See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323576927

Skin mechanical properties and modeling: A review

Article in Proceedings of the Institution of Mechanical Engineers Part H Journal of Engineering in Medicine · March 2018
DOI: 10.1177/0954411918759801

CITATIONS READS

399 33,226

2 authors:

Hamed Joodaki Matthew B Panzer

University of Virginia University of Virginia
16 PUBLICATIONS 637 CITATIONS 127 PUBLICATIONS 3,716 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Hamed Joodaki on 11 March 2018.

The user has requested enhancement of the downloaded file.

Review Article

Proc IMechE Part H:

J Engineering in Medicine
1–21
Skin mechanical properties and Ó IMechE 2018
Reprints and permissions:
modeling: A review sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0954411918759801
journals.sagepub.com/home/pih

Hamed Joodaki and Matthew B Panzer

Abstract
The mechanical properties of the skin are important for various applications. Numerous tests have been conducted to
characterize the mechanical behavior of this tissue, and this article presents a review on different experimental methods
used. A discussion on the general mechanical behavior of the skin, including nonlinearity, viscoelasticity, anisotropy, load-
ing history dependency, failure properties, and aging effects, is presented. Finally, commonly used constitutive models for
simulating the mechanical response of skin are discussed in the context of representing the empirically observed
behavior.

Keywords
Literature review, skin, mechanical properties characterization, constitutive modeling, flesh

Date received: 2 December 2016; accepted: 23 January 2018

Introduction organized structure consisting of three major layers,

called the epidermis, dermis, and hypodermis or subcu-
Importance tis (Figure 1). The topmost layer of the skin is called
The skin, which is the largest organ in surface area, has epidermis which is approximately 75–150 mm in thick-
several functions including the protection from environ- ness11 and is made of four sub-layers, namely, stratum
ment, temperature regulation, and adaptation of con- spinosum, stratum basale, stratum granulosum, and
tours of the body during movement. This tissue consists stratum corneum.
of several layers, each layer made of several compo- The second layer is the dermis which is a dense
nents, which makes it mechanically complicated. The fibrous tissue. The dermis and the epidermis together
mechanical properties of the skin are of importance for are also referred to as cutis, which has a thickness of
various cosmetic, clinical, and biomechanical applica- 1.5–2.5 mm.13 The dermis is a moderately dense con-
tions. Cosmetic applications include the efficiency nective tissue which comprises three fibrous proteins,
assessment of cosmetic products in terms of emolliency collagen, elastin, and minute quantities of reticulin and
and hydration.1–3 Clinicians use the mechanical proper- a supporting matrix or ground substance. The collagen
ties of skin to better understand the laceration process, fibers contribute to 75% of the fat-free dry mass and
techniques in plastic surgery and skin closure, treat- 18%–30% of the volume of dermis.14 The collagen
ments for skin diseases, and the formulation of scar tis- fibers are within the ground substance and unconnected
sue.4,5 Finally, the biomechanical applications of skin along most of their length.15 The elastin fibers comprise
material properties include the design and production of 4% of the fat-free dry mass and 1% of the volume of
skin simulant materials for physical anthropomorphic the dermis.14 Elastin dictates the mechanical behavior
devices such as crash test dummies and surgical simula- of skin at small stresses and strains.16 Especially, the
tors.6,7 In addition to use with physical biomechanical elastin fibers are in charge of the recoiling mechanism
models, skin material properties are used in the develop-
ment and improvement of computational human body Department of Mechanical and Aerospace Engineering, University of
models for biomechanical simulation.8–10 Virginia, Charlottesville, VA, USA

Corresponding author:
Structure of skin Hamed Joodaki, Department of Mechanical and Aerospace Engineering,
University of Virginia, 4040 Lewis and Clark Dr., Charlottesville, VA
To understand the mechanical behavior of skin, its 22911, USA.
structure needs to be considered. The skin is an Email: [email protected]
2 Proc IMechE Part H: J Engineering in Medicine 00(0)

modulus, n Poisson’s ratio, and subscript i denoting the

indenter properties)

1 1  n2 1  n2i
= + ð3Þ
Er E Ei
Hertz formulation is based on the theory of linear
elasticity, and hence, it must be possible to define mea-
sures of stress and strain that satisfy a Hookean relation-
Figure 1. Schematic view of cross section of the human skin ship. Widely accepted definitions of indentation stress (or
showing the distinct layers.12 mean pressure, s ) and strain (e ) are given by22
P
s = ð4Þ
after deformation. At tensile strains around 30%, the pa2
a
undulated collagen fibers are straightened.5 e = 0:2 ð5Þ
Rsi
The third layer is the hypodermis (also called subcu-
taneous fat or subcutis), mainly made of adipose cells. The strain prefactor of 0.2 was empirically determined
The thickness of this layer varies notably over the sur- by Tabor23 and has since been verified by other investi-
face of the body.5,17 It functions as an insulating and gators.24–26
cushion layer and constitutes about 10% of the body The measurement of viscoelastic mechanical proper-
mass.18 ties can be done by the principle of elastic–viscoelastic
correspondence.21,27,28 In case of relaxation test, the
relaxation modulus is substituted for elastic modulus
Skin testing (E in equation (3)) and d is considered as a function of
In vivo time (deformation time history in the relaxation test).
Studies have exhibited that the measured mechanical
In vivo experiments provide data on tissue in its natural behavior depends on the indentation probe size.20,29
state, i.e., permeated with blood, and in a typical stress/ Crichton et al.20 did micro-indentation tests on the skin
strain state.19 In some of the in vivo experiments, the and found that the measured elastic modulus decreases
measured behavior is attributed to the dermis. with the increase in the diameter of used probe. They
However, as a result of connections between the vari- proposed that this may be a result of the spreading of
ous skin layers, it is difficult to isolate the contribution the stress over a larger area than the probe tip facili-
of the dermis from that of the epidermis and subcuta- tated by the hard corneocytes in the outermost skin
neous tissues, such as muscle and fat.5 Four main layers and softer living cell layers below. This will cause
methods used for in vivo testing of the skin are indenta- a greater effective radius than the actual probe tip
tion, torsion, tension, and suction. In what follows, radius.
each method is described briefly.
Torsion. In torsion tests, usually a constant rotation or
Indentation. In indentation experiments, a rigid indenter torque is applied to the skin via an intermediary disk.
(with a cylindrical, conic, or spherical indentation tip) Vlasblom30 and Sanders31 used an apparatus which
is used to apply a known deformation force to the skin. transmitted torque (8.3 3 1024 N m) of a coil moving in
Sub-micron atomic force microscopy indentation can a magnetic field under the influence of an electric cur-
be used to determine the mechanical properties of dif- rent bar to a circular disk (diameter 8.7 mm) that was
ferent layers.20 From the Hertz contact theory using a attached to the skin by a piece of adhesive tape sticky
spherical indenter, and assuming elastic behavior, there on both sides. The rotation of the skin resulted in an
is a mathematical relationship between indentation opposed moment and the coil moved until an equili-
depth, indentation force, and material properties20,21 brium position was reached. At given times, the torsion
angle was recorded by means of a light beam which was
4 1 3 reflected by a mirror attached to the aforementioned
P= Er R2si d2 ð1Þ
3 bar. Holding the applied moment constant, Sanders31
observed the creep response of the skin. The modulus
a2 of elasticity, E, was driven by the formula
d= ð2Þ
Rsi
2T(1 + n)
where P is the load applied giving the resultant com- E= ð6Þ
4hR2d u
pression, Rsi the radius of the spherical indenter, d the
displacement of the indenter, and a the contact radius where T is the torque, n the Poisson’s ratio, h the skin
of the indenter on the material being indented. Er is the thickness, Rd the radius of disk, and u the rotation
reduced modulus, where (with E being Young’s angle.
Joodaki and Panzer 3

Agache et al.32 and Leveque et al.33 conducted tor- However, it also has some disadvantages. The sample
sion test on human skin in vivo by transmitting the tor- is usually preserved between the excision and testing
que to the skin using a disk (25 mm diameter) covered time which might alter the mechanical properties of the
with a double band tape and surrounded by a guard tissue. For example, Caro-Bretelle et al.38 reported that
ring which was used to isolate a well-defined part of the the conservation of skin in a saline solution and freez-
skin under study. It was assured that there is constant ing without any caution alter the irreversible part of
skin contact by applying a certain pressure. The rota- the global mechanical behavior of the tissue. The other
tion of skin between the disk and guard ring was mea- disadvantage is that handling and holding the sample is
sured, and the elastic modulus was calculated using the tricky as the soft tissue tends to slip in the clamp.
following formula Moreover, due to the soft nature of the test material, it
is difficult to insert samples without subjecting them to
T
E= ð7Þ an axial load. The two methods of in vitro testing that
0:432phR1 R2 u are widely used are uniaxial and biaxial tensile tests.
where R1 and R2 are the disk and inner guard ring radii,
respectively. Biaxial tension. In biaxial tests, samples are cut into a
square shape, and the load is applied by means of
Tension. In tensile testing, the skin is mainly loaded par- hooks and strings to prevent development of shear
allel to its surface using two tabs which are attached to strains.39,40 These tests allow the skin to be stretched in
the skin. The attachment of the tabs to the skin may sig- both directions, similar to how it would be on the body.
nificantly influence the results as many of the double- However, this method of testing can be challenging—
sided adhesive tapes exhibit creep deformation. Rapidly the hooks affixed to the edges of the sample tend to tear
bonding cyanoacrylate adhesives are useful to avoid the skin. For the same reason, this method is not suit-
these effects.5,12 able for measuring the failure properties of the skin.

Suction. In suction testing, the skin is elevated by apply- Uniaxial tension. For uniaxial tensile test, skin samples
ing a partial vacuum using a circular aperture and the are cut into dog-bone shape and then they undergo ten-
deformation is quantified by optical or ultrasound sion while held by clamps.41–43 Tests in multiple direc-
devices. Two suction devices that are commercially tions would be required to determine the anisotropic
available and used by the researchers are Dermaflex behavior of the material.43–45 Specimen wrinkling can
and Cutometer.5,34–37 Numerous studies are done on also complicate uniaxial tests and cause one portion of
skin via suction method. In one of them, Diridollou the material to be loaded greater than the other.
et al.2 developed an equation that relates the suction Furthermore, to prevent the slippage of the sample
pressure applied, Pext (t), with the vertical displacement from the clamps, fixation techniques such as using
of the skin’s surface d(t) freeze clamp,46 serrated jaw clamp,47 wrapping the tis-
 sue in sandpaper or tissue paper at the ends,48–51 and
4d(t)  h E air dried ends48 should be recruited. However, uniaxial
Pext (t) =  
d(t)2 + r20 1  n tensile testing is the only feasible method to measure
"  !# !  the failure properties of soft tissue among the discussed
d(t)2 + r20 2  r0  d(t)
 arcsin  1 + s0 methods.
2r0 :d(t) d(t)2 + r20
ð8Þ
Mechanical behavior of skin
where h is the thickness of combined dermis and epider-
mis, r0 the aperture radius that indicates the measure-
Nonlinear stress–strain relationship
ment area of the skin, E the Young’s modulus, s0 the The mechanical response of skin tissue is highly non-
initial stress, and n the Poisson’s ratio. linear due to the makeup of its microstructural consti-
tuents (Figure 2).52 Under uniaxial tension, skin is
relatively soft, and much of the structural response of
In vitro the skin at low strain levels (Figure 2, Stage I) is carried
In vitro testing has a number of advantages over the in through the elastin components as the collagen fibers
vivo method. One advantage of this method is that the are slack (sometimes referred to as wavy) and non-
tissue sample can be tested under different loading val- load-bearing. When the skin is stretched to higher
ues and rates and the stress–strain curve can be easily strain levels (Figure 2, Stage II), the stiffness of the
obtained. Another advantage is that the tissue can be material increases rapidly as the collagen fibers are
tested until failure, and consequently the failure proper- recruited, straightened, and begin to carry the major
ties can be determined. Furthermore, it provides stress– part of the load.53 In the third phase, all collagen fibers
strain relationships that can be modeled and quantified are straight and the system has its highest stiffness
easily, since the boundary conditions are well-defined. (Figure 2, Stage III). By comparing the results of
4 Proc IMechE Part H: J Engineering in Medicine 00(0)

Figure 2. Stress–strain diagram for skin showing the different Figure 3. The engineering stress versus stretch ratio response
stages.52 Stage I: collagen fibers are still wavy and elastin fibers of pig skin to uniaxial compression at different rates and the
are the load-bearing components; Stage II: collagen fibers are corresponding Ogden hyperelastic models.59
gradually getting aligned and contribute to load-bearing; Stage III:
all the collagen fibers are aligned and the tissue has its highest
stiffness. model constants at each strain rate (the Ogden model is
explained in section ‘‘Ogden model’’).
Agache58 and Khatyr et al.60 showed in a creep test
testing on collagen and skin in high strains, it is shown that if a constant stress is imposed on the skin, the
that the collagen is the main support structure in the curve obtained can be decomposed into three phases
linear region of the stress–strain curve (Figure 2, Stage (Figure 4(a): the first phase corresponds to a purely
III).17,54 Ridge and Wright55 tracked this behavior to elastic deformation ee, the second phase of variable
higher strains and observed a decrease in the stiffness creep corresponds to the viscoelastic phase eve, and
of tissue after the stage in which all fibers are aligned the third phase corresponds to the constant creep
which is a result of gradual breaking of the collagen phase es. Piérard et al.61 conducted in vivo suction
fibers. tests on human forearm skin by applying a 500-mbar
Markenscoff and Yannas56 used a simple model of suction for 5 s followed by a stress-off time of 5 s and
the collagen fiber network in the dermis to calculate the measured the vertical skin deformation. The results
strain level at which the stiffness of skin abruptly characterized the maximum deformation (MD) and
increases. The collagen network was modeled as con- the residual deformation (RD) under this testing con-
sisting of small straight fibers of unit length attached to dition (Figure 4(b).
each other. The angle theta at attachment points was Skin also exhibits the relaxation behavior character-
considered as a random variable distributed with equal istic of viscoelastic materials. When stretched and held
probability between 0° and 180°. They predicted that in to a constant strain level, the stress within the skin
strain of e = 57%, all fibers will become straightened. material will decay over time (Figure 5). Karimi et al.62
Nevertheless, it should be noted that this conclusion is demonstrated the relaxation response of the rat back
based on the assumption that the directions of fibers and abdomen skin to the ramp and hold uniaxial strain
are completely random which is inconsistent with many of 40% after preparing the sample by imposing to 10
regions on the body due to Langer’s lines (see section cycles of preconditioning up to 30% strain and subse-
‘‘Anisotropy’’). quent 30% strain for 2 min as a physiological preload.

Viscoelasticity Anisotropy
Several studies have shown the strain rate dependency In vivo skin is normally in a state of tension. When a
of skin mechanical properties.57–59 Figure 3 shows the skin sample is excised from the body, it retracts and the
result of Shergold et al.’s59 study on pig skin that under- amount of retraction depends on the incision direction
went uniaxial compression with different strain rates. and site. This phenomenon was originally demonstrated
This figure illustrates the rate dependency of stress– by A.K. Langer. Langer identified lines of maximum
strain relationship for the skin. It is clear that fitting skin tension over the entire human body, which have
hyperelastic models to these curves, which do not take since been called Langer’s lines (Figure 6). The higher
into account the viscoelasticity, will result in different tensile forces along the Langer’s lines are due to the
Joodaki and Panzer 5

Figure 4. Skin creep curve under constant stress (a) schematic representation of three phases of skin creep test58 and
(b) deformation versus time curve obtained under the creep mode procedure.61

Figure 5. The response of rat back and abdomen skin to a

constant uniaxial strain of 40%.62 Figure 7. Uniaxial tensile test on human back skin parallel and
perpendicular to the Langer’s lines.66

vivo. Hence, when the skin is stretched in the direction

of Langer’s lines, it will extend only a small amount
before all fibers which are parallel to the Langer’s lines
orient themselves. Nevertheless, when the skin is
stretched across the Langer’s lines, it will extend much
further before those fibers, which are oriented across
the Langer’s lines, become straight and aligned in the
direction of stretch.40,64
Figure 7 shows the results of in vitro tensile test on
the human back skin at a strain rate of 0.012/s in two
directions: parallel and perpendicular to Langer’s
lines.66 This figure illustrates that the skin is stiffer
when testing parallel to Langer’s lines compared to the
transverse direction. Moreover, by comparing the
slopes of the two curves, it can be concluded that
the tissue gets to the maximum stiffness at lower stretch
ratios when testing along the Langer’s lines compared
to the transverse direction. Hence, the fibers become
Figure 6. Langer’s lines.58 straight at lower stretch levels when stretching the tis-
sue in parallel direction.
Conducting in vitro uniaxial tensile test on cat skin,
collagen fibers in the dermis being predominately obtained from the spinal and dorsal lateral area of
oriented parallel to these lines.40,63–65 healthy cats, Veronda and Westmann67 observed that
Those collagen fibers that are parallel to Langer’s while testing in longitudinal direction (parallel to the
lines are stretched more than those across the lines in spine) up to extension ratio, l1, of 2.1, the lateral
6 Proc IMechE Part H: J Engineering in Medicine 00(0)

Failure properties
Several studies, mainly by in vitro uniaxial tension tests,
have attempted to measure the failure properties of
skin. A summary of failure properties of skin found by
different studies is shown in Table 1. Because of the vis-
coelasticity and anisotropy, the skin is expected to have
different failure properties in different strain rates and
loading directions.
Strain rate is among the predominant factors which
affect the soft tissue failure properties. Dombi et al.75
conducted low- and high-rate tensile failure experi-
ments (0.3–60/s) using the dorsal skin of rats 1.5–
6 months of age parallel and perpendicular to body axis
at cephalad and caudad locations on the dorsum. They
found that when testing at high rates skin appears
Figure 8. Preconditioning results of a typical swine skin nearly twice as strong as at static rates of loading. They
sample.68 stated that the total skin strength is the summation of
the collagen fibril strength and the collagen–matrix
interaction.74,84 At low rates of loading, the strength of
extension ratio was 0.3 4 l2 4 1 and through-thickness the tissue is primarily due to the cohesive forces (i.e.
direction was 1 4 l3 4 1.2. Taking into account these crosslinks) between collagen fibrils themselves since the
observations and the fact that for an isotropic material bonds between the collagen fibrils and the glycosami-
l2 = l3, it was concluded that skin is an anisotropic noglycan matrix are broken early as the collagen fibrils
material on through-thickness plane as well. deform along the axis of loading. However, the increase
in tissue strength at high rates of loading was attributed
to the viscous shearing effect between the matrix and
Preconditioning effect collagen fibers.
Connective tissues such as skin, ligament, and tendon The dependency of failure properties on strain rate
which consist of elastin and collagen fibers show not was observed in other studies consistently. For exam-
only rate dependency but also load history–dependent ple, Arumugam et al.76 conducted in vitro tensile test
behavior.57,68,69 Load history dependency means that on 20 samples of goat butt skin with strain rates of
the shape of the load–elongation curve will vary 0.0008, 0.0083, 0.0333, 0.0833, and 0.1667/s and
depending on the previous history of loading (irrespec- observed a continuous increase in ultimate tensile stress
tive of the viscoelastic effects). During cyclic stretching (UTS) with increasing strain rate. In addition, using
of connective tissues, the nonlinear loading and unload- three different compression test machines with different
ing curves (forming a loop called hysteresis) change strain rate capabilities, Shergold et al.59 measured the
from a cycle to another and follow different paths. failure properties of porcine skin in strain rates ranging
Figure 8 shows the stress–strain relationship for a typi- from 0.004 to 4000/s and found that UTS and failure
cal swine skin sample which is imposed to cyclic defor- strain have positive and negative relationships, respec-
mation to a certain value of stress.68 As shown, the tively, with the strain rate (Figure 3). Also, when com-
sample experiences different hysteresis loops in each paring the results of Nı́ Annaidh et al.41 with Gallagher
subsequent cycle. Generally, each loop gradually shifts et al.,42 former with low strain rate and latter with high
to the right (strain offset) and the amount of this shift strain rate, done on specimens excised from human
decreases with subsequent cycles until a converged hys- back, it can be realized that the reported UTS and fail-
teresis loop is obtained. ure stretch ratios are higher and lower, respectively, in
Because of the effect described above (which is com- high rate tests. Besides, doing uniaxial tensile test on
mon in many soft biological tissues), preconditioning is cadaveric back skin, Ottenio et al.83 showed that the
a necessary step to obtain consistent and repeatable UTS (15.9 6 5.7, 24.1 6 7.1, and 25.8 6 8.2 MPa in
experimental results. Such a procedure orientates the 0.06, 53, and 167/s strain rates, respectively) and strain
microstructure of tissues to its natural in vivo align- energy (4.3 6 1.8, 4.5 6 1.3, and 8.2 6 3.5 MJ/m3 in
ment, allowing tissues to gradually adapt to loading 0.06, 53, and 167/s strain rates, respectively) of the skin
and hence, resulting in more consistent data from increase with increased strain rate. Nevertheless, they
mechanical testing.70 Thus, before in vitro testing of argued that stretch ratio at failure (1.3–1.5) is indepen-
the skin, samples should undergo cyclic loading with dent of the strain rate.
strains considered to be within the physiologic regime. Because of the anisotropic properties of skin, its fail-
Karimi et al.62 reported that the preconditioning of the ure properties are expected to depend on loading direc-
rat skin can be fulfilled after 10 cycles. tion. Ankersen et al.78 obtained skin specimens from
back and belly of pig, parallel and transverse to the
Joodaki and Panzer 7

Table 1. Summary of the results of some studies on the skin failure properties.

Study Sample source Rate UTS (MPa) Failure stretch ratio

Jansen and Rottier71 Human abdomen 0.01/s 1.0–24.0 1.17–3.07

Dunn and Silver72 Human abdomen 2.0–15.0
Vogel73 Various human sources 5–32 1.30–2.15
Haut74 Rat skin 60/s 1.2–3.2 1.519–2.248
Dombi et al.75 Rat skin, parallel 0.3|60/s 3–8|6–12
Rat skin, transverse 0.3|60/s 5–17|10–24
Arumugam et al.76 Goat skin 0.00083/s 102.5 1.353
Goat skin 0.0083/s 106.6 1.297
Goat skin 0.033/s 110.8 1.263
Goat skin 0.083/s 117.3 1.313
Goat skin 0.16/s 126.5 1.300
Greven et al.77 Anuran skin 0.014/s 6.21–10.09 1.427–1.519
Ankersen et al.78 Porcine back|belly skin, parallel to spine 0.01/s 15 6 6.7|19.0 6 11.0 1.42 6 0.11|1.30 6 0.03
Porcine back|belly skin, transverse to spine 0.01/s 15.2 6 4.7|6.3 6 1.5 1.37 6 0.10|1.96 6 0.19
Silver et al.79 Human abdomen and thorax 0.001/s 2.0–15.0
Shergold et al.59a Porcine skin 0.004/s 20 0.42
Porcine skin 0.4/s 25 0.50
Porcine skin 40/s 31 0.53
Porcine skin 4000/s 57 0.58
Jacquemoud et al.80 human forehead and arm 55/s 5.7–12.6 1.27–1.59
Zhou et al.81 Porcine skin 0.0025–0.1/s 0.25–1.0 2.23–2.26
Lim et al.82 Porcine skin 1700–3500/s 0.1–0.8 1.16–1.30
Nı́ Annaidh et al.41 Human back, parallel 0.012/s 17–28 1.46–1.74
Human back, perpendicular 0.012/s 10–16 1.52–1.61
Human back, 45° 0.012/s 20.2–28.8 1.43–1.52
Gallagher et al.42 Human back, parallel 1–2 m/s 30.5 6 6.53 1.22 6 0.02
Human back, perpendicular 1–2 m/s 29.82 6 12.77 1.26 6 0.04
Human back, 45° 1–2 m/s 23.51 6 6.53 1.28 6 0.06
Ottenio et al.83 Human back, parallel 0.06|53|167/s 20.9|32.1|30.9 1.50|1.30|1.30
Human back, perpendicular 0.06|53|167/s 10.4|16.7|19.7 1.50|1.40|1.50
Human back, 45° 0.06|53|167/s 16.4|23.7|26.3 1.40|1.30|1.50

UTS: ultimate tensile stress.

a
As the type of testing was compression, the failure stretch ratios are less than 1.

spine and conducted uniaxial testing in vitro. They Aging effects

found the anisotropy to be significant in belly skin Age is a factor affecting the stiffness of human skin.
(greater UTS and smaller failure stretch ratio in the Sanders31 did in vivo torsion test on human skin and
longitudinal direction compared to the perpendicular found that between 6 and 61 years, the stiffness of skin
direction) but negligible in the back skin. Nı́ Annaidh under torsional conditions decreases with age (0.03–
et al.41 showed that the UTS is higher in the direction 0.10 MPa for \ 30 years and 0.02–0.05 MPa for . 30
of Langer’s lines compared to the transverse condition. years). The loss of elasticity in a direction parallel to the
In addition, they observed that the stretch ratio at UTS skin surface was attributed to fraying and fragmenta-
is influenced by the orientation of the sample. tion of elastic fibers at advancing age. Alexander and
Similarly, Ottenio et al.83 showed that UTS (28.0 6 5.7, Cook85 tested the back and forearm of 116 healthy male
22.5 6 8.2, and 15.6 6 5.2 in parallel, 45°, and perpen- and female subjects, ranging in age from 2 to 67 years,
dicular to the Langer’s lines, respectively) and strain via in vivo suction method at constant pressurization
energy (7.1 6 2.8, 5.4 6 3.5, and 4.9 6 1.8 MJ/m3 in par- rate. They observed that the average slope of the stress–
allel, 45°, and perpendicular to the Langer’s lines, strain curve decreases through maturation, reaches a
respectively) depend on the testing direction. minimum between the ages of 15 and 25 years, and then
Nevertheless, by testing on specimens excised from left appears to increase with advancing age (age . 25
and right side° of the human back, they showed that years). In contrary with Sanders,31 Agache58 did in vivo
there is no significant difference of failure parameters torsion test on human forearm skin and found that the
between the left and right samples which is indicative skin elastic modulus increases with increased age
of a line of symmetry about the spine. The authors dis- (0.42 MPa for \ 30 years and 0.85 MPa for . 30
cussed that most of the collagen fibers are oriented in years). In addition, Belkoff and Haut86 did quasi-static
the direction of Langer’s lines, and hence, the tissue has uniaxial tensile tests on skin specimens taken
a higher strength in this direction.
8 Proc IMechE Part H: J Engineering in Medicine 00(0)

Figure 9. (a) Typical viscoelastic behavior of young and old soft tissues88 and (b) relaxation time constants as a function of age for
high torque and low torque.89

longitudinally and transversely to the spine of rats age Skin constitutive modeling
1–4 months and adopted a microstructurally based
model and found the fiber stiffness to increase laterally
Linear-elastic modeling
during maturation. The results also indicated the migra- The simplest constitutive model used to represent the
tion of the heel portion of the response curve (Stage II mechanical behavior of skin and its layers is the linear-
of Figure 2) toward the origin during maturation to be elastic model as proposed by Hooke’s law. This type of
due to less crimping in the fibers. Having data on the model cannot represent the nonlinearity, viscoelasticity,
lower back skin of six post mortem human subjects and load history dependency of the tissue. However,
(PMHS), Tonge et al.87 observed that specimen from knowing the Young’s modulus of the skin provides a
older donors exhibited a stiffer response than those rough estimation of the response of tissue to mechani-
from younger donors. The stiffness ratio (fiber/perpen- cal loading.
dicular direction) was 1.34 6 0.41 and 1.36 6 0.42 in toe
region for younger (43–59 years) and older (61– Whole skin models. Whole skin models refer to constitu-
83 years) specimens, respectively; this value was tive models that look at the effective properties of the
7.27 6 5.20 and 1.57 6 1.08 in the linear region for skin without differentiating the mechanical properties
younger and older specimens, respectively. Hence, it by each layer that makes up the skin. Table 2 provides
was concluded that the older tissues are more isotropic the summary of reported elastic modulus for the whole
than younger ones. skin using various testing techniques, namely, indenta-
Bader and Bowker88 investigated the effect of age on tion, torsion, tension, and suction. Knowing the com-
skin viscoelasticity by conducting test on two groups of plicated mechanical behavior of the soft tissue, the test
subjects: a young (20–26 years) healthy group and an conditions under which the elastic moduli are estimated
aged (57–79) group comprising both healthy volunteers should be considered for these values to be used accord-
and hospital patients. A constant load was applied ingly. In the following paragraphs, brief descriptions of
through a plane-ended indenter onto the soft tissue sur- these studies are provided.
face of a limb resting in a fixed position for a 2-min Several studies have used indentation method to
period and then removed and allowed the tissue to estimate the skin’s Young’s modulus. Bader and
recover for further 2 min. Significant differences in both Bowker88 did in vivo indentation tests on two groups
the instantaneous response and creep slope due to the of young and old human subjects; 11.7 kN/m2 load was
viscoelasticity between two age groups were observed. applied through a 20-mm indenter at the forearm and
While the young tissue showed higher instantaneous 7.0 kN/m2 through a 40-mm indenter at the thigh.
stiffness, it had lower creep slope. Figure 9(a) illustrates Comparing to Figure 2, it seems that the stiffness has
the typical viscoelastic behavior of young and old soft measured at the end of region 1 and beginning of
tissues. In addition, Escoffier et al.89 did in vivo torsion region 2. Elleuch et al.90 applied load domain of range
test on ventral face of the forearm skin of 54 men and from 20 to 80 mN via a spherical indenter of 6.35-mm
69 women and observed that the creep relaxation time radius with the fixed loading speed of 50 mN/s on three
(when only one relaxation time constant was defined, young women (24–26 years old) in situ. Then, they
see section ‘‘Viscoelastic modeling’’) decreases linearly developed a finite element model which simulated the
with age (Figure 9(b)). This observation is consistent experimental apparatus and showed that with the load-
with Bader and Bowker.88 ing speed of 50 mN/s, the human skin has an elastic
Joodaki and Panzer 9

Table 2. Summary of some studies on Young’s modulus of whole skin.

Indentation test study E (MPa) Test specimen Probe size (mm) Type

Bader and Bowker88 0.00111 Aged female forearm 2000 In vivo

0.00109 Young female forearm 2000
0.00151 Young male forearm 2000
0.00199 Young male thigh 4000
Elleuch et al.90 0.014 Female forearm 6350 In vitro
Iivarinen et al.91 0.210 FR 2000 In vivo
0.446 IFL 2000
0.651 IEL 2000
0.254 VO 2000
Crichton et al.92 29.70 6 21.26 Ear of mice 5.00E201 In vitro
11.15 6 4.90 Ear of mice 1
8.19 6 5.26 Ear of mice 2.5
3.26 6 2.01 Ear of mice 5
1.31 6 0.80 Ear of mice 10
0.86 6 0.44 Ear of mice 20

Torsion test study E (MPa) Test specimen Type

Sanders31 0.03–0.10 Forearm (\ 30 years) In vivo

0.02–0.05 Forearm ( . 30 years)
Agache et al.32 0.42 Dorsal forearm (\ 30 years) In vivo
0.85 Dorsal forearm ( . 30 years)
Escoffier et al.89 1.12 Ventral forearm In vivo
Grebeniuk and Utenkin93a 0.58–1.33 Ventral forearm In vivo

Tension test study E (MPa) Test specimen Direction with respect to Langer’s lines Type

Manschot12 22 Human leg Parallel In vivo

6 Human leg Perpendicular

Suction test study E (MPa) Test specimen Type

Grahame and Holt94 18–57 Forearm In vivo

Barel et al.95 0.13–0.26 Male thigh In vivo
Diridollou et al.96 0.11|0.12 Human male|female forearm In vivo
0.21|0.25 Human male|female forehead

FR: forearm at rest; IFL: isometric flexor loading; IEL: isometric extensor loading; VO: venous occlusion.
a
The reported values are shear moduli (G). In case of incompressible material behavior, E = 3G.

behavior until a ‘‘critical normal load’’ (PP = 25 mN). finite element model of the indentation experiment was
Based on their observation, for any indentation load created and the model response was matched with the
less than PP, there is no hysteresis and consequently, experimental results to determine the elastic modulus
the skin has elastic behavior. In an effort to quantify for each tissue in the model.
the skin’s elastic properties at a micro-scale, Crichton Torsion is the other major testing method used for
et al.92 probed full-thickness skin with customized skin stiffness measurement. Sanders31 measured the
probes of scales from sub- to super-cellular (0.5–20 mm torsional elasticity of intact human skin in vivo by
radius) and observed an inverse relationship between transmitting the twisting moment of 8.3 3 1024 N m to
the used probe and elastic modulus. Iivarinen et al.91 a circular disk with a diameter of 8.7 mm that was
measured stiffness of forearm soft tissue (skin and adi- attached to the forearm skin. Although they observed
pose layers) under four different indentation test proto- that the skin response is a function of time, they argued
cols: (1) forearm at rest (FR), (2) forearm under that the initial quick response of the skin to the applied
isometric flexor loading (IFL), (3) forearm under iso- load can be described as a truly elastic component. By
metric extensor loading (IEL), and (4) forearm under assuming the Poisson’s ratio of n = 0:5 and skin thick-
venous occlusion (VO). In equations (2) and (3), the ness of h = 0:1 cm, and considering the elastic compo-
loading forces were monitored using a dynamometer, nent only, they calculated the elastic modulus using
and in equation (4) the soft tissue swelling was induced equation (6). Agache et al.32 applied a higher torque
by VO using a pressure cuff. The maximum indenter (28.6 3 1023 N m) on human forearm skin in vivo using
force was 0.33, 0.40, 0.55, and 0.70 N for FR, VO, IFL, a disk, which resulted in a twist angle of 2°–6° between
and IEL, respectively. Having the thickness of each the disk and surrounding ring, and evaluated the elastic
layer measured by ultrasound imaging, a multi-layer modulus by the slope of the linear portion (first stage
10 Proc IMechE Part H: J Engineering in Medicine 00(0)

of Figure 2) of the stress–strain curve. Using the same probes, respectively. Geerligs et al.97 also found the
method as Agache et al.,32 Escoffier et al.89 applied two Young’s modulus for this layer to be 2.6 6 0.6 MPa by in
torques, a low one at 2.3 3 1023 N m and a higher one vitro indentation testing on abdominal skin of women
at 10.4 3 1023 N m, to the ventral forearm of 123 with probe radius of 500 mm. Similarly, Crichton et al.20
healthy volunteers and measured the torsion angle. reported the elastic modulus of epidermis equal to 0.751–
Grebeniuk and Utenkin93 applied a constant torque to 2.420 MPa by means of 6.62- and 1.9-mm-diameter sphe-
various anatomical sites on 12-year-old children for rical probes, respectively. In addition, Kendall et al.29
several seconds, resulting in 7°–10° rotation, causing no reported the murine ear skin epidermis Young’s modulus
wrinkles. to be equal to 2.9–11.1 MPa by means of 5 and 2 mm
Manschot12 performed uniaxial tensile tests on the indentation probes.
human calf, both across and along the tibial axis. Two
square tabs (10 3 10 mm) were attached to the skin Dermal components. The dermis is believed to be the
(thickness of 1.2 mm) with cyanoacrylate adhesive with main load-bearing component in the skin and is stiffer
a distance of 5 mm in between. Four saw tooth–shaped compared to the epidermis. Using indentation method,
loads (maximum 12 N) were applied with a loading and Crichton et al.20 reported the Young’s modulus of der-
interval time of 10 and 20 s, respectively. They observed mis to be 7.33–13.48 MPa for 6.62- and 1.90-mm-dia-
nonlinear stress–strain curves in both directions and meter spherical probes, respectively, which are higher
estimated the elastic moduli by fitting lines to these than the reported epidermal components’ stiffness in the
curves. As the estimated elastic moduli were in a high same study. Finding the elastic modulus of the intact
range of stress (up to 5 and 10 MPa for parallel and human skin and having the volume fraction (0.01 from
perpendicular directions, respectively), the reported val- Tregear99), Sanders31 estimated this value for the elastin
ues were higher than most of the other studies included fibers in the living human skin to be 2–10 GPa. For the
in Table 2. collagen fibers, an elastic modulus from 0.112 to 1 GPa52
Some other studies have used suction method for (in linear region) is reported in literature.
this purpose. Grahame and Holt94 found values for the
modulus of elasticity from 18 to 57 MPa from in vivo Bi-linear elastic modeling
measurements of human forearm skin by suction test-
ing. Barel et al.95 used this method and measured the Based on the behavior discussed in section ‘‘Nonlinear
deformation–time and pressure–deformation curves. stress–strain relationship,’’ some studies modeled the
Although the pressure–deformation curve was non- skin as a bi-linear elastic material. In fact, that formu-
linear, they found a linear part between pressures of lation models the skin in a way that when the strain is
150 and 500 mbar and estimated the elastic modulus in less than a threshold, eeq, the stress–strain relationship
that part. Diridollou et al.96 used an ultrasound device obeys Hooke’s law with Young’s modulus of E1 , and
to measure the vertical displacement of the surface of when the strain passes that threshold, the material
the skin and its thickness under suction. The suction responds to the strain linearly with new Young’s modu-
was done by a linear increase in vacuum pressure up to lus of E2 (Figure 10).87,100 A summary of bi-linear
100 mbar in 10 s and a subsequent linear unloading in models is presented in Table 3.
10 s. Consequently, the elastic modulus was estimated Delalleau et al.100 conducted in vivo suction defor-
in relatively low stresses (first stage of Figure 2) and it mation on the volar aspect of the human forearm.
explains the relatively small reported value. Diridollou Then, they modeled this test with the finite element
et al.96 reported slightly larger skin elastic modulus for method to compare the experimental and simulated
females compared to males. curves using an inverse method that allows the identifi-
cation of skin mechanical parameters. To account for
the nonlinear behavior of skin, they modeled the tissue
Models of skin layers as a bi-linear elastic material. The skin was assumed to
Epidermal components. The epidermis layer is a suitable be a homogeneous material, and the viscous and aniso-
site for pain-free drug delivery due to some of its physio- tropic components were not considered. In addition,
logical characteristics. Consequently, attention has been they assumed the skin to be quasi-incompressible
paid to mechanical properties of the epidermal compo- (n = 0:45), and hence, the effect of the bulk modulus
nents in literature. Geerligs et al.97 found the Young’s was considered to be insignificant. Then, the elasticity
modulus for the whole epidermis to be 1.1 6 0.2 MPa via was defined by two slopes, which are related to the first
in vitro indentation using a spherical probe with a radius and third phases of skin behavior as shown in Figure
of 500 mm. Using suction measurements combined with 10(a). Showing that there is a relationship between
ultrasound and optical coherence tomography, Hendriks thickness and mechanical properties, model parameters
et al.98 measured the elastic modulus for the upper skin for experimental curves corresponding to minimal,
layer (defined as epidermis and papillar dermis in that average, and maximal thicknesses of tested forearm
article) equal to 0.66 kPa. Crichton et al.20 measured elas- dermis were determined.
tic modulus of stratum corneum to be 1.033–2.346 MPa Dunn and Silver72 and Nı́ Annaidh et al.41 reported
using 6.62- and 1.9-mm-diameter spherical indentation two different stiffness, one called the initial slope and
Joodaki and Panzer 11

Figure 10. (a) Bi-linear elasticity due to the orientation of the collagen fibers and (b) corresponding numerical model.100

Table 3. A summary of bi-linear elastic skin models.

Study Sample Direction wrt Type Thickness E1 (MPa) E2 (MPa) eeq (%)
Langer’s line (mm)

Dunn and Silver72 Human abdomen In vitro tension 0.1 18.8

Delalleau et al.100a Human forearm In vivo suction 0.86 0.131 0.365 3.8
Human forearm 1.08 0.064 0.273 3.9
Human forearm 1.51 0.041 0.187 6.8
Nı́ Annaidh et al.41b Human middle back Parallel In vitro tension 2.56 1.21 112.47
Human lower back Parallel 2.56 1.95 73.81
Human middle back Perpendicular 2.56 0.91 63.75
Human lower back Perpendicular 2.56 0.54 37.66
Human upper back 45° 2.56 0.99 82.62
Human middle back 45° 2.56 1.33 103.49
Human lower back 45° 2.56 1.30 82.81
Tonge et al.87 Human lower back Parallel In vitro bulge testing 4.62 0.128 13.863 7
Human lower back Perpendicular 4.62 0.095 1.439 16
Human upper back Parallel 5.18 0.103 11.471 10
Human upper back Perpendicular 5.18 0.073 4.145 21
Human—unknown side Parallel 2.46 0.131 10.621 11
Human—unknown side Perpendicular 2.46 0.094 11.781 15.33
a
The shown thickness is the thickness of dermis.
b
An average thickness with standard deviation of 60.39 mm has been reported for the whole skin.

the other elastic modulus, for excised human skin which W = C1 (I1  3) + C2 (I2  3) ð9Þ
based on their method are equivalent to E1 and E2 as
where C1 and C2 are the constant and material prop-
shown in Figure 10(b). Tonge et al.87 investigated the erty, respectively, and I1 and I2 are the first and second
nonlinear anisotropic properties of PMHS lower back invariants of the right Cauchy–Green deformation ten-
skin using bulge testing. Samples were loaded from a sor, respectively. If C2 is equal to 0, the model would
baseline pressure of 0.276 kPa to a maximum pressure be equivalent to the neo-Hookean model. For small
of 5.516 kPa at a rate of 0.069 kPa/s. They observed strains, the contribution of the second term is negligible
greater transition strain and higher stiffness in linear and it can be shown that C1 can be related to the
region in the fiber direction compared to the parallel Young’s modulus using E = 6C1 equation. Shergold
direction which is consistent with the discussion in sec- et al.59 used the results of uniaxial tensile testing on
tion ‘‘Anisotropy.’’ human abdomen skin obtained from Jansen and
Rottier,71 taken 80 mm laterally from the median line
between the umbilic and pubic area. Their effort to
Hyper-elastic modeling
model the specimen as a Mooney–Rivlin material
Mooney–Rivlin and neo-Hookean model. For a Mooney– resulted in C1 = 300 kPa and C2 ’0. Also, Delalleau
Rivlin material, the strain energy density function is et al.100 identified the human forearm neo-Hookean
defined by the following formulation parameter, C1 , equal to 57, 39, and 19 kPa for dermis
12 Proc IMechE Part H: J Engineering in Medicine 00(0)

Table 4. Summary of Ogden model parameters for the skin.

Study Sample Type Strain rate (1/s) a1 a2 m1 (kPa) m2 (kPa)

Shergold and Fleck102 Human skin In vitro tension 9 110

Shergold et al.59 Pig skin In vitro compression 0.004 12 400
Pig skin In vitro compression 0.4 12 1200
Pig skin In vitro compression 40 12 2200
Pig skin In vitro compression 4000 12 7500
Lim et al.82a Pig skin In vitro tension 0.005 11|7 10|3
Pig skin In vitro tension 0.5 11|7 20|8
Pig skin In vitro tension 1700 11|7 180|40
Pig skin In vitro tension 2500 11|7 230|200
Pig skin In vitro tension 3500 11|7 300|370
Flynn et al.103b Human skin In vivo tension 2.10 35.24 49.8 0.0003
a
x–y means parallel (x) and perpendicular (y) direction to the spine in pig sample.
b
The reported parameters are average of each parameter found for five volunteers.

with thickness of 0.86, 1.08, and 1.51 mm, respectively. ground substance dictates the value of m, and these
Nevertheless, these models were unable to fully repre- deformation mechanisms are sensitive to the strain rate.
sent the nonlinear skin stress–strain curve. They discussed that since the Ogden model accurately
describes a wide range of strain hardening behaviors
while Mooney–Rivlin model only gives an accurate
Ogden model. For an incompressible Ogden101 model, description of the constitutive behavior of solids that
the strain energy density function is defined by the fol- have a low rate of strain hardening, Ogden model is a
lowing formulation more accurate representative of skin. It is especially
X
N because their fitted Mooney–Rivlin model had ended
2m
W= i
(la1 i + la2 i + la3 i  3) ð10Þ up with C2 = 0 and consequently decreased to a neo-
i=1
a2i Hookean model.
where l1 , l2 , and l3 are the principle stretch ratios, a is
a strain hardening exponent, and m has the interpreta- Polynomial model. Lapeer et al.104 used three constitutive
tion of the shear modulus under infinitesimal straining. models, namely, general polynomial (GP), reduced
A summary of the Ogden models for the skin is pro- polynomial (RP), and Ogden, to fit the stress–strain
vided in Table 4. relationship of human skin under in vitro tension. The
In an effort to develop a model for the deep penetra- following formulation represents GP model
tion of human skin, Shergold and Fleck102 formulated !
the mechanical behavior of tissue via a one-term Ogden XN
1
i j
W= cij (I1  3) (I2  3)  p(I3  1) ð11Þ
model (i = 1 only). The results showed that this model i+j=1
2
gives a reasonable approximation to the concave stress–
strain curve observed for skin. where Ik , k = 1, 2, 3, is the kth strain tensor principal
In another study, Shergold et al.59 used the experi- invariant, cij are constants determined experimentally,
mental data on pig skin and showed that if the skin is and p is the hydrostatic pressure. Since biological materi-
modeled as an Ogden material, the strain hardening als are usually assumed to be incompressible (I3 = 1), the
exponent, a, is almost independent of strain rate while last term will be 0 automatically. First- and second-order
m increases monotonically with increasing strain rate. GP models were used to fit the skin stress–strain curves.
Consistently, Lim et al.82 tried to fit the results of uniax- Similarly, the RP formulation for describing the
ial tensile experiments on pig skin at different rates with stress–strain relationship of soft tissue can be written as
the Ogden model and found a to be independent of !
strain rate. Flynn et al.103 conducted in vivo testing on XN
1
i
W= Ci (I1  3)  p(I3  1) ð12Þ
facial skin using a micro-robotic device which subjects 2
i=1
the skin to a rich set of deformation cycles with a fre-
quency of 0.1 Hz. Presuming that the experimentally where Ci is a constant.
observed anisotropy is entirely due to the in vivo tension Two criteria were to be taken into account to ensure
field, they modeled the skin assuming it to be isotropic. a good model fit to experimental data, namely, the
Shergold et al.59 argued that the value of a is dic- quality of fit using the coefficient of determination, R2,
tated by the geometric evolution of the collagen net- and stability using the Drucker’s105 stability test. The
work and is thereby independent of strain rate. In condition of Drucker stability is satisfied by requiring
contrast, the resistance to rearrangement by bending of the tangential material stiffness matrix to be positive
the collagen fibers and by shearing of the intervening definite.
Joodaki and Panzer 13

Table 5. Fitted parameters of reduced polynomial model for the human skin.104

C1 (kPa) C2 (kPa) C3 (kPa) C4 (kPa) C5 (kPa) C6 (kPa) Directiona

14.40 415 1970 P

12.1 356 1770 A
10.85 152.3 72.15 2482 P
10860 150.2 762.5 2256 354.5 P
12.4 351 1020 12,300 –44,200 44,600 A
14.7 390 1100 16,900 –67,300 77,800 P
a
P: parallel with Langer’s lines; A: across Langer’s lines.

Among the three recruited constitutive models to fit abdominal skin conducted by Lanir and Fung.39
their experimental data, only RP models, ranging from Instead of the method of least squares, they determined
first to sixth order, were found to be stable. Based on the constants by requiring the model fits exactly the
RP formulation and using in vitro tensile tests on experimental data at some selected points. The number
abdominal skin of middle-aged females, Table 5 sum- of experimental data points required depends on the
marizes the results for models with different orders. number of constants to be determined. Since only ten-
sile testing data were used, only those constants associ-
Exponential models. According to Veronda and ated with tensile strains could be determined. As the
Westmann,67 if a compressible material is of Mooney– shear strain, e12 , was 0 in all experiments, a3 and a3
Rivlin character (i.e. ∂w=(∂Ii ) = const:), a plot of could not be determined.
s=2(l1  (l2 l3 )=l1 ) versus l2 l3 would yield a straight Table 6 shows the list of optimized Fung-type expo-
line. Nevertheless, this plot for cat skin that underwent nential model constants fitted to the results of biaxial
uniaxial testing exhibited an exponential or power law tests on three different rat skins. For each set of fitting,
behavior rather than linear. After several iterations to two tests on a single specimen were done, once tension-
model skin mechanical behavior, they selected the fol- ing in x (longitudinal)-direction while the stretch in y
lowing form of strain energy function (transverse)-direction was held constant, and once
stretching in the y-direction when the stretch in the
W = c1 (eb(I1 3)  1) + c2 (I2  3) + g(I3 ) ð13Þ x-direction was constant. While in most of the sets all g
s were assumed to be 0, for some of the tests on speci-
where c1 , c2 , and b are constant and since in unde- men number 40, fitting was done with two types of
formed state W = 0, g(1) = 0. Using the isotropic assumptions, once assuming all g s to be 0 and once
model and considering the material as incompressible, assuming g 4 = g 5 to be finite while other g s were 0. In
the following strain energy function was obtained all cases, a1 was assumed equal to a2 .
W = 0:0134½e4:4(I1 3)  1  0:0295(I2  3) + g(I3 )
ð14Þ Gasser–Ogden–Holzapfel model. The Gasser–Ogden–
Holzapfel (GOH) formulation,66,107 which can be used
Another form of exponential model is Fung-type expo-
to model an incompressible solid with two preferred
nential model,106 in which a pseudo strain potential,
directions (e.g. with two families of fibers), has the fol-
ro W, for tissue is defined so that the stress can be
lowing strain energy density
derived from the expression Sij = (∂r0 W)=(∂eij ), where
ro is the density of the material in the initial unde- m X ki1 2
formed state, W is the strain energy per unit mass, and W= (I1  3) + m (eki2 ½tr(Hi C)1  1)
2 i = 1, 2
2ki2
eij are the components of Green’s strain tensor. This
pseudo strain potential is represented by the form ð18Þ
r0 W = f(a, e) + c exp½F(a, e) ð15Þ where the index i is used to describe the mechanical
behavior of two families of fibers separately. m is posi-
where
tive material constants, ki1 and ki2 are dimensionless
f(a, e) = a1 e211 + a2 e222 + a3 e212 + 2a4 e11 e22 ð16Þ stiffness parameters, C is the right Cauchy–Green strain
tensor, and I1 = tr(C). H1 and H2 are structure tensors
F(a, e) = a1 e211+ a2 e222 + a3 e212 + 2a4 e1 e2
ð17Þ which are defined as
+ g 1 e31 + g 2 e32 + g 4 e21 e2 + g5 e1 e22
Hi = ki I + (1  3ki )a1 3a2 ð19Þ
and a1 , . . . , a4 , a1 , . . . , a4 , g 1 , . . . , g 5 , and c are
constants. where k1 and k2 are dispersion factors, and a1 and a2
Tong and Fung106 used this model to fit the experi- are the unit vectors of the material preferred directions.
mental data obtained by biaxial tensile testing on rabbit In case of homogeneous uniaxial tensile test, when two
14 Proc IMechE Part H: J Engineering in Medicine 00(0)

Table 6. A list of Fung-type exponential model constants fitted to three rat skin specimens (nos 32, 36, and 40) in different tests.106

Exp. no. Fixed transverse a1 a2 a4 c a1 = a2 a4 g4 = g5

x exp y exp

36 X1(III) and Y1(II) 1 0.974 7.33 20.37 6.81 1.44E203 5.98 2.24 0
36 X1(III) and Y1(III) 1 0.974 9.49 68.1 6.57 7.30E206 6.23 2.27 0
36 X1(IV) and Y1(IV) 1 0.974 9.66 31.92 9.01 1.63E206 5.56 2.24 0
36 X6 and Y7 1.039 1.057 7.99 53.58 6.07 7.78E205 4.70 0.63 0
32 X7 and Y20 1.361 1.273 5.10 9.79 22.3 3.42E210 14.55 4.2 0
32 X6 and Y19 1.313 1.17 8.1 14.64 11.57 5.99E208 14.46 3.86 0
32 X4 and Y19 1.189 1.17 5.07 16.61 0.91 9.26E204 10.19 6.84 0
32 X3 and Y18 1.127 1.083 4.34 12.25 2.02 1.89E203 10.61 5.93 0
32 X2 and Y17 1.058 1 3.38 11.74 1.27 1.60E202 11.54 5.18 0
40 X11 and Y6 1 1.45 3.78 23.8 2.21 0.02 15.6 8.63 0
40 X13 and Y2 1 1.04 3.24 11.2 0.37 0.0244 10.18 2.91 0
40 X8 and Y10 1.08 1.1 4.45 13.2 0.63 0.00631 9.53 2.14 0
40 X6 and Y8 1.23 1.28 7.73 17.1 1.83 0.000134 10.3 2.81 0
40 X13 and Y11 1 1 3.79 12.7 0.58 7.94E203 10.4 2.59 0
3.79 18.4 0.58 7.94E203 10.4 2.59 15.6
40 X5 and Y5 1.31 1.52 9.36 27.2 0 4.46E205 10.5 3.13 0
9.32 27.2 0 4.69E205 9.6 4.24 9.13
40 X6 and Y6 1.23 1.45 8.2 23.1 0.79 8.90E205 10.8 2.66 0
8.18 22.9 0.66 1.01E203 9.62 4.99 6.24
40 X7 and Y7 1.15 1.38 6.44 23 0.25 4.40E204 10.8 2.45 0
6.49 25.2 0.10 4.10E204 11.8 21.6 27.4
40 X8 and Y11 1.08 1.04 4.12 11.5 0.82 1.17E202 9.27 2.09 0
4.13 12 0.84 1.14E202 9.56 0.8 14.9
40 X9 and Y9 1.04 1.18 3.3 14.4 16.6 5.11E202 8.87 5.27 0
3.3 14.4 16.6 5.11E202 8.87 5.28 8.85

x-Experiments are those in which the stretch in the y-direction is constant; y-experiments are those in which the stretch in the x-direction is
constant. In some sets of data, all gs are assumed 0; in others, once the fitting is done with all gs as 0 and once when g4 = g5 is assumed finite while
other gs are 0.

families of fibers are equivalent to the same dispersion

Table 7. The results of curve-fitting for three human skin
factor, k11 = k21 [k1 , k12 = k22 [k2 , and k1 = k2 [k. samples.66
Nı́ Annaidh et al.66 conducted uniaxial tensile test on
human back skin samples when the tension was applied Sample m (MPa) k1 k2 k
along the bisector of a1 and a2 . They fitted a GOH
model to the stress–strain curves for three samples: par- Parallel 0.2014 243.6 0.1327 0.1535
Perpendicular 0.2014 243.6 0.1327 0.1456
allel, perpendicular, and non-symmetric (135°) with Non-symmetric 0.2014 243.6 0.1327 0.1602
respect to Langer’s lines. Defining the preferred direc-
tions as the mean of the orientations of two families of
interweaving fibers, the authors estimated the angle constants ti , and the last term, t=h, represents a viscous
between a1 and tension direction to be 41° through his- creep term.
tology in their extracted samples. The other structural Del Prete et al.108 measured the viscoelastic proper-
parameter, k, was also determined by histological anal- ties of skin samples in three types of mice skin (Tsk,
ysis. Table 7 shows the values found for k and the control, and Mov-13) using uniaxial stretching and
results of curve-fitting (stress–strain relationship) for measuring the creep response. Table 8 summarizes the
the parameters m, k1 , and k2 . parameters of fit models.
As skin shows nonlinear instantaneous elastic
response, quasi-linear viscoelastic (QLV) formulation
Viscoelastic modeling can be a more suitable constitutive model to describe
the behavior of this tissue.109 In this model, stress, s, as
The formulation below models a linear viscoelastic
a function of strain and time is
material via the creep compliance J(t)
! ðt
1 X‘
1 tt t e dse de(t)
J(t) = + ei + ð20Þ s(e, t) = s (0)G(t) + G(t  t) dt ð21Þ
E0 E h de dt
i=1 i 0+

where the term 1=E0 represents the initial elastic where se is the instantaneous elastic response (which is
response; Ei are the viscoelastic terms with time a nonlinear function of strain in a QLV material), and
Joodaki and Panzer 15

Table 8. The fit parameters of linear viscoelastic model of mice skin.108

Skin type E0 (kPa) E1 (kPa) t 1 (s) E2 (kPa) t 2 (s) h (MPa s)

Control 335 6 48 1731 6 490 6.13 6 0.32 1542 6 315 84.2 6 3.2 1375 6 275
Tsk 269 6 46 2103 6 460 5.75 6 0.31 1639 6 301 86.3 6 3.1 1678 6 263
Mov-13 206 6 39 2030 6 394 6.46 6 0.26 1574 6 253 85.6 6 2.6 1209 6 221

G is the reduced relaxation function which can be writ- response to nano-indentation at sub-cellular and bulk-
ten in the form of Prony series cellular scale was measured. A QLV model with the
  Ogden form instantaneous response was used.
X
N
t Assuming incompressible material, it can be shown
G(t) = G‘ + Gi exp  ð22Þ
ti that22
i=1
"  a   #
where ti are the relaxation times, Gi are the modulus 40Einst a2 a  21 a a1
P= 1  0:2  1  0:2
coefficients, G‘ is the long-term modulus, and 1  v2 Rsi Rsi
P
G‘ + N i = 1 Gi = 1. Table 9 shows the summary of ð24Þ
some QLV skin models.
In an effort to develop a multi-layer model of skin where P, a, Rsi, v, and a are the indentation load
consisting of the stratum corneum, dermis, and under- applied, contact radius, radius of spherical indenter,
lying hypodermis, Flynn and McCormack110 imple- Poisson’s ratio, and material constant, respectively.
mented constitutive equation of each layer into a finite Einst can be written as
element model. The data on rapid unconfined compres-     
t t
sion of human calcaneal fat samples were used to model Einst = Er 1  G1 1  e t1  G2 1  e t2 ð25Þ
the hypodermis layer. Modeling the viscoelasticity by
QLV formulation, the hyperelastic immediate response where Er can be obtained from equation (3).
was modeled via Yeoh model, whose strain energy Liu et al.112 studied the strain rate–dependent beha-
potential is vior of pig dorsal skin in uniaxial stretch and analyzed
the viscoelasticity of skin using QLV formulation. The
W = C10 (I1  3) + C20 (I1  3)2 + C30 (I1  3)3 instantaneous response was described by the following
1 1 1 expression
+ (J  1)2 + (J  1)4 + (J  1)6
D1 D2 D3 se = A(eBe  1) ð26Þ
ð23Þ
where A and B are material constants. The strain was
where Ci0 and Di are the material parameters and given by a constant ramp with different rates, followed
J = det(F), F being the deformation gradient tensor. by a hold phase. They argued that the relaxation time
With the assumption of incompressibility J = 1 and of skin tissue depends on strain rate.
thus, determination of Di is unnecessary. Wang et al.113 found compressive viscoelastic para-
Flynn et al.111 developed a finite element model con- meters of freshly excised mouse skin (139 samples) to
sisting of an Ogden strain energy function (as the be dependent on specimen thickness, strain level, and
instantaneous response) and QLV formulation to simu- strain rate. After preconditioning, compression was
late the in vivo experiments done on the anterior fore- applied with a ramp-up phase at the desired strain rate,
arm and the posterior upper arm skin. For comparison a hold phase at maximum displacement for 6 s, and an
purpose, the instantaneous response was modeled using unloading phase of the same rate as the ramp-up. A
neo-Hookean formulation as well. The effort to fit the QLV model with first-order Ogden form instantaneous
instantaneous response of posterior upper arm with elastic response was fitted to the experimental data. As
neo-Hookean strain energy (resulted in C1 = the specimen thickness ranged from 211 to 671 mm,
0:02197 MPa, G1 = 0:46496, and t1 = 1:86102 s) ended notable variation in both QLV parameters, the relaxa-
up with 34.8% error-of-fit. Nevertheless, the model tion time constant (t 1 = 0:1960:10 s) and steady-state
with Ogden strain energy function had an error-of-fit residual stress ratio (G‘ = 0:2860:13), was observed.
of 9.3% for the same region. Thus, it was concluded When t1 was fixed, a positive correlation between G‘
that the neo-Hookean formulation is not a suitable and skin thickness was noted. In addition, as steady-
model for skin instantaneous elastic response when state stretch, l‘ , was increased from 0.22 to 0.81, t1
modeling skin as a QLV material. (0:2660:14 s) and G‘ (0:4760:17) varied significantly.
Crichton et al.20 did atomic force microscopy inden- With t1 fixed, it was observed that G‘ increases as
tation to quantify separately the viscous and elastic stretch level increases. Furthermore, as the strain rate
behavior of stratum corneum, epidermis, and dermis of was increased from 0.06 to 22.81/s, the median time
mouse skin. By accessing to each layer directly, the constant t1 (with a variety of thicknesses) decreased
16 Proc IMechE Part H: J Engineering in Medicine 00(0)

C10 = 1.649 kPa, C20 = 21.136 kPa, C30 = 21.792 kPa

from 1.90 to 0.31 s, indicating that t1 is negatively cor-
related with strain rate. The reported values in Table 9
for this study are the median fit parameters for all

Instantaneous elastic response parameters

experimental conditions.

E = 0.489|1.510 MPa, a1 = 20:68|20.58

E = 7.33|13.48 MPa, a1 = 57:89|24.23
E = 0.752|1.618 MPa, a1 = 8:68|5.77
Aging and microstructural modeling
To model aging skin and study the effect of different

a1 = 10:703, m1 = 6.422 kPa

a1 = 33:452, m1 = 39.8 kPa

a1 = 6:511, m1 = 7.162 kPa

a1 = 7:924, m1 = 7.189 kPa

a1 = 3:447, m1 = 4.057 kPa

A = 0.268 MPa, B = 0.9.575
a1 = 35:993, m1 = 9.6 kPa
a1 = 35:883, m1 = 2.6 kPa
factors on wrinkle formation, Flynn and

A = 0.481 MPa, B = 9.229

A = 0.506 MPa, B = 8.660
A = 1.125 MPa, B = 6.692
McCormack114 developed a finite element model of the
skin consisting of three layers, stratum corneum, der-
mis, and hypodermis, while modeling each layer with a
different formulation. To form a wrinkle, a region of
skin was compressed by in-plane translation of the skin
layer which resulted in the skin collecting in the com-
pressed region. Then, the changes in physical properties
observed in real aging skin, including change in stratum

4.68|2.16
6.96|4.59
4.71|3.70

403.693
258.711
228.824
corneum moisture level, the density of collagen fibers,

24.024

164.41
t 2 (s)
and natural tension, were applied to the parameters of
the constitutive models of layers to see the effect of
each parameter on wrinkling. The results of simulations

0.19|0.11
0.31|0.33
0.67772

0.2|0.12
exhibited that reduction in the moisture content of stra-

29.826
16.897
12.895
1.685
0.828
0.757

7.384

0.236
0.180
0.310
t 1 (s)
tum corneum, increase in the dermal collagen fiber den-

1.90
sity, and reduction in the pre-stress of the skin result in
the formation of larger wrinkles.

0.45|0.58
0.28|0.35
0.11572

0.2|0.18
None of the aforementioned constitutive models is

0.289
0.251
0.192
0.184
able to represent the skin behavior under precondition-
G2

ing. However, a microstructural formulation developed

by Lokshin and Lanir115 is able to model the viscoelas-
0.35|0.36
0.39|0.38
0.63928

0.22|0.3
ticity of skin and the preconditioning effect. This model
0.3398
0.459
0.398

0.620
0.668
0.755
0.781
0.684
0.548
0.748
0.666
assumes that the load-bearing components of the skin
G1

are collagen and elastin fibers. Each fiber is modeled as

a QLV material which is wavy at the relaxed condition
In vitro compression
Test type|strain rate

and starts load-bearing once it is no longer wavy and

In vitro indentation

reaches the ‘‘straightening length.’’ The straightening

In vivo tension

length increases by applying strain on the fiber and that

is how preconditioning influences the model response.
10%/min
25%/min
50%/min
5%/min

This formulation is based on the observation that dur-

0.06/s
1.47/s
3.54/s
22.8/s

ing the preconditioning process, the collagen fibers’

stress-free length increases as a function of stretch and
time. Nevertheless, the slope of fibers’ stress–strain rela-
In vitro tension

In vitro tension
Table 9. Summary of QLV model parameters for the skin.

x|x means by probe size of 6.62|1.9 mm; E refers to equation (3).

tionship seems to be preserved, that is, the stiffness of

collagen fibers does not change.116 Finally, the load on
Mice stratum corneum
Human upper forearm
Human lower forearm

the whole tissue equals the sum of fiber loads.

Mice viable epidermis
Human hypodermis
Human upper arm

Discussion
Mice dermis
Porcine skin

Mouse skin

Human skin is an anisotropic, nonlinear viscoelastic,

Sample

loading history–dependent material. Because of the

complex structure and behavior of skin, and since dif-
ferent studies have used different specimens and differ-
110

ent testing methods, there is a wide range of reported

Flynn and McCormack

mechanical properties in the literature.

The literature is consistent that increasing strain rate
Crichton et al.20a

corresponds to higher UTS regardless of testing direc-

Wang et al.113
Flynn et al.111

tion.59,75,76,83 Also, stretching human skin along the

Liu et al.112

Langer’s lines results in higher UTS.41,42,83 The failure

Study

stretch ratio is typically higher in the quasi-static tests

compared to the dynamic tests.59,74,83 It seems the
a
Joodaki and Panzer 17

maximum failure strain reported by Jansen and comparing it to the results of another test with different
Rottier71 is too high compared to other studies in nature (e.g. ramp or creep).
Table 1. Not considering Jansen and Rottier’s71 study, Finite element analysis can be used to simulate an
the maximum human skin failure stretch ratio and experiment, optimize the parameters of a constitutive
UTS reported by the studies shown in Table 1 are 1.74 model,117–119 and compare different formulations in
and 42.5 MPa, respectively. terms of representing the tissue behavior.120 In case of
As a fibrous tissue, the preconditioning effect on in vivo experiments, it is essential to take into account
skin is very significant. When the skin is imposed to in the effect of other tissues (e.g. adipose tissue and mus-
vitro constant strain, the decay in the stress is affected cles) on the overall load–deformation response. Finite
both by viscoelasticity and preconditioning. However, element analysis facilitates the mathematical difficulties
the predominant factor for stress decay in the subse- associated with the mixed boundary value problems
quent relaxation tests for the same specimen will be and the interaction between different layers.121 Once a
viscoelasticity.115 suitable constitutive model is identified, it can be imple-
While Sanders31 observed a reduction in human skin mented into finite element models for various applica-
stiffness with aging, Agache et al.,32 Belkoff and tions, including surgical,122,123 aging,114 and wrinkling
Haut,86 and Tonge et al.87 reported higher stiffness simulation.110
with older specimens. Aging also influences the skin Although a more complicated formulation (e.g.
viscoelasticity by increasing creep slope.88,89 QLV model compared to elastic model) is able to char-
A wide range of elastic moduli, from 0.001 to acterize the tissue mechanical behavior better, it is com-
57 MPa, have been reported by different studies with putationally more expensive. Having both these factors
different testing methods. The reported Young’s mod- in mind, choosing the constitutive model should be
uli for a linear model are heavily influenced by the based on the application.
strain-level range, in which the specimen is tested, and
the strain rate. These values can be especially useful if Acknowledgements
the stiffness is measured by applying a load in the range
of aimed application. The authors would like to thank Kevin Kong for his
Among the reviewed studies which have used bi- insightful feedback.
linear elastic modeling, Dunn and Silver72 and Tonge
et al.87 have reported roughly similar fitted model para- Declaration of conflicting interests
meters. Based on these two studies, the elastic modulus The author(s) declared no potential conflicts of interest
in lower strains, E1 , is 0.07–0.13 MPa and in higher with respect to the research, authorship, and/or publi-
strains, E2 , is 4.1–18.8 MPa. Furthermore, both Nı́ cation of this article.
Annaidh et al.41 and Tonge et al.87 showed that both
E1 and E2 are higher in the Langer’s lines’ direction
compared with transverse direction. As for isotropic Funding
hyperelastic models, Ogden model is a better represen- The author(s) disclosed receipt of the following finan-
tative of skin behavior compared to Mooney–Rivlin cial support for the research, authorship, and/or publi-
and neo-Hookean models.59,111 cation of this article: The authors are grateful to the
A more complete constitutive model for skin, which University of Virginia School of Engineering and
can capture the nonlinear stress–strain response and Applied Sciences and Luna Innovation, Inc., for sup-
the viscoelastic response, is the QLV model. This type porting the work leading to this review.
of constitutive model, using an Ogden formulation for
the instantaneous elastic function, has proven effective References
for representing skin simulants that have a similar
1. De Rigal J and Leveque J-L. In vivo measurement of the
behavior.6 Although this model is supposed to take
stratum corneum elasticity. Bioeng Skin 1985; 1: 13–23.
into account the rate dependency of soft tissue, Liu 2. Diridollou S, Patat F, Gens F, et al. In vivo model of the
et al.112 and Wang et al.113 have reported strain level mechanical properties of the human skin under suction.
and rate dependency of relaxation time constants. The Skin Res Technol 2000; 6: 214–221.
QLV assumption for the material is suitable when the 3. Salter D, McArthur H, Crosse J, et al. Skin mechanics
reduced relaxation function is only depending on time. measured in vivo using torsion: a new and accurate
To investigate whether the skin behaves as a QLV model more sensitive to age, sex and moisturizing treat-
material, relaxation tests with different strain levels ment. Int J Cosmet Sci 1993; 15: 200–218.
should be conducted to see whether the ratios of stres- 4. Hochberg J, Meyer KM and Marion MD. Suture choice
ses as a function of time remain constant. If it behaves and other methods of skin closure. Surg Clin North Am
2009; 89: 627–641.
like a QLV material, a better approach to the problem
5. Hendriks F. Mechanical behaviour of human skin in
might be trying to fit the QLV model with the results vivo—a literature review. Nat.Lab. unclassified report
obtained from relaxation tests with various strain levels 2001/820. Philips Research Laboratories, https://pdfs.
for a certain specimen and then assessing the model by
18 Proc IMechE Part H: J Engineering in Medicine 00(0)

semanticscholar.org/975e/38a4599ac81ad440dcca6ebee6a 22. Lin DC, Shreiber DI, Dimitriadis EK, et al. Spherical
31417655e.pdf indentation of soft matter beyond the Hertzian regime:
6. Wood GW, Panzer MB, Bass CR, et al. Viscoelastic numerical and experimental validation of hyperelastic
properties of hybrid III head skin. SAE Int J Mater models. Biomech Model Mechanobiol 2009; 8: 345–358.
Manuf 2010; 3: 186–193. 23. Tabor D. The hardness of metals. Oxford: Oxford Univer-
7. Joodaki H, Forman J, Forghani A, et al. Comparison of sity Press, 2000.
kinematic behaviour of a first generation obese dummy 24. Field J and Swain M. Determining the mechanical prop-
and obese PMHS in frontal sled tests. In: Proceedings of erties of small volumes of material from submicrometer
the 2015 IRCOBI conference, Lyon, 9–11 September spherical indentations. J Mater Res 1995; 10: 101–112.
2015, pp.9–11. International Research Council on Bio- 25. Taljat B, Zacharia T and Kosel F. New analytical proce-
mechanics of Injury (IRCOBI). dure to determine stress-strain curve from spherical
8. Koch RM, Gross MH, Carls FR, et al. Simulating facial indentation data. Int J Solids Struct 1998; 35: 4411–4426.
surgery using finite element models. In: Proceedings of 26. Herbert E, Pharr G, Oliver W, et al. On the measurement
the 23rd annual conference on computer graphics and of stress–strain curves by spherical indentation. Thin
interactive techniques, New Orleans, LA, 4–9 August Solid Films 2001; 398–399: 331–335.
1996, pp.421–428. New York: ACM. 27. Lee E and Radok JRM. The contact problem for viscoe-
9. Iwamoto M, Kisanuki Y, Watanabe I, et al. Develop- lastic bodies. J Appl Mech 1960; 27: 438–444.
ment of a finite element model of the total human model 28. Mattice JM, Lau AG, Oyen ML, et al. Spherical indenta-
for safety (THUMS) and application to injury recon- tion load-relaxation of soft biological tissues. J Mater Res
struction. In: Proceedings of the International Research 2006; 21: 2003–2010.
Council on the biomechanics of injury conference, Munich, 29. Kendall MA, Chong Y-F and Cock A. The mechanical
18–20 September 2002, p.12. Zurich: International properties of the skin epidermis in relation to targeted
Research Council on Biomechanics of Injury. gene and drug delivery. Biomaterials 2007; 28: 4968–4977.
10. Untaroiu CD, Yue N and Shin J. A finite element model 30. Vlasblom D. Skin elasticity. In vivo measurements of
of the lower limb for simulating automotive impacts. Ann small deformation. In: Proceedings of the digest of the 7th
Biomed Eng 2013; 41: 513–526. international conference on medical and biological engi-
11. Odland G and Goldsmith I. Physiology, biochemistry and neering, Stockholm, 14–19 August 1967, p.369. Stock-
molecular biology of the skin. Oxford: Oxford University holm: The Royal Academy of Engineering Sciences.
Press, 1991. 31. Sanders R. Torsional elasticity of human skin in vivo.
12. Manschot JFM. The mechanical properties of human skin Pflüg Arch 1973; 342: 255–260.
in vivo. Nijmegen: Radboud University, 1985. 32. Agache P, Monneur C, Leveque J, et al. Mechanical
13. Laurent A, Mistretta F, Bottigioli D, et al. Echographic properties and Young’s modulus of human skin in vivo.
measurement of skin thickness in adults by high fre- Arch Dermatol Res 1980; 269: 221–232.
quency ultrasound to assess the appropriate microneedle 33. Leveque J, De Rigal J, Agache P, et al. Influence of age-
length for intradermal delivery of vaccines. Vaccine 2007; ing on the in vivo extensibility of human skin at a low
25: 6423–6430. stress. Arch Dermatol Res 1980; 269: 127–135.
14. Ebling F, Eady R and Leigh I. Anatomy and organiza- 34. Fong S, Hung L and Cheng J. The cutometer and ultra-
tion of human skin. In: Champion RH, Burton JL and sonography in the assessment of postburn hypertrophic
Ebling FJG (eds) Textbook of dermatology. London: scar—a preliminary study. Burns 1997; 23: S12–S18.
Blackwell, 1992, p.3160. 35. Jemec GB, Selvaag E, Ågren M, et al. Measurement of
15. Wilkes G, Brown I and Wildnauer R. The biomechanical the mechanical properties of skin with ballistometer and
properties of skin. CRC Crit Rev Bioeng 1973; 1: suction cup. Skin Res Technol 2001; 7: 122–126.
453–495. 36. Pedersen L, Hansen B and Jemec G. Mechanical proper-
16. Oxlund H, Manschot J and Viidik A. The role of elastin ties of the skin: a comparison between two suction cup
in the mechanical properties of skin. J Biomech 1988; 21: methods. Skin Res Technol 2003; 9: 111–115.
213–218. 37. Dobrev HP. A study of human skin mechanical proper-
17. Goldsmith LA. Biochemistry and physiology of the skin. ties by means of Cutometer. Folia Med 2002; 44: 5–10.
Oxford: Oxford University Press, 1991. 38. Caro-Bretelle A, Gountsop P, Ienny P, et al. Effect of
18. Burton JL and Cunliffe WJ. Subcutaneous Fat. In: sample preservation on stress softening and permanent
Champion RH, Burrington JL, Ebling FJG (eds) Text- set of porcine skin. J Biomech 2015; 48: 3135–3141.
book of Dermatology, 5th edition. New York, NY, USA: 39. Lanir Y and Fung Y. Two-dimensional mechanical prop-
Blackwell Scientific Publications, 1992. erties of rabbit skin. II. Experimental results. J Biomech
19. Brouwer I, Ustin J, Bentiey L, et al. Measuring in vivo 1974; 7: 171–182.
animal soft tissue properties for haptic modeling in surgi- 40. Martin PG. Properties of human skin. Charlottesville,
cal. In: Westwood JD, Hoffman HM, Mogel GT, et al. VA: University of Virginia, 2000.
(eds) Medicine meets virtual reality. Amsterdam: IOS 41. Nı́ Annaidh A, Bruyère K, Destrade M, et al. Characteri-
Press, 2001, pp.69–74. zation of the anisotropic mechanical properties of excised
20. Crichton ML, Donose BC, Chen X, et al. The viscoelas- human skin. J Mech Behav Biomed Mater 2012; 5:
tic, hyperelastic and scale dependent behaviour of freshly 139–148.
excised individual skin layers. Biomaterials 2011; 32: 42. Gallagher A, Nı́ Annaidh A and Bruyère K. Dynamic
4670–4681. tensile properties of human skin. In: Proceedings of the
21. Johnson K. Contact mechanics. London: Cambridge Uni- 2012 IRCOBI conference, Dublin, 12–14 September 2012.
versity Press, 1985. Zurich: International Research Council on the Biomecha-
nics of Injury.
Joodaki and Panzer 19

43. Karimi A, Rahmati SM and Navidbakhsh M. Mechani- rubber at low and high strain rates. Int J Impact Eng
cal characterization of the rat and mice skin tissues using 2006; 32: 1384–1402.
histostructural and uniaxial data. Bioengineered 2015; 6: 60. Khatyr F, Imberdis C, Vescovo P, et al. Model of the vis-
153–160. coelastic behaviour of skin in vivo and study of aniso-
44. Karimi A, Haghighatnama M, Navidbakhsh M, et al. tropy. Skin Res Technol 2004; 10: 96–103.
Measurement of the axial and circumferential mechanical 61. Piérard GE, Piérard S, Delvenne P, et al. In vivo evalua-
properties of rat skin tissue at different anatomical loca- tion of the skin tensile strength by the suction method:
tions. Biomed Tech 2015; 60: 115–122. pilot study coping with hysteresis and creep extension.
45. Karimi A, Navidbakhsh M, Haghighatnama M, et al. ISRN Dermatol 2013; 2013: 841217.
Determination of the axial and circumferential mechani- 62. Karimi A, Haghighatnama M, Shojaei A, et al. Measure-
cal properties of the skin tissue using experimental testing ment of the viscoelastic mechanical properties of the skin
and constitutive modeling. Comput Methods Biomech tissue under uniaxial loading. Proc IMechE, Part L: J
Biomed Engin 2015; 18: 1768–1774. Materials Design and Applications 2016; 230: 418–425.
46. Riemersa D and Schamhardt H. The cryo-jaw, a clamp 63. Flint M. The biological basis of Langer’s lines. In: Long-
designed for in vitro rheology studies of horse digital acre JJ and Thomas CC (eds) The ultrastructure of col-
flexor tendons. J Biomech 1982; 15: 619–620. lagen. Springfield, IL: Charles C Thomas, 1976,
47. Shi D, Wang D, Wang C, et al. A novel, inexpensive and pp.132–140.
easy to use tendon clamp for in vitro biomechanical test- 64. Cox H. The cleavage lines of the skin. Br J Surg 1941; 29:
ing. Med Eng Phys 2012; 34: 516–520. 234–240.
48. Ng B, Chou S and Krishna V. The influence of gripping 65. Langer K. On the anatomy and physiology of the skin. I.
techniques on the tensile properties of tendons. Proc The cleavability of the cutis. Br J Plast Sur 1978; 31: 3–8.
IMechE, Part H: J Engineering in Medicine 2005; 219: 66. Nı́ Annaidh A, Bruyere K, Destrade M, et al. Automated
349–354. estimation of collagen fibre dispersion in the dermis and
49. Velardi F, Fraternali F and Angelillo M. Anisotropic its contribution to the anisotropic behaviour of skin. Ann
constitutive equations and experimental tensile behavior Biomed Eng 2012; 40: 1666–1678.
of brain tissue. Biomech Model Mechanobiol 2006; 5: 67. Veronda D and Westmann R. Mechanical characteriza-
53–61. tion of skin—finite deformations. J Biomech 1970; 3:
50. Karimi A and Navidbakhsh M. Measurement of the uni- 111–124.
axial mechanical properties of rat skin using different 68. Liu Z and Yeung K. The preconditioning and stress
stress-strain definitions. Skin Res Technol 2015; 21: relaxation of skin tissue. J Biomed Pharm Eng 2008; 2:
149–157. 22–28.
51. Karimi A, Faturechi R, Navidbakhsh M, et al. A non- 69. Graf BK, Vanderby R, Ulm MJ, et al. Effect of precon-
linear hyperelastic behavior to identify the mechanical ditioning on the viscoelastic response of primate patellar
properties of rat skin under uniaxial loading. J Mech Med tendon. Arthroscopy 1994; 10: 90–96.
Biol 2014; 14: 1450075. 70. Fung Y-C. Bioviscoelastic solids. In: Fung Y-C (ed.) Bio-
52. Maurel W, Wu Y, Thalmann D, et al. Biomechanical mechanics. New York: Springer, 1993, pp.242–320.
models for soft tissue simulation. Berlin; Heidelberg: 71. Jansen L and Rottier P. Some mechanical properties of
Springer, 1998. human abdominal skin measured on excised strips. Der-
53. Daly CH. Role of elastin in the mechanical behavior of matologica 1958; 117: 65–83.
human skin. In: Proceedings of the 8th international con- 72. Dunn MG and Silver FH. Viscoelastic behavior of
ference on medicine and biology engineering; 22nd annual human connective tissues: relative contribution of viscous
conference on engineering in medicine and biology, Chi- and elastic components. Connect Tissue Res 1983; 12:
cago, IL, USA, 20–25 July 1969. Chicago, IL, USA: Con- 59–70.
ference Committee. 73. Vogel H. Age dependence of mechanical and biochemical
54. Barbenel J, Evans J and Finlay J. Stress-strain-time rela- properties of human skin. I: stress-strain experiments,
tions for soft connective tissues. In: Kenedi RM (ed.) Per- skin thickness and biochemical analysis. Bioeng Skin
spectives in biomedical engineering. Berlin: Springer, 1973, 1987; 3: 67–91.
pp.165–172. 74. Haut R. The effects of orientation and location on the
55. Ridge M and Wright V. Mechanical properties of skin: a strength of dorsal rat skin in high and low speed tensile
bioengineering study of skin structure. J Appl Physiol failure experiments. J Biomech Eng 1989; 111: 136–140.
1966; 21: 1602–1606. 75. Dombi GW, Haut RC and Sullivan WG. Correlation of
56. Markenscoff X and Yannas I. On the stress-strain rela- high-speed tensile strength with collagen content in con-
tion for skin. J Biomech 1979; 12: 127–129. trol and lathyritic rat skin. J Surg Res 1993; 54: 21–28.
57. Eshel H and Lanir Y. Effects of strain level and proteo- 76. Arumugam V, Naresh M and Sanjeevi R. Effect of strain
glycan depletion on preconditioning and viscoelastic rate on the fracture behaviour of skin. J Biosciences 1994;
responses of rat dorsal skin. Ann Biomed Eng 2001; 29: 19: 307–313.
164–172. 77. Greven H, Zanger K and Schwinger G. Mechanical prop-
58. Agache P. Collection explorations fonctionnelles erties of the skin of Xenopus laevis (Anura, Amphibia). J
humaines. In: Agache P (ed.) Physiologie de la peau et Morphol 1995; 224: 15–22.
explorations fonctionnelles cutanées. Paris: Tec Doc 78. Ankersen J, Birkbeck A, Thomson R, et al. Puncture
Lavoisier, 2000. resistance and tensile strength of skin simulants. Proc
59. Shergold OA, Fleck NA and Radford D. The uniaxial IMechE, Part H: J Engineering in Medicine 1999; 213:
stress versus strain response of pig skin and silicone 493–501.
20 Proc IMechE Part H: J Engineering in Medicine 00(0)

79. Silver FH, Freeman JW and DeVore D. Viscoelastic 98. Hendriks F, Brokken D, Oomens C, et al. The relative
properties of human skin and processed dermis. Skin Res contributions of different skin layers to the mechanical
Technol 2001; 7: 18–23. behavior of human skin in vivo using suction experi-
80. Jacquemoud C, Bruyere-Garnier K and Coret M. Meth- ments. Med Eng Phys 2006; 28: 259–266.
odology to determine failure characteristics of planar soft 99. Tregear RT. Physical functions of skin. London: Aca-
tissues using a dynamic tensile test. J Biomech 2007; 40: demic Press, 1966.
468–475. 100. Delalleau A, Josse G, Lagarde JM, et al. A nonlinear
81. Zhou B, Xu F, Chen C, et al. Strain rate sensitivity of elastic behavior to identify the mechanical parameters of
skin tissue under thermomechanical loading. Philos Trans human skin in vivo. Skin Res Technol 2008; 14: 152–164.
A Math Phys Eng Sci 2010; 368: 679–690. 101. Ogden R. Large deformation isotropic elasticity—on
82. Lim J, Hong J, Chen WW, et al. Mechanical response of the correlation of theory and experiment for incompres-
pig skin under dynamic tensile loading. Int J Impact Eng sible rubberlike solids. P Roy Soc Lond A Mat 1972;
2011; 38: 130–135. 326: 565–584.
83. Ottenio M, Tran D, Nı́ Annaidh A, et al. Strain rate and 102. Shergold OA and Fleck NA. Mechanisms of deep pene-
anisotropy effects on the tensile failure characteristics of tration of soft solids, with application to the injection
human skin. J Mech Behav Biomed Mater 2015; 41: and wounding of skin. P Roy Soc Lond A Mat 2004;
241–250. 460: 3037–3058.
84. Minns R, Soden P and Jackson D. The role of the fibrous 103. Flynn C, Taberner AJ, Nielsen PM, et al. Simulating the
components and ground substance in the mechanical three-dimensional deformation of in vivo facial skin. J
properties of biological tissues: a preliminary investiga- Mech Behav Biomed Mater 2013; 28: 484–494.
tion. J Biomech 1973; 6: 153–165. 104. Lapeer R, Gasson P and Karri V. Simulating plastic sur-
85. Alexander H and Cook T. Variations with age in the gery: from human skin tensile tests, through hyperelastic
mechanical properties of human skin in vivo. In: Kenedi finite element models to real-time haptics. Prog Biophys
RM and Cowden JM (eds) Bed sore biomechanics. Berlin; Mol Biol 2010; 103: 208–216.
Heidelberg: Springer, 1976, pp.109–117. 105. Drucker DC. A definition of stable inelastic material.
86. Belkoff S and Haut R. A structural model used to evalu- DTIC Document, September 1957, http://www.dtic.mil/
ate the changing microstructure of maturing rat skin. J dtic/tr/fulltext/u2/143756.pdf
Biomech 1991; 24: 711–720. 106. Tong P and Fung Y-C. The stress-strain relationship for
87. Tonge TK, Atlan LS, Voo LM, et al. Full-field bulge test the skin. J Biomech 1976; 9: 649–657.
for planar anisotropic tissues: part I—experimental meth- 107. Gasser TC, Ogden RW and Holzapfel GA. Hyperelastic
ods applied to human skin tissue. Acta Biomater 2013; 9: modelling of arterial layers with distributed collagen
5913–5925. fibre orientations. J R Soc Interface 2006; 3: 15–35.
88. Bader D and Bowker P. Mechanical characteristics of 108. Del Prete Z, Antoniucci S, Hoffman A, et al. Viscoelas-
skin and underlying tissues in vivo. Biomaterials 1983; 4: tic properties of skin in Mov-13 and Tsk mice. J Bio-
305–308. mech 2004; 37: 1491–1497.
89. Escoffier C, de Rigal J, Rochefort A, et al. Age-related 109. Fung Y-C. Stress-strain-history relations of soft tissues
mechanical properties of human skin: an in vivo study. J in simple elongation. In: Fung YC, Perrone N and Anli-
Invest Dermatol 1989; 93: 353–357. ker M (eds) Biomechanics: its foundations and objectives,
90. Elleuch K, Elleuch R and Zahouani H. Comparison of vol. 7. Englewood Cliffs, NJ: Prentice Hall, 1972,
elastic and tactile behavior of human skin and elastomeric pp.181–208.
materials through tribological tests. Polym Eng Sci 2006; 110. Flynn C and McCormack BA. Finite element modelling
46: 1715–1720. of forearm skin wrinkling. Skin Res Technol 2008; 14:
91. Iivarinen JT, Korhonen RK, Julkunen P, et al. Experi- 261–269.
mental and computational analysis of soft tissue stiffness 111. Flynn C, Taberner A and Nielsen P. Mechanical charac-
in forearm using a manual indentation device. Med Eng terisation of in vivo human skin using a 3D force-
Phys 2011; 33: 1245–1253. sensitive micro-robot and finite element analysis.
92. Crichton ML, Chen X, Huang H, et al. Elastic modulus Biomech Model Mechanobiol 2011; 10: 27–38.
and viscoelastic properties of full thickness skin charac- 112. Liu F, Li C, Liu S, et al. Effect of viscoelasticity on skin
terised at micro scales. Biomaterials 2013; 34: 2087–2097. pain sensation. Theor Appl Mech Lett 2015; 5: 222–226.
93. Grebeniuk L and Utenkin A. The mechanical properties 113. Wang Y, Marshall KL, Baba Y, et al. Compressive vis-
of human skin. I. Fiziologiia Cheloveka 1994; 20: 157. coelasticity of freshly excised mouse skin is dependent
94. Grahame R and Holt P. The influence of ageing on the on specimen thickness, strain level and rate. PLoS ONE
in vivo elasticity of human skin. Gerontologia 1969; 15: 2015; 10: e0120897.
121–139. 114. Flynn C and McCormack BA. Simulating the wrinkling
95. Barel AO, Courage W, Clarys P, et al. Handbook of non- and aging of skin with a multi-layer finite element
invasive methods and the skin. Boca Raton, FL: CRC model. J Biomech 2010; 43: 442–448.
Press, 1995. 115. Lokshin O and Lanir Y. Viscoelasticity and precondi-
96. Diridollou S, Vabre V, Berson M, et al. Skin ageing: tioning of rat skin under uniaxial stretch: microstruc-
changes of physical properties of human skin in vivo. Int tural constitutive characterization. J Biomech Eng 2009;
J Cosmet Sci 2001; 23: 353–362. 131: 031009.
97. Geerligs M, Van Breemen L, Peters G, et al. In vitro 116. Sverdlik A and Lanir Y. Time-dependent mechanical
indentation to determine the mechanical properties of behavior of sheep digital tendons, including the effects
epidermis. J Biomech 2011; 44: 1176–1181. of preconditioning. J Biomech Eng 2002; 124: 78–84.
 Joodaki and Panzer 21

117. Delalleau A, Josse G, Lagarde J-M, et al. Characteriza- linear mechanical behaviour of human skin. Skin Res
tion of the mechanical properties of skin by inverse anal- Technol 2003; 9: 274–283.
ysis combined with the indentation test. J Biomech 2006; 121. Zhang M, Zheng Y and Mak AF. Estimating the effec-
39: 1603–1610. tive Young’s modulus of soft tissues from indentation
118. Kvistedal Y and Nielsen P. Investigating stress-strain tests—nonlinear finite element analysis of effects of fric-
properties of in-vivo human skin using multiaxial load- tion and large deformation. Med Eng Phys 1997; 19:
ing experiments and finite element modeling. In: Pro- 512–517.
ceedings of the 26th annual international conference of the 122. Lapeer RJ, Gasson PD and Karri V. A hyperelastic
IEEE engineering in medicine and biology society, 2004 finite-element model of human skin for interactive real-
(IEMBS’04), San Francisco, CA, 1–5 September 2004, time surgical simulation. IEEE Trans Biomed Eng 2011;
pp.5096–5099. New York: IEEE. 58: 1013–1022.
119. Evans SL and Holt CA. Measuring the mechanical 123. Molinari E, Fato M, De Leo G, et al. Simulation of the
properties of human skin in vivo using digital image biomechanical behavior of the skin in virtual surgical
correlation and finite element modelling. J Strain Anal applications by finite element method. IEEE Trans
Eng 2009; 44: 337–345. Biomed Eng 2005; 52: 1514–1521.
120. Hendriks F, Brokken D, Van Eemeren J, et al. A
numerical-experimental method to characterize the non-

View publication stats