essentially heterogeneous
character of strain values with a limited band of fluctuations

Table 3. Estimated average of radial (ERR), circumferential (ECC) and longitudinal (ELL) strains ðK ¼ 70 kPaÞ.

Strain Septal Anterior Lateral Inferior

ERR
Basal 0.20 ^ 0.07 0.27 ^ 0.07 0.20 ^ 0.07 0.13 ^ 0.07
Equatorial 0.27 ^ 0.07 0.13 ^ 0.07 0.20 ^ 0.07 0.20 ^ 0.07
Apical 0.27 ^ 0.07 0.07 ^ 0.07 0.13 ^ 0.07 0.20 ^ 0.07
ECC
Basal 20.13 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07
Equatorial 20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07 20.27 ^ 0.07
Apical 20.20 ^ 0.07 20.20 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07
ELL
Basal 20.27 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07 20.20 ^ 0.07
Equatorial 20.27 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07 20.20 ^ 0.07
Apical 20.27 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07 20.27 ^ 0.07
 FE model of the human left ventricular systole 331

Figure 7. Simulation of the systolic contraction ðK ¼ 600 kPaÞ using the second Piola–Kirchhoff active stress tensor of equation (23) including
longitudinal fibre forces, cross fibre forces and shear.

could be determined also in short-axis sections. We of normal strains in septal, anterior, lateral and inferior
calculated also radial (ERR), circumferential (ECC) and from epi-cardium to endo-cardium. Additionally, esti-
longitudinal (ELL) strains in basal, equatorial and apical mated average values of radial, circumferential and
areas of the LV. Figures 12 –14 shows transmural variation longitudinal strains are summarised in table 6. In most

Table 4. Estimated average of principal strains (EMax, EInt, EMin) ðK ¼ 600 kPaÞ.

Strain Septal Anterior Lateral Inferior

EMax
Basal 0.40 ^ 0.07 0.40 ^ 0.07 0.40 ^ 0.07 0.27 ^ 0.07
Equatorial 0.60 ^ 0.07 0.40 ^ 0.07 0.47 ^ 0.07 0.33 ^ 0.07
Apical 0.60 ^ 0.07 0.20 ^ 0.07 0.40 ^ 0.07 0.47 ^ 0.07
EInt
Basal 20.07 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.07 ^ 0.07
Equatorial 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07
Apical 20.13 ^ 0.07 20.07 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07
EMin
Basal 20.27 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07
Equatorial 20.20 ^ 0.07 20.20 ^ 0.07 20.27 ^ 0.07 20.20 ^ 0.07
Apical 20.27 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07 20.20 ^ 0.07
332 F. Dorri et al.

Figure 8. Colour-coded maps of strain pattern in a septal–lateral long-axis section at end-systole ðK ¼ 600 kPaÞ. These figures show distribution of
strain along: (A) fibre direction F ðEff Þ; (B) cross-fibre direction S ðEss Þ; (C) cross-fibre direction N ðEnn Þ; (D –F) show principal strains ðEMax ; EInt ; EMin Þ.
Components of the strain tensor are referenced to the end-diastole.
 FE model of the human left ventricular systole 333

Table 5. Estimated average of normal strains along fibre (Eff) and cross-fibre(Ess, Enn) directions ðK ¼ 600 kPaÞ.

Strain Septal Anterior Lateral Inferior

Eff
Basal 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.07 ^ 0.07
Equatorial 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.20 ^ 0.07
Apical 20.13 ^ 0.07 20.07 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07
Ess
Basal 0.40 ^ 0.07 0.33 ^ 0.07 0.33 ^ 0.07 0.27 ^ 0.07
Equatorial 0.53 ^ 0.07 0.40 ^ 0.07 0.40 ^ 0.07 0.33 ^ 0.07
Apical 0.53 ^ 0.07 0.20 ^ 0.07 0.40 ^ 0.07 0.47 ^ 0.07
Enn
Basal 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07
Equatorial 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.07 ^ 0.07
Apical 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07

regions the ERR was found to vary around 0.33, ECC around
2 0.07 and ELL around 2 0.13 within a range of about 0.07.
These results show that ERR and ELL are in good agreement
with MRI measurements (see table 2 of Moore et al.
(2000), Tseng et al. (2000)), ECC on the other hand is larger
than measured values (around 2 0.20).
7. Discussion
7.1 Determination of the fibre field by the SPOT
technique
Myocardial fibres are multiply branched. In histologic
preparation, no beginnings and endings of fibre strands
can furthermore be found. Nevertheless, the tissue is
highly anisotropic and a preferred direction of the average
fibre orientation can clearly be discerned, because
myocytes form endless chains, which are densely meshed
by bridges to their paralleling neighbours. Due to this
syncytial nature of the tissue, however, the mean fibre
orientation, which is considered to coincide with the axis
of anisotropy cannot be determined in a straightforward
fashion. Accordingly, whatever method one applies in the
attempt to analyse the main alignment of myocardial
fibres, one has to destroy essential tissue structures. Mall’s
(1911), approach was to peel out of the ventricular muscle
distinct strands which he understood as being prominent
and functionally privileged parts of a continuum. By so
doing he accepted to destroy an unnumbered amount of
linkages which join the sequester with its bed.
In a previous study (Lunkenheimer et al. 1997), we
combined the peeling technique with an electromagnetic
digitising procedure by which the boundaries of the
myocardial fibre strands, once isolated were compiled in
their spatial location and orientation. Then, the digitised
strands were taken away to give access to deeper layers, a
step which offered a rough access to the complex system
of interlayer connections. Finally, all digitised fibre
strands were compiled to reconstruct the original ventricle
in a computer simulation as composed of the totality of the
individual strands in their spatial orientation. The
drawback of our method was, first, the manual digitising
procedure which is susceptible to artefacts as any other
hand-guided procedure and second, the peeling procedure
itself which isolates but randomly selected fibre
populations of variable thickness, length and alignment,
which sequesters is currently called fibres or bundles. The
totality of myocytes contained in such a bundle is
excluded from their proper spatial analysis, and, more
prejudicial, the totality of the spatial contractile linkages
of the bundle with its surroundings abstains from being
displayed, except that one along which the strand
arbitrarily had been severed.
In consideration of all aspects, the SPOT technique,
although associated with a number of drawbacks, can be
considered to expose the fibrous architecture realistically
in that it can be assumed that a careful peeling, allowing
strands to follow the direction of least mechanical
resistance will expose the axis of mechanical anisotropy.
7.2 Ejection fraction
In a healthy adult heart, the ejection fraction, i.e. the ratio
of the stroke volume to the end diastolic volume under
resting condition, is about 0.60. Such a value was obtained
when an increased effective compressibility for the
myocardial tissue was applied. In case of weekly
compressible material, which is more realistic, however,
the ejection fraction was considerably lower (33%). This
finding needs to be interpreted. One major mechanism for
the ejection of blood is wall thickening, which contributes
around 45% to the LV stroke volume. Longitudinal
shortening and radial inward motion of the wall
(circumferential shortening), in turn, have been estimated
to contribute around 55% of the stroke volume (Dumesnil
and Shoucri 1991). While wall thickening reached
realistic values in our weekly compressible model, in
particular longitudinal shortening was insufficient.
Somewhat higher stroke volumes were reached by
increasing the fibre forces. A further simulation was made
with the following incremental second Piola– Kirchhoff
active stress tensor
0 1
9 0 0
B C
inc
S0 active ¼ B
@0 0 0:3 C
A: ð26Þ
0 0:3 5:4
334 F. Dorri et al.

Figure 9. Colour-coded maps of strain pattern in a basal short-axis section at end-systole ðK ¼ 600 kPaÞ. These figures show distribution of strain
along: (A) fibre direction F ðEff Þ; (B) cross-fibre direction S ðEss Þ; (C) cross-fibre direction N ðEnn Þ; (D– F) show principal strains ðEMax ; EInt ; EMin Þ.
Components of the strain tensor are referenced to the end-diastole.

The number of increments and the bulk modulus K was
at the same time increased to 90 and 6 MPa, respectively,
such that the relative change in the volume of the
ventricular wall due to contraction remained about 4%. In
this simulation the volume of the LV decreased from the
initial value of 118 ml to the final value of 58 ml, which
corresponds to a stroke volume of 60 ml. As a result,
however, the stresses increased to unrealistically high
values while longitudinal shortening was still low.
To assess furthermore, the effect of material parameters
in our calculations we used material constants of six other
specimens suggested by Lin and Yin, (see table 1 of Lin
 FE model of the human left ventricular systole 335

Figure 10. Colour-coded maps of strain pattern in an equatorial short-axis section at end-systole ðK ¼ 600 kPaÞ. These figures show distribution of
strain along: (A) fibre direction F ðEff Þ; (B) cross-fibre direction S ðEss Þ; (C) cross-fibre direction N ðEnn Þ; (D –F) show principal strains ðEMax ; EInt ; EMin Þ.
Components of the strain tensor are referenced to the end-diastole.
336 F. Dorri et al.

Figure 11. Colour-coded maps of strain pattern in an apical short-axis section at end-systole ðK ¼ 600 kPaÞ. These figures show distribution of strain
along: (A) fibre direction F ðEff Þ; (B) cross-fibre direction S ðEss Þ; (C) cross-fibre direction N ðEnn Þ; (D– F) show principal strains ðEMax ; EInt ; EMin Þ.
Components of the strain tensor are referenced to the end-diastole.
 FE model of the human left ventricular systole
Figure 12. Variation of radial ðERR Þ, circumferential ðECC Þ and longitudinal ðELL Þ components of elastic strain (^0.07) in septal, anterior, lateral and
inferior of basal area from epi-cardium to endo-cardium ðK ¼ 600 kPaÞ.
and Yin (1998)). Resulting changes in ejection fraction
and stress-strain pattern were minor. Using material
parameters (g/cm2) of (equation 27) increased the ejection
fraction to about 0.38, simultaneously higher strain
domains were reached in the wall without changing the
longitudinal shortening.
C 1 ¼ 213:14; C2 ¼ 17:05; C 3 ¼ 28:94;
ð27Þ
C 4 ¼ 9:48; C5 ¼ 0:35
Material parameters are likely to influence the results of
simulation slightly but they do not change the pattern of
deformation significantly.
Insufficient longitudinal shortening and radial inward
displacement (circumferential shortening) were obviously
the major causes for the low stroke volume in our weekly
compressible FE model. A first reason for this derives
from the geometrical structure of apical wall, which
acuminates to a thin membrane-like structure. Due to
technical restrictions for the digitisation of the epi-
cardium this geometrical feature was insufficiently
337
reflected in our model. As a result, the mesh in the apical
region consisted of brick elements, which, under large
curvature conditions, resist deformation generally. It
should, however, be noted that most mathematical models
have treated the apex as thick walled and only in a recent
ventricular model in diastole it was attempted to construct
a more realistic structure of the apex (Stevens et al. 2003).
Due to the weak compressibility of the myocardium
extensive rearrangement of muscle fibres during systole is
mandatory to yield sufficient systolic wall thickening,
circumferential and overall longitudinal shortening (Wald-
man et al. 1985, Rademakers et al. 1994, MacGowan et al.
1997). Whereas the amount of fibre shortening is
comparable from the base to the apex throughout the
myocardium, local wall thickening is quite inhomo-
geneous and increases significantly from sub-epi-cardium
to sub-endo-cardium (Sabbah et al. 1981, Waldman et al.
1985, Rademakers et al. 1994). Realignment and
interposition of fibres, causing an increase of the number
of fibre cross-sections between epi- and endo-cardium
during systole, appears to be necessary to produce these
effects, in particular longitudinal shortening (Hort 1960,
338
Figure 13. Variation of radial ðERR Þ, circumferential ðECC Þ and longitudinal ðELL Þ components of elastic strain (^0.07) in septal, anterior, lateral and
inferior of equatorial area from epi-cardium to endo-cardium ðK ¼ 600 kPaÞ.
Spotnitz et al. 1974). The intrinsic structure of the
myocardium including, cleavage clefts are major features
to this end. It is therefore necessary to include some
mechanism describing this properties; adapting material
constants or orthotropic material models alone cannot
change the pattern of deformation essentially (Usyk et al.
2000).
The simulation using a constitutive law with a relatively
high effective compressibility showed that within the
framework of our model, systolic interposition of fibres
can artificially be modelled by increasing compressibility
of the material. Comparison of tables 1– 3 with tables 4–
6, as well as figures 4 – 6 with figures 12 –14, respectively,
show that assumption of compressibility of the myocar-
dium increases ejection fraction. This increase was
provided essentially by increasing circumferential and
longitudinal shortening, but it was accompanied by
reduction of other strain components such that important
measures like wall thickening and fibre shortening were
on their limits of the measured experimental range. This
emphasises that the assumption of compressibility cannot
replace the effect of fibre rearrangement completely.
F. Dorri et al.
Measurements (Sabbah et al. 1981, Waldman et al. 1985,
Rademakers et al. 1994) showed furthermore that strain
components, especially radial strain, in the sub-endo-
cardial region increases considerably which is a direct
result of gliding or fibre rearrangement. The decreasing
strain components in our model in sub-endo-cardial areas,
shown in figures 4 –6, 12– 14, can be attributed to the fact
that local curvature increases from epi-cardium to endo-
cardium and, as a result, the elements of the mesh in sub-
endo-cardial region resist deformation more than the
elements in sub-epi-cardial region.
7.3 Comparison with other models
During the last decades a number of FE models have been
put forward to assess cardiodynamics theoretically and to
calculate the stress and strain distribution in the
ventricular wall. These models have mostly analysed the
end-diastolic state and were based on idealised and
smoothed geometrical representations, furthermore, ani-
mal hearts (dogs, pigs, and rats) were often modelled
because available measurements are more abundant than
 FE model of the human left ventricular systole 339

Figure 14. Variation of radial ðERR Þ, circumferential ðECC Þ and longitudinal ðELL Þ components of elastic strain (^0.07) in septal, anterior, lateral and
inferior of apical area from epi-cardium to endo-cardium ðK ¼ 600 kPaÞ.

in humans. A typical feature common to many models of
systolic contraction (Horowitz et al. 1986, Huyghe et al.
1992, McCulloch et al. 1992, Bovendeerd et al. 1992,
1994, Guccione and McCulloch 1993, Guccione et al.
1995, Nash and Hunter 2000, Usyk et al. 2000) is the
application of a symmetric geometry such as a thick
ellipsoid of revolution. A local cardiac-specific coordinate
system could then be defined with the help of the
symmetry properties in the radial, circumferential and
longitudinal directions. As a further simplification, the
fibre-specific coordinate system xF, yS, zN was defined
such that fibres were assumed as parallel to the epi-cardial
surface. Another assumption was that active stresses
develop only along tangential fibre directions. These

Table 6. Estimated average of radial (ERR), circumferential (ECC) and longitudinal (ELL) strains ðK ¼ 600 kPaÞ.

Strain Septal Anterior Lateral Inferior

ERR
Basal 0.40 ^ 0.07 0.33 ^ 0.07 0.33 ^ 0.07 0.27 ^ 0.07
Equatorial 0.53 ^ 0.07 0.33 ^ 0.07 0.27 ^ 0.07 0.33 ^ 0.07
Apical 0.60 ^ 0.07 0.20 ^ 0.07 0.20 ^ 0.07 0.47 ^ 0.07
ECC
Basal 20.07 ^ 0.07 20.13 ^ 0.07 20.07 ^ 0.07 20.07 ^ 0.07
Equatorial 20.13 ^ 0.07 20.13 ^ 0.07 20.07 ^ 0.07 20.20 ^ 0.07
Apical 20.13 ^ 0.07 20.07 ^ 0.07 20.07 ^ 0.07 20.20 ^ 0.07
ELL
Basal 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07
Equatorial 20.13 ^ 0.07 20.13 ^ 0.07 20.07 ^ 0.07 20.07 ^ 0.07
Apical 20.20 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07 20.13 ^ 0.07
340 F. Dorri et al.
models were mostly evaluated with experimental results
of strain using epi-cardial markers at a limited number of
measurement sites (Guccione et al. 1995, Nash and Hunter
2000, Usyk et al. 2000). Yet, some of these models could
not represent the global deformation of the LV correctly:
A low ejection fraction was reported in one case
(Guccione et al. 1995) (in some other models no ejection
fraction is given), furthermore longitudinal elongation
instead of shortening was predicted (Bovendeerd et al.
1992, Guccione et al. 1995). An accurate comparison of
the results obtained with such idealised models with our
calculations, which were performed on an unsymmetrical
geometry including non surface-parallel fibres and a more
advanced active stress tensor is not straightforward.
Nevertheless, as far as comparisons were possible, general
agreement was observed.
The development of cross-fibre stresses was studied
experimentally (Lin and Yin 1998). These measurements
along with other observations suggest that fibre, cross-
fibre and shear stresses develop simultaneously in the
myocardium and should be taken into account for the
modelling of systolic contraction (Mazhari and McCul-
loch 1999, Usyk et al. 2000). Yet, the configuration of an
active stress tensor depends on the reference coordinate
system; in an appropriately defined local coordinate
system an active stress tensor can be implemented with
components corresponding to the local fibre field. In our
mathematical model we furthermore concluded that
S0active ðS; SÞ should be negligible in comparison with the
(F, F) and (N, N) components because an active
compressive (S, S) component would counteract wall
thickening and longitudinal shortening, apart from that,
there is no experimental support for the existence of
tensile (S, S) active components.
In this work we attempted to determine deformation
patterns of the LV by inverting the modelling process, i.e.
to extrapolate stresses from deformations rather than
determining deformations from assumed fibre stresses.
Given the relation between the heart muscle’s oxygen
consumption and wall stress, the clinical intention is to
evaluate the cardiac load and work performed by the heart.
The ultimate goal of our work therefore consists of an
estimation of wall stresses from a known deformation
pattern. Our results indicate that this goal might
successfully be reached; however, a more advanced
constitutive description of the heart muscle tissue is
necessary. Nevertheless, our model included essential
anatomical details of a human heart, namely surface
irregularities and a realistic, albeit individual fibre
orientation pattern. As expected, the analysis reveals that
a realistic geometrical model along with a typical fibre
arrangement leads to considerable inhomogeneities in the
systolic deformation pattern. This result holds indepen-
dent of further features influencing the constitutive
behaviour of cardiac tissue. Our data support the
assumption that accurate measurement of local systolic
wall thickening, twisting and longitudinal shortening is of
diagnostic value in so far as any pathologic fibre
arrangement leads to distinct abnormalities in the
ventricular deformation pattern. The interpretation of
such abnormalities can essentially be improved by
applying mathematical approaches. Even when assuming
that the ventricular fibre architecture is optimal with
respect to normal cardiac performance (Rijcken et al.
1997, 1999), this is at best a thumb rule which holds for a
healthy heart, yet, in case of pathologies such as fibrosis,
infarction or cardiomyopathy the fibre field must be
expected to deviate significantly from an optimal pattern.
References
T. Arts, P.C. Veenstra and R.S. Reneman, “Epicardial deformation and
left ventricular wall mechanics during ejection in the dog”, Am.
J. Physiol., 243, pp. H379–H390, 1982.
P.H. Bovendeerd, T. Arts, J.M. Huyghe, D.H. van Campen and R.S.
Reneman, “Dependence of local left ventricular wall mechanics on
myocardial fiber orientation: a model study”, J. Biomech., 25,
pp. 1129–1140, 1992.
P.H. Bovendeerd, J.M. Huyghe, T. Arts, D.H. van Campen and R.S.
Reneman, “Influence of endocardial–epicardial crossover of muscle
fibers on left ventricular wall mechanics”, J. Biomech., 27,
pp. 941–951, 1994.
P.H. Bovendeerd, T. Arts, T. Delhaas, J.M. Huyghe, D.H. van Campen
and R.S. Reneman, “Regional wall mechanics in the ischemic left
ventricle: numerical modeling and dog experiments”, Am. J. Physiol.,
270, pp. H398–H410, 1996.
R.S. Chadwick, “Mechanics of the left ventricle”, Biophys. J., 39,
pp. 279–288, 1982.
R.S. Chadwick, J. Ohayon and M. Lewkowicz, “Wall-thickness and
midwall-radius variations in ventricular mechanics”, Proc. Natl.
Acad. Sci. USA, 86, pp. 2996–2999, 1989.
C.J. Chuong and Y.C. Fung, Frontiers in Biomechanics, G.W. Schmid-
Schönbein, S.L.-Y. Woo and B.W. Zweifach, Eds., New York, NY:
Springer-Verlag, 1986, pp. 117 –129.
K.D. Costa, Y. Takayama, A.D. McCulloch and J.W. Covell, “Laminar
fiber architecture and three-dimensional systolic mechanics in canine
ventricular myocardium”, Am. J. Physiol., 276, pp. H595–H607,
1999.
C.W. Cryer, H. Navidi-Kasmai, P.P. Lunkenheimer and K. Redmann,
“Computation of the alignment of myocardial contractile pathways
using a magnetic tablet and an optical method”, Technol. Health
Care, 5, pp. 79 –93, 1997.
F. Dorri, “A finite element model of the human”, 2004, PhD Thesis, Swiss
Federal Institute of Technology Zurich.
J.G. Dumesnil and R.M. Shoucri, “Quantitative relationships between left
ventricular ejection and wall thickening and geometry”, J. Appl.
Physiol., 70, pp. 48 –54, 1991.
H. Feneis, “Das Gefüge des herzmuskels bei systole und diastole”,
Morphol. Jahrb., 89, pp. 371–406, 1943.
J.M. Guccione and A.D. McCulloch, “Mechanics of active contraction in
cardiac muscle: Part I—constitutive relations for fiber stress that
describe deactivation”, J. Biomech. Eng., 115, pp. 72–81, 1993.
J.M. Guccione, A.D. McCulloch and L.K. Waldman, “Passive material
properties of intact ventricular myocardium determined from a
cylindrical model”, J. Biomech. Eng., 113, pp. 42–55, 1991.
J.M. Guccione, K.D. Costa and A.D. McCulloch, “Finite element stress
analysis of left ventricular mechanics in the beating dog heart”,
J. Biomech., 28, pp. 1167–1177, 1995.
J.M. Guccione, S.M. Moonly, P. Moustakidis, K.D. Costa, M.J. Moulton,
M.B. Ratcliffe and M.K. Pasque, “Mechanism underlying mechanical
dysfunction in the border zone of left ventricular aneurysm: a finite
element model study”, Ann. Thorac. Surg., 71, pp. 654–662, 2001a.
J.M. Guccione, S.M. Moonly, A.W. Wallace and M.B. Ratcliffe,
“Residual stress produced by ventricular volume reduction surgery
has little effect on ventricular function and mechanics: a finite
element model study”, J. Thorac. Cardiovasc. Surg., 122,
pp. 592–599, 2001b.
P.J. Hanley, A.A. Young, I.J. LeGrice, S.G. Edgar and D.S. Loiselle, “3-
Dimensional configuration of perimysial collagen fibres in rat cardiac
 FE model of the human left ventricular systole
muscle at resting and extended sarcomere lengths”, J. Physiol., 517,
pp. 831–837, 1999.
K.B. Harrington, F. Rodriguez, A. Cheng, F. Langer, H. Ashikaga, G.T.
Daughters, J.C. Criscione, N.B. Ingels and D.C. Miller, “Direct
measurement of transmural laminar architecture in the anterolateral
wall of the ovine left ventricle: new implications for wall thickening
mechanics”, Am. J. Physiol. Heart Circ. Physiol., 288,
pp. H1324–H1330, 2005.
A. Horowitz, M. Perl, S. Sideman and E. Ritman, “Comprehensive model
for the simulation of left ventricle mechanics. Part 2. Implementation
and results analysis”, Med. Biol. Eng. Comput., 24, pp. 150– 156,
1986.
W. Hort, “Makroskopische und mikrometrische Untersuchungen am
Myokard verschieden stark gefüllter linker Kammern”, Virchows.
Arch. Pathol. Anat. Physiol., 333, pp. 569– 581, 1960.
J.D. Humphrey, R.K. Strumpf and F.C. Yin, “Determination of a
constitutive relation for passive myocardium: II. Parameter
estimation”, J. Biomech. Eng., 112, pp. 340 –346, 1990a.
J.D. Humphrey, R.K. Strumpf and F.C. Yin, “Determination of a
constitutive relation for passive myocardium: I. A new functional
form”, J. Biomech. Eng., 112, pp. 333 –339, 1990b.
P.J. Hunter, P.M. Nielsen, B.H. Smaill, I.J. LeGrice and I.W. Hunter, “An
anatomical heart model with applications to myocardial activation
and ventricular mechanics”, Crit. Rev. Biomed. Eng., 20,
pp. 403–426, 1992.
P.J. Hunter, A.D. McCulloch and H.E. Keurs, “Modelling the mechanical
properties of cardiac muscle”, Prog. Biophys. Mol. Biol., 69, pp.
289 –331, 1998.
J.M. Huyghe, T. Arts, D.H. van Campen and R.S. Reneman, “Porous
medium finite element model of the beating left ventricle”, Am.
J. Physiol., 262, pp. H1256–H1267, 1992.
N.B. Ingels, “Myocardial fiber architecture and left ventricular function”,
Technol. Health Care, 5, pp. 45–52, 1997.
I.J. LeGrice, B.H. Smaill, L.Z. Chai, S.G. Edgar, J.B. Gavin and P.J.
Hunter, “Laminar structure of the heart: ventricular myocyte
arrangement and connective tissue architecture in the dog”, Am.
J. Physiol., 269, pp. H571 –H582, 1995.
I.J. LeGrice, P.J. Hunter and B.H. Smaill, “Laminar structure of the heart:
a mathematical model”, Am. J. Physiol., 272, pp. H2466–H2476,
1997.
D.H.S. Lin and F.C. Yin, “A multiaxial constitutive law for mammalian
left ventricular myocardium in steady-state barium contracture or
tetanus”, J. Biomech. Eng., 120, pp. 504–517, 1998.
P.P. Lunkenheimer, K. Redmann, K.-H. Dietl, C. Cryer, K.-D. Richter,
W.F. Whimster and P. Niederer, “The heart’s fibre alignment assessed
by comparing two digitizing systems. Methodological investigation
into the inclination angle toward wall thickness”, Technol. Health
Care, 5, pp. 65–77, 1997.
P.P. Lunkenheimer, K. Redmann, J. Florek, U. Fassnacht, C.W. Cryer, F.
Wübbeling, P. Niederer and R.H. Anderson, “The forces generated
within the musculature of the left ventricular wall”, Heart, 90,
pp. 200–207, 2004.
G.A. MacGowan, E.P. Shapiro, H. Azhari, C.O. Siu, P.S. Hees, G.M.
Hutchins, J.L. Weiss and F.E. Rademakers, “Noninvasive measure-
ment of shortening in the fiber and cross-fiber directions in the normal
human left ventricle and in idiopathic dilated cardiomyopathy”,
Circulation, 96, pp. 535 –541, 1997.
F.P. Mall, “On the muscular architecture of the ventricles of the human
heart”, Am. J. Anat., 11, pp. 211– 266, 1911.
R. Mazhari and A.D. McCulloch, Proceeding of ASME 4th Summer
Bioengineering Conference, ASME, ed., 1999, pp. 43–44.
A.D. McCulloch, L.K. Waldman and J.M. Rogers, “Large-scale finite
element analysis of the beating heart”, Crit. Rev. Biomed. Eng., 20,
pp. 427–449, 1992.
C.C. Moore, C.H. Lugo-Olivieri, E.R. McVeigh and E.A. Zerhouni,
“Three-dimentional systolic strain patterns in the normal human left
ventricle: characterization with tagged MR imaging”, Radiology,
214, pp. 453– 466, 2000.
P. Moustakidis, R. Pyo and R.P. Scheri, “Stress distribution analysis of an
experimental model of the left ventricular aneurysm using magnetic
resonance imaging and finite element modeling”, Circulation, 100,
p. 644, 1999.
M.P. Nash and P.J. Hunter, “Computational mechanics of the heart”,
J. Elasticity, 61, pp. 113 –141, 2000.
P.M. Nielsen, I.J. Le Grice, B.H. Smaill and P.J. Hunter, “Mathematical
model of geometry and fibrous structure of the heart”, Am. J. Physiol.,
260, pp. H1365–H1378, 1991.
341
V.P. Novak, F.C. Yin and J.D. Humphrey, “Regional mechanical
properties of passive myocardium”, J. Biomech., 27, pp. 403 –412,
1994.
F.E. Rademakers, W.J. Rogers, W.H. Guier, G.M. Hutchins, C.O. Siu,
M.L. Weisfeldt, J.L. Weiss and E.P. Shapiro, “Relation of regional
cross-fiber shortening to wall thickening in the intact heart. Three-
dimensional strain analysis by NMR tagging”, Circulation, 89,
pp. 1174–1182, 1994.
J. Rijcken, P.H. Bovendeerd, A.J. Schoofs, D.H. van Campen and T. Arts,
“Optimization of cardiac fiber orientation for homogeneous fiber
strain at beginning of ejection”, J. Biomech., 30, pp. 1041– 1049,
1997.
J. Rijcken, P.H. Bovendeerd, A.J. Schoofs, D.H. van Campen and T. Arts,
“Optimization of cardiac fiber orientation for homogeneous fiber
strain during ejection”, Ann. Biomed. Eng., 27, pp. 289–297, 1999.
R.S. Rivlin and D.W. Saunders, “Large elastic deformations of isotropic
materials, VII experiments on the deformation of rubber”, Philos.
Trans. R. soc., A243, pp. 251–288, 1951.
D.F. Rogers, p An q Introduction to NURBS with Historical
Perspective, San Francisco, CA: Morgan Kaufmann, 2001.
H.N. Sabbah, M. Marzilli and P.D. Stein, “The relative role of
subendocardium and subepicardium in left ventricular mechanics”,
Am. J. Physiol., 240, pp. H920–H926, 1981.
R.M. Shoucri, “Active and passive stresses in the myocardium”, Am.
J. Physiol. Heart Circ. Physiol., 279, pp. H2519–H2528, 2000.
A.J. Sinusas, X. Papademetris, R.T. Constable, D.P. Dione, M.D. Slade, P.
Shi and J.S. Duncan, “Quantification of 3-D regional myocardial
deformation: shape-based analysis of magnetic resonance images”,
Am. J. Physiol. Heart Circ. Physiol., 281, pp. H698–H714, 2001.
A.J.M. Spencer, Deformations of Fibre-Reinforced Materials, Oxford:
Clarendon Press, 1972.
A.J.M. Spencer, Continuum Theory of the Mechanics of Fibre-Reinforced
Composites, Wien: Springer, 1984.
H.M. Spotnitz, W.D. Spotnitz, T.S. Cottrell, D. Spiro and E.H.
Sonnenblick, “Cellular basis for volume related wall thickness
changes in the rat left ventricle”, J. Mol. Cell Cardiol., 6,
pp. 317–331, 1974.
C. Stevens, E. Remme, I. LeGrice and P. Hunter, “Ventricular mechanics
in diastole: material parameter sensitivity”, J. Biomech., 36,
pp. 737–748, 2003.
D.D. Streeter, JR and D.L. Basset, “An engineering analysis of
myocardial fibre orientation in pig’s left ventricle in systole”, Anat.
Rec., 155, pp. 503 –511, 1966.
D.D. Streeter, JR, D.P.P. Spotnitz, J.J. Ross and E.H. Sonnenblick, “Fiber
orientation in the canine left ventricle during diastole and systole”,
Circ. Res., 24, pp. 339–347, 1969.
M. Stuber, Quantification of the Human Heart Wall Motion by Ultra-fast
Magnetic Resonance Myocardial Tagging Techniques, Zurich: Swiss
Federal Institute of Technology, 1997.
M. Stuber, M.B. Scheidegger, S.E. Fischer, E. Nagel, F. Steinemann,
O.M. Hess and P. Boesiger, “Alternations in the local myocardial
motion pattern in patients suffering from pressure overload due to
aortic stenosis”, Circulation, 100, pp. 361–368, 1999.
F. Torrent-Guasp, W.F. Whimster and K. Redmann, “A silicone rubber
mould of the heart”, Technol. Health Care, 5, pp. 13 –20, 1997.
F. Torrent-Guasp, G.D. Buckberg, C. Clemente, J.L. Cox, H.C. Coghlan
and M. Gharib, “The structure and function of the helical heart and its
buttress wrapping. I. The normal macroscopic structure of the heart”,
Semin. Thorac. Cardiovasc. Surg., 13, pp. 301 –319, 2001.
W.Y. Tseng, T.G. Reese, R.M. Weisskoff, T.J. Brady and V.J. Wedeen,
“Myocardial fiber shortening in humans: initial results of MR
imaging”, Radiology, 216, pp. 128– 139, 2000.
T.P. Usyk, R. Mazhari and A.D. McCulloch, “Effect of laminar
orthotropic myofiber architecture on regional stress and strain in the
canine left ventricle”, J. Elasticity, 61, pp. 143 –164, 2000.
L.K. Waldman, Y.C. Fung and J.W. Covell, “Transmural myocardial
deformation in the canine left ventricle. Normal in vivo three-
dimensional finite strains”, Circ. Res., 57, pp. 152–163, 1985.
J.A. Weiss, “A constitutive model and finite element representation for
transversely isotropic soft tissue”, PhD Thesis, The University of
Utha. 1994.
F.C. Yin, C.C. Chan and R.M. Judd, “Compressibility of perfused passive
myocardium”, Am. J. Physiol. Heart Circ. Physiol., 40,
pp. H1864–H1870, 1996.
A.A. Young, I. LeGrice, M.A. Young and B.H. Smaill, “Extended
confocal microscopy of myocardial laminae and collagen network”,
J. Microsc., 192, pp. 139–150, 1998.