ARTICLE IN PRESS

Journal of Biomechanics 40 (2007) 693696

Short communication

www.elsevier.com/locate/jbiomech
www.JBiomech.com

Inverse elastostatic stress analysis in pre-deformed biological

structures: Demonstration using abdominal aortic aneurysms
Jia Lua,, Xianlian Zhoua, Madhavan L. Raghavanb
a

Department of Mechanical and Industrial Engineering, Center for Computer Aided Design, The University of Iowa, Iowa City, IA 52242-1527, USA
b
Department of Biomedical Engineering, The University of Iowa, Iowa City, IA 52242, USA
Accepted 16 January 2006

Abstract
In stress analysis of membrane-like biological structures, the geometry constructed from in vivo image, which often corresponds
to a deformed state, is routinely taken as the initial stress-free geometry. In this paper, we show that this limitation can be
completely removed using an inverse elastostatic approach, namely, a method for nding the initial geometry of an elastic body from
a given deformed state. We demonstrate the utility of the inverse approach using a patient-specic abdominal aortic aneurysm
model, and identify the scope of error in stress estimation in the conventional approach within a realistic range of material
parameter variations.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Inverse elastostatics; Inverse nite element method; Aortic aneurysm; Patient-specic analysis

1. Introduction
A recent trend in aneurysm stress analysis is to utilize
patient-specic models with aneurysm geometry constructed from diagnostic images (Wang et al., 2002;
Raghavan et al., 2000; Thubrikar et al., 2001; Fillinger
et al., 2002, 2003). The image-based geometry, however,
corresponds to a pressurized state rather than the initial
stress-free state. A common approach is to take the in
vivo geometry as the initial conguration, and proceed
to nd the stress using the standard forward nite
element analysis. In other words, the pre-deformation
due to in vivo pressure is not accounted for. This is
certainly a limitation in the current analyses; the induced
error could be nontrivial if the pre-deformation is
signicant, as what could happen under hypertension
conditions.
In this communication, we demonstrate that the
assumption made to the initial conguration is comCorresponding author. Tel.: +1 319 3356405; fax: +1 319 3355669.

E-mail address: jia-lu@uiowa.edu (J. Lu).

0021-9290/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jbiomech.2006.01.015

pletely unnecessary. The initial stress-free conguration

of an elastic body can be determined from a
given deformed state. The feasibility of nding the
initial geometry is a unique feature of elastostatic
problem. Inverse solutions have been reported in the
literature. For example, Green and Zerna (1954)
provided solutions to uniform expansion of thick
spherical or cylindrical shell that allow either the
deformed or the undeformed conguration to be
found knowing the other. Recently, Govindjee and
Mihalic (1996, 1998) and Yamada (1995) proposed a
procedure of solving the inverse problem via the
nite element method (FEM). Govindjees formulation,
in particular, involves only minor changes to the
elements designed for forward analysis. Here, we
implement a material model reported specically for
abdominal aortic aneurysm (AAA) in Govindjees
inverse framework. We use the inverse method on a
representative patient-specic AAA model and demonstrate its ability to accurately determine the load-free
geometry. We further investigate the scope of error in
the conventional forward approach within a realistic

ARTICLE IN PRESS
J. Lu et al. / Journal of Biomechanics 40 (2007) 693696

694

range of AAA wall material behavior in the patient

population.

2. Methods
The inverse formulation by Govindjee et al. starts
from the Eulerian weak form of the static equilibrium
problem
Z
Z
Z
ti Zi da,
sij Zi;j dv
rbi Zi dv
(1)
O

qOt

where O is the given current conguration, r is the

Cauchy stress, b is the body force, t is the prescribed
surface traction and qOt is the boundary where the
traction is applied. The solution for the initial geometry
is facilitated via the introduction of the inverse motion
U : X Ux, which is the mathematical inverse of the
forward motion u : x uX, the primary variable in a
forward problem. The cornerstone in Govindjees
approach is to reparameterize the Cauchy stress, which
is normally a function of the forward deformation
gradient F, in inverse deformation gradient
f:qU=qx F1 . In this manner, the weak form is
formulated in the inverse motion, which can be solved to
nd the initial geometry. In implementation, this scheme
results in a FEM formulation that involves minimum
change to the standard element. The details of
implementation, including a mixed treatment of the
pressure eld, are described in Govindjee and Mihalic
(1996, 1998).
We implemented a hyperelastic material model
reported specically for abdominal aortic aneurysms
(Raghavan and Vorp, 2000) in the inverse nite element
framework. The model is specied by the energy
function
W aI 1  2 log J  3 bI 1  32 klog J2 .

1
1 1
where in index notation, Ic1 ijkl 12 c1
ik cjl cil cjk ,
and  means the standard tensor product. The inverse
procedure and this material model have been implemented in an in-house version of a nonlinear FEM code
(FEAP) originally developed at University of California
Berkeley (Taylor, 2003).

3. Results
The surface mesh shown in Fig. 1 was constructed from
computed tomography (CT) images of an abdominal
aortic aneurysm. The triangular surface mesh was
extruded outward to create the 3D nite element model
with one layer of elements in the thickness direction. The
wall thickness was assumed to be 1.9 mm. We employed
the material model (2) with the population mean material
parameters a 17:39 N=cm2 and b 188:08 N=cm2 . The
penalty parameter was set to k 100 000 N=cm2 .
Assuming that the CT-reconstructed geometry corresponds to the deformed state at 100 mmHg mean aortic
pressure, we predicted the load-free geometry and the
stress in this state. The load-free geometry computed
using the inverse approach is shown in Fig. 1 superimposed on the CT-reconstructed geometry. As a
validation, the in vivo pressure was subsequently applied
to the recovered load-free conguration and a forward
stress analysis was performed. This forward analysis
exactly recovered the in vivo shape, and resulted in a
stress distribution that exactly matches those from the
inverse approach.
It is of practical interest in assessing the error in the
conventional approach. For this, we performed forward
stress analysis assuming the CT-reconstructed geometry

(2)

Here, I 1 tr F F, J det F, and a; b; k are material

constants. Comparing to the original energy function in
Raghavan and Vorp (2000), a volumetric penalty term is
augmented for enforcing quasi-incompressibility constraint. The Cauchy stress in inverse kinematics can be
obtained by converting the normal forward stress
function, and this gives
r 2jac1  I 4jbtr c1  3c1  2kj log jI,

(3)

where c f f is the inverse deformation tensor, I is the

second-order identity tensor, and j det f. The material
tangent tensor, which is needed a NewtonRaphsonbased iterative solver, is dened as the derivative of r
relative to c and is
2

qr
 j4a 8btr c1  3Ic1 8bc1  c2
qc
2kI  c1  r  c1 ,

(a)

(b)

Fig. 1. Views of the stress-free conguration of a pressurized

aneurysm. The predicted initial geometry is depicted in solid and is
visibly smaller than the in vivo shape by surface mesh. In the FE
model, the nodes at the top and bottom cross sections are xed in zdirection, in addition to a set of six constraints necessary to eliminate
the rigid body motion.

ARTICLE IN PRESS
J. Lu et al. / Journal of Biomechanics 40 (2007) 693696

695

24.80
22.73
20.65
18.58
16.51
14.44
12.36
10.29
8.22
6.15
4.07
2.00

21.58
19.80
18.02
16.24
14.46
12.68
10.90
9.12
7.34
5.56
3.78
2.00

(a)

(b)

Fig. 2. von Mises stress at 100 mmHg arterial pressure. (a) Inverse analysis; (b) Forward analysis using the in vivo conguration as reference.

as itself stress-free, as has been done in this eld so far,

and estimated the stress distribution under a pressure
100 mmHg. Fig. 2 shows the stress distribution from the
inverse approach and the conventional forward approach. The stress distributions have remarkably similar
patterns but the latter is elevated uniformly. To further
investigate the scope of error within a realistic range of
material parameter variation, we performed repetitive
analyses by varying the parameters a and b within the
reported 95% condence interval (Raghavan and Vorp,
2000). Specically, we assigned a to the upper and lower
limits in the condence domain while holding b xed at
the population mean, and vice versa. The stress distributions at 100 mmHg arterial pressure were computed using
both approaches. The maximum von Mises stresses are
listed in Table 1. The conventional approach for this
representative AAA was found to result in 13.6%17.1%
error within the given range of parametric variation.
Although the error does not uctuate signicantly over
the considered range, one can clearly see that error
increases as the tissue becomes more compliant.

4. Discussion and conclusion

In this communication, we demonstrated that the
load-free conguration of an AAA can be exactly
recovered using the inverse elastostatic approach, and
thus conclusively addressed the issue of nding the
reference conguration in imaging-based patient-specic
stress analysis. We found that the conventional
approach over-predicts the stress, and the error magnies as the material becomes more compliant. This is
expected, because a compliant wall would result in
greater differences between the load-free and in vivo
geometries, which is precisely what the conventional
approach neglects.

Table 1
Comparison of the maximum von Mises stress at 100 mmHg arterial
pressure predicted from the present approach smax and the
conventional approach smax units: N=cm2
(a; b)

(14.40,
(20.40,
(17.39,
(17.39,
(17.39,

188.08)
188.08)
188.08)
115.20)
261.00)

smax
Present

smax
Conventional

Error (%)
smax  smax
 100
smax

21.9
21.4
21.6
21.1
21.9

25.4
24.3
24.8
24.7
24.9

16.0
13.6
14.8
17.1
13.7

There are some limitations in the present inverse

methodology in AAA applications. First, the inverse
methodology as we have described here exists on the
premise that the residual stress is indeed negligible. If
residual stress was found to be signicant, further
modications to the approach not described in this
report would be needed. Second, the use of a single layer
of prism elements can affect stress estimation slightly.
However, because all comparisons used the same prism
elements, they should remain reasonably valid.
Although we have used AAA as a test case for
demonstration, the inverse approach is applicable to a
variety of biological structures that are eternally loaded
and could be treated as residual stress free. Membranelike structures such as brain aneurysms, pericardium,
urinary bladder, cell membranes are potential application areas for inverse elastostatics.

Acknowledgments
We would like to thank Dr. Mark F. Fillinger
(Dartmouth-Hitchcock Medical Center) for providing

ARTICLE IN PRESS
696

J. Lu et al. / Journal of Biomechanics 40 (2007) 693696

the AAA surface mesh. We also thank Mr. Wenyi Hou

and Mr. Weixue Yang for assistance in 3D mesh
generation. The rst author (J. L.) acknowledges the
support by the National Science Foundation Grant
CMS 03-48194.
References
Fillinger, M.F., Raghavan, M.L., Marra, S.P., Cronenwett, J.L.,
Kennedy, F.E., 2002. In vivo analysis of mechanical wall stress and
abdominal aortic aneurysm rupture risk. Journal of Vascular
Surgery 36, 589597.
Fillinger, M.F., Marra, S.P., Raghavan, M.L., Kennedy, F.E., 2003.
Prediction of rupture risk in abdominal aortic aneurysm during
observation: wall stress versus diameter. Journal of Vascular
Surgery 37, 724732.
Govindjee, S., Mihalic, P.A., 1996. Computational methods for inverse
nite elastostatics. Computer Methods in Applied Mechanics and
Engineering 136, 4757.
Govindjee, S., Mihalic, P.A., 1998. Computational methods for inverse
deformations in quasi-incompressible nite elasticity. International
Journal for Numerical Methods in Engineering 43, 821838.

Green, A.E., Zerna, W., 1954. Theoretical Elasticity. Clarendon Press,

Oxford.
Raghavan, M.L., Vorp, D.A., 2000. Toward a biomechanical
tool to evaluate rupture potential of abdominal aortic
aneurysm: identication of a nite strain constitutive model and
evaluation of its applicability. Journal of Biomechanics 33,
475482.
Raghavan, M.L., Vorp, D.A., Federle, M.P., Makaroun, M.S.,
Webster, M.W., 2000. Wall stress distribution on three-dimensionally reconstructed models of human abdominal aortic aneurysm.
Journal of Vascular Surgery 31, 760769.
Taylor, R.L., 2003. FEAP User manual: v7.5. Technical Report,
Department of Civil and Environmental Engineering, University of
California, Berkeley.
Thubrikar, M.J., Al-Soudi, J., Robicsek, F., 2001. Wall stress studies
of abdominal aortic aneurysm in a clinical model. Annals of
Vascular Surgery 15, 355366.
Wang, D.H., Makaroun, M.S., Webster, M.W., Vorp, D.A., 2002.
Effect of intraluminal thrombus on wall stress in patient-specic
models of abdominal aortic aneurysm. Journal of Vascular Surgery
36, 17.
Yamada, T., 1995. Finite element procedure of initial shape
determination for rubber-like materials. Technical Report No.
20, Res. Lab. Eng. Mat. Tokyo Institute of Technology.