Faisal and Luo BioMed Eng OnLine (2017) 16:116

DOI 10.1186/s12938-017-0407-y BioMedical Engineering

OnLine

RESEARCH Open Access

Study of the variations of fall induced

hip fracture risk between right and left femurs
using CT‑based FEA
Tanvir R. Faisal1,2 and Yunhua Luo3*

*Correspondence:
[email protected] Abstract
3
Department of Mechanical Background: Hip fracture of elderly people—suffering from osteoporosis—is a
Engineering, Faculty
of Engineering, University severe public health concern, which can be reduced by providing a prior assessment
of Manitoba, Winnipeg, MB of hip fracture risk. Image-based finite element analysis (FEA) has been considered an
R3T 2N2, Canada effective computational tool to assess the hip fracture risk. Considering the femoral
Full list of author information
is available at the end of the neck region is the weakest, fracture risk indicators (FRI) are evaluated for both single-
article legged stance and sideways fall configurations and are compared between left and
right femurs of each subject. Quantitative Computed Tomography (QCT) scan datasets
of thirty anonymous patients’ left and right femora have been considered for the FE
models, which have been simulated with an equal magnitude of load applied to the
aforementioned configurations. The requirement of bilateral hip assessment in predict-
ing the fracture risk has been explored in this study.
Results: Comparing the sideways fall and single-legged stance, the FRI varies by
64 to 74% at the superior aspects and by 14 to 19% at the inferior surfaces of both
the femora. The results of this in vivo analysis clearly substantiate that the fracture is
expected to initiate at the superior surface of femoral neck region if a patient falls from
his/her standing height. The distributions of FRI between the femurs vary considerably,
and the variability is significant at the superior aspects. The p value (= 0.02) obtained
from paired sample t-Test yields p value ≤ 0.05, which shows the evidence of variability
of the FRI distribution between left and right femurs. Moreover, the comparison of FRIs
between the left and right femur of men and women shows that women are more
susceptible to hip fracture than men.
Conclusions: The results and statistical variation clearly signify a need for bilateral hip
scanning in predicting hip fracture risk, which is clinically conducted, at present, based
on one hip chosen randomly and may lead to inaccurate fracture prediction. This
study, although preliminary, may play a crucial role in assessing the hip fractures of the
geriatric population and thereby, reducing the cost of treatment by taking predictive
measure.
Keywords: Quantitative Computed Tomography, Fracture risk indicator, Hip fracture,
Single-stance, Sideways fall

Background
One of the fatal injuries observed in elderly people is hip fracture, which is the most seri-
ous complication of osteoporosis. Due to reduction in bone mass with age, a sideways

© The Author(s) 2017. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium,
provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and
indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdo-
main/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 2 of 17

fall from the standing height often triggers the hip fracture associated with a high
level of morbidity and mortality in elderly population. Hence, the elder people often
undergo screening to identify the quality of bone by densitometric technique, such as
Dual-energy X-ray absorptiometry (DXA) [1], which is recognized by the World Health
Organization (WHO). During a clinical densitometric screening, only one hip is typi-
cally scanned instead of bilateral hips for osteoporosis assessment, reporting a good cor-
relation between the bone mineral density (BMD) of left and right hips. However, an
accurate assessment of fracture risk is a vital step for initial screening, which can help
reducing fall induced hip fracture by providing and creating a prior awareness to the
patients.
Although DXA is primarily used as the reference method to measure BMD, the BMD
based fracture assessment lacks desired accuracy. For instance, half of the patients
who suffer from fracture have BMD above the conventional osteoporotic threshold [2].
Accepting the fact, the DXA is still used clinically concerning lower potential health risk
due to radiation dose and associated cost [1, 3], and only one hip is generally scanned
for the assessment. It is argued that scanning of bilateral hips requires additional time
to repositioning a patient and thereby, exposing the individual to unnecessary radia-
tion. Compared with the early/first generation scanning systems, the modern scanning
modalities are fast; and a single hip can be scanned in a few minutes. The scanning of
the second hip in an automated sequential bilateral hip mode adds only about 1–2 min,
and both hips can be scanned in less than 5 min after the patient is positioned correctly
[4–8]. Thus, the ability to perform bilateral hip scans may not be constrained by time
issues. Albeit, clinical practice is to scan patient’s “non-dominant hip” (determined by
asking the patient if he or she is right- or left- handed) irrespective of imaging modalities
to assess the hip fracture risk.
Several studies have been conducted on subjects with different race [9], sex [10], age
[11] to justify the requirement of scanning one or both femurs for assessing osteopo-
rosis to predict hip fracture [12, 13]. A number of studies have shown a significant cor-
relation between the two hip BMD measurements at femoral neck of normal Chinese
women [6], and the Caucasian women of age above 65 years [8]. On the contrary, no
significant difference has also been reported for the groups of 36 healthy women of the
United Kingdom and the USA [14, 15]. Although these studies show no systematic dif-
ferences between the hips, the correlation may not be held true for a larger population
of patients. However, researchers have also observed BMD variation between the femora
[14, 16, 17], where the variation may be originated due to unilateral hip disorders, arthri-
tis, hemiplegia and osteoporosis. It is still unclear whether osteoporotic hip fracture can
be assessed equally or not, conducting bilateral scanning of both left and right femora.
Nevertheless, the decision of scanning mostly depends on technologist’s personal pref-
erence and the physical location of the scanner, even though single leg scanning may
underestimate the osteoporosis and thereby the fracture [17].
Until today, to the best of our knowledge, only densitometric studies have been con-
ducted to investigate the necessity for mono or bilateral hip scanning. Due to the nature
of 2D imaging modality, bone mass measured by DXA fail to incorporate the essential
factors such as geometry, microarchitecture, and material properties of bone tissue
[18] as well as loading condition as a consequence of falling. Hence, it is important to
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 3 of 17

investigate the requirement of mono or bilateral hip assessment in predicting the risk
of hip fracture considering more robust imaging technique—Quantitative Computed
Tomography (QCT). The QCT can predict the strength of femoral bone more accurately
since the imaging modality can adequately capture the bone’s 3D anatomic structure.
The strength of a femoral bone largely depends on its 3D anatomic structure, which is
not correctly reflected in DXA-based FEA. The strength of the femoral bone predomi-
nantly depends on its geometry and structural property, the distribution of bone mate-
rial and its properties within the entire structure. Hence, QCT-based FEA can include
the factors that influence the hip/femoral fracture. However, a large number of CT based
3D FE analyses have been conducted on proximal femur to predict bone fracture risk,
where von Mises failure criterion based on distortion energy theory [19–23] has been
adopted regardless of its limitation of predicting other than yielding in isotropic duc-
tile materials. Therefore, the goal of this work is to set a criterion, which can lead us to
justify the requirement of mono or bilateral hip analysis for assessing hip fracture more
reliably using FEA. In this preliminary study, we only considered femoral neck region
for assessing and comparing the fracture risk indicator (FRI). In this work, the mechan-
ics of femoral neck fracture due to stress variations considering the maximum tensile
and compressive stresses generated at the neck owing to the single-stance and sideways
fall have been investigated on both left and right femora. The hip fracture risk has been
assessed in terms of FRI, η, a ratio of bone stress to strength. Therefore, the compari-
son of FRIs of left and right femora will justify the requirement of bilateral fracture risk
assessment.
The specific objectives of this paper are to: (1) conduct the patient-specific FEA of both
left and right femora for the single-stance and sideways fall configurations; (2) compare
the fracture risk indicators at the femoral neck of both the femora to investigate the jus-
tification of bilateral hip assessment using CT based FEA.

Subjects and methods

Subjects
In this study, QCT-scan images of 30 anonymous Canadian male and female have
been selected randomly. The CT dataset of both left and right femora of the patients
was obtained in Digital Imaging and Communications in Medicine (DICOM) format,
removing all personal information as required under human research ethics approval,
from the Great-West Life PET/CT Centre located at the Health Science Centre, Win-
nipeg, Canada. The mean age of the subjects was 63.7 ± 7.8 years, within the ranges
of 51–78 years. Most of the patients (63%) were 60 and above, and the rest were below
60. The mean weight of the patients was 82.7 ± 16.8 kg, which varied between 51.7 to
111.4 kg.

Imaging
The images were scanned by SIEMENS S5VB40B CT scan machine and acquired in the
Great-West Life PET/CT at Winnipeg. The acquisition and reconstruction parameters
were 120 kVp, 244 mAs, and image matrix of 512 × 512 pixels. Both high and low-res-
olution protocols with a slice thickness of 1 and 3 mm, respectively, were used with in
plane spatial resolutions varying between approximately 0.78 and 0.98 mm. However,
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 4 of 17

1 mm slice thickness is more suitable for constructing a 3D model of a femur from the
CT dataset by extracting the femur from pelvis, ensuring proper segmentation. The local
bone density—expressed in Hounsfield Unit (HU)—correlated with CT voxel results in
an inhomogeneous density distribution. A calcium hydroxyapatite calibration phantom
(Mindways Inc., Austin, TX, USA) was incorporated during the time of image acquisi-
tion to correct the scanner drift and to attain an accurate estimation of bone mineral
density.

FE modelling of 3D reconstructed bone

Medical image processing software-Mimics® 16.0 (Materialise N.V., Leuven, Bel-
gium)—has been used to generate the three-dimensional geometry of each femur from
its DICOM image dataset. The 2D slices are stacked and converted into a 3D model
using the interpolation algorithm embedded in Mimics. The 3D anatomy of a bone is
extracted through careful segmentation, excluding soft tissue prior to the generation of
a FE meshed model of the femur. The FE mesh is generated using the 3-matic module
in Mimics (Fig. 3b). The 4-node linear tetrahedral element has been used in this study.
The tetrahedral element can simulate irregular and complex geometric models such as
those produced from various CAD/CAM systems. The element has four nodes with six
degrees of freedom at each node: translations in the nodal x, y, and z directions and rota-
tions about the nodal x, y, and z directions.
The FE mesh quality was checked and edited by using the mesh processing tools of
Mimics. To investigate model convergence, FE models with different maximum element
edge lengths were created. The generated mesh is then exported to MATALB® 14.0 (The
MathWorks Inc, Natick, MA), where in-house built codes have been used to map mate-
rial properties and to conduct the FEA. For each FE model, the maximum von Mises
stress at the smallest femoral neck cross-section was calculated under the same loading
and boundary conditions. The maximum element edge length that produced converged
finite element solutions [24] was obtained and used in all the rest FE simulations. Hence,
an adequate mesh density has been conserved to achieve model convergence.

Material properties and its mapping on the FE model

Although bone is anisotropic [25, 26], the widely accepted existing empirical modulus-
density relationships consider bone to be isotropic in literature [22, 27–30]. In this work,
the properties of bone material have also been considered isotropic and inhomogene-
ous to commensurate with a reliable modeling approach. The material properties can be
mathematically derived from CT data [31–35] by correlating HU and CT gray values.
However, material models for bone either directly depend on the HU number [28, 36]
or on the apparent density [31–34, 36]. In this work, density based material models have
been considered to estimate Young’s modulus and compressive yield strength of a femur
[31, 32], and they are,
2.29
E = 10, 500ρash (MPa) (1)

1.88
ρash < 0.317 g/cm3 
  
137ρash (MPa),
σyC = 1.72 (2)
114ρash (MPa), ρash > 0.317 g/cm3
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 5 of 17

where ρash is the bone ash density and is related to the equivalent K2 HPO4 density
ρK2 HPO4 correlating HU by empirical equations [34, 35]. The tensile yield strength [37]
 

is

σyT = 0.8σyC (3)

A constant Poisson’s ratio of 0.4 is assumed for the bone material in the present analy-
sis [34, 38]. A detail procedure to obtain the isotropic inhomogeneous material prop-
erties can be found in [39] and the references cited there. A brief description of the
material mapping on to the FE mesh has been described below.
The material properties have been mapped following a node-wise approach [39]
instead of element-wise [40–42] in the current work. This mapping approach is benefi-
cial, because only one material definition is required depending on the assigned densities
and resulting in a cost-effective and time-saving calculation. Nodal coordinates of the FE
mesh are retrieved using the in-house built mapping algorithm. Each nodal coordinate
is determined from the stack of CT slices considering the location of transverse plane (x
and y coordinates of each pixel) and the axial height (z values), and the corresponding
material data are mapped on to the nodes. To accommodate error due to the conversion
of node coordinates, from the spatial coordinates (x, y, and z) into pixel indices (i and
j), the HU at the corresponding locations in the CT slices are averaged within a definite
zone surrounding the nodal coordinates. A longitudinal QCT image of both right and
left femora of a patient, corresponding 3D meshed models, and the distribution of iso-
tropic inhomogeneous material properties throughout the femoral bones are shown in
Fig. 1. It is to be noted that the patient specific distribution of modulus might vary con-
siderably depending on many factors such as age, sex, and even non-uniform bone loss
between femurs of each patient.

Fig. 1 a A QCT image showing right and left femurs of a patient including pelvis and muscle, b 3D FE
meshed models of right and left femora, c isotropic inhomogeneous material properties distributions in the
femora
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 6 of 17

Loading and boundary conditions

Two loading conditions— single-stance and sideways fall, have been adopted to assess
and compare the fracture risk of left with right hips. The fall typically results in the
development of a peak stress at the femoral neck. A vertically downward load is applied
to the superior surface of femoral head (Fig. 2b) for the single-stance configuration,
whereas a lateral load at the greater trochanter, while constraining the nodes of the
femoral head, simulates the sideways fall (Fig. 2c). The nodes at the distal condyles of
the femur are completely constrained in all three directions for both the configurations
[43–45]. Although the magnitude of load in one-legged stance is reasonably less than the
sideways fall in real life [46, 47], in this work, an equal magnitude of load is applied to
both the configurations to explore the effect of sideways fall, which frequently causes the
fracture by developing stress variations in the femoral neck region.

Results
Femoral neck is the most critical and vulnerable section, where tensile and compressive
stress becomes higher for both the loading configurations and is, therefore, considered in
the present work. The femoral neck undergoes constant bending during the one-legged
stance, developing tensile and compressive stress at the superior and inferior surfaces

Fig. 2 Boundary conditions of the QCT-based finite element modeling—a a QCT scan of a representative
subject’s left femur, b single-stance load, c sideways fall load
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 7 of 17

of the femoral neck (Fig. 2b). From structural viewpoint, downward force through the
femoral head causes tensile stress in the superior surface of the neck and compressive
stresses inferiorly, which result in highest stresses in sub-capital and mid-femoral neck
regions. On the contrary, the stress patterns are reversed in the neck region during the
sideways fall due to lateral impact on the greater trochanter as shown in Fig. 2c. The fall
consequently develops large compressive stress at the superior-posterior surface of the
neck and tensile stress in the inferior region. Therefore, depending on the state of load-
ing, the principal stresses, σ1 and σ3, are considered at the superior and inferior surfaces
of the femoral neck region. The ratio of the maximum tensile and compressive stresses
to appropriate tensile and compressive yield strengths given by Eqs. (2) and (3) have
been considered to predict the fracture risk, and the associated fracture risk indicators
are expressed as [39].

ηT = |σ1 | σYT (4a)


ηC = |σ3 | σYC (4b)

Figure 3a shows a comparison between (ηT )stance and (ηC )fall at the superior aspects of
the femoral neck, and Fig. 3b displays the fracture risk indicators, (ηC )stance and (ηT )fall, at

Fig. 3 a Comparison of (ηT )stance and (ηC )fall at the superior surface, b the comparison of (ηC )stance and
(ηT )fall at the inferior surface of femoral neck of right femur during single-stance and sideways fall configura-
tions
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 8 of 17

the inferior aspects of the femoral neck of the right femur for the one-legged stance and
sideways fall configurations, respectively. Figures 4a, b show the similar comparisons for
the left femur. It is apparent that (ηC )fall is significantly higher than (ηT )stance for both left
and right femora even the magnitude of applied load is same for both the loading configu-
rations. Comparing the sideways fall and single-legged stance, the fracture risk indicators
vary by 64 to 74% at the superior aspects and by 14 to 19% at the inferior surfaces of both
femurs. Therefore, the results of this in vivo analysis clearly substantiate that the fracture
is expected to initiate at the superior surface of femoral neck region if a patient experi-
ences the impact of fall; and the fracture at this surface is as well observed both clinically
and experimentally [48, 49]. The variations of FRI in the superior and inferior aspects of
left and right femurs are shown in Figs. 5 and 6, respectively. The variation of FRI in the
superior and inferior aspects of left and right femurs may reasonably manifest that the
location of the impact load and its direction are largely responsible for hip fracture in dif-
ferent magnitude between the femurs and obviously more severe in actual scenario as the
magnitude of fall load is much higher than the one-legged stance.
The statistical distribution of FRI and its variability during fall at the superior and infe-
rior aspects of left and right femurs have also been shown in Fig. 7. Figures 7a and b
show box and whisker plots of FRI at the superior and inferior aspects of the femoral
neck of left and right femurs, respectively. The upper and lower boundaries of the box

Fig. 4 a Comparison of (ηT )stance and (ηC )fall at the superior surface, b the comparison of (ηC )stance and
(ηT )fall at the inferior surface of femoral neck of left femur during single-stance and sideways fall configura-
tions
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 9 of 17

Fig. 5 Comparison of ηT and ηC during single-stance (a) and the comparison of ηC and ηT during sideways
fall (b) configurations at the superior and inferior surfaces of left femur

indicate upper (75th percentile) and lower (25th percentile) quartile, whereas the inter-
nal line indicates the median FRI; and the cross sign represents the mean FRI for each
dataset. The lines extending vertically from the boxes, known as whiskers, illustrate the
variability outside the upper and lower quartiles. The box and whisker plots in Fig. 7a
describe that the distribution of FRI at the superior aspect of left femur is left-skewed,
whereas the FRI of the right femur follows a near normal distribution. The distributions
of FRI at the superior surface of the two femurs are further tested and compared using
Anderson–Darling normality test in Fig. 8, which shows that the FRI distribution of left
femurs in Fig. 8a rejects normality (p ≤ 0.05), whereas the distribution in Fig. 8b exhibits
normality for the right femurs. Figure 7b also resembles near normal distributions of
FRIs at the inferior aspects of left and right femurs, respectively. The normality of the
FRI distribution at the inferior surface of left and right femurs is also supported by the
Anderson–Darling test shown in Fig. 8c and d, respectively. It is evident that the prob-
ability of fracture and its variability is significantly different between the two sides.

Discussion
The non-invasive procedure of developing a CT based FE model, considering 3D bone
geometry, material properties, and loading effect can be an excellent tool for estimating
a priori the in vivo hip fracture risk of geriatric population and osteoporotic patients.
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 10 of 17

Fig. 6 Comparison of ηT and ηC during single-stance (a) and the comparison of ηC and ηT during sideways
fall (b) configurations at the superior and inferior surfaces of right femur

The maximum and minimum principal stresses have been found to be the characteristic
of sideways fall in comparison with one-legged stance. Total body weight is supported by
a femur during the one-legged posture, whereas the sideways fall from a standing height
impacted on a higher load than the body weight. Hence, the simulations with different
amount of loads may be mimicking the two configurations but lack a conclusive correla-
tion between the resultant stress distributions [46, 47, 50, 51] for predicting the fracture
risk. Therefore, an equal magnitude of load is considered for both the configurations to
distinguish the effect of sideways fall from the stance in conjunction with bone quality
and thereby evaluating the fracture risk and its variation between the femurs, if there is
any.
The physiology of bone demonstrates that bone typically comprises different tensile
and compressive strength due to which bone loss preferentially occurs at the superior
aspect of the femoral neck [47]. The inferior cortex of the femoral neck is usually thicker
than the superior, and the thinner superior cortex could make the femoral neck suscep-
tible to failure, while being subjected to greater compressive stress in consequence of the
impact of fall. It is obvious that the effect of stress is higher at the weaker region—the
region with lower mass density—and is demonstrated in Table 1 for both types of femur.
For both left and right femurs, the fracture risk indicators, ηT and ηC are consider-
ably higher at the superior cortex, where more bone loss is generally observed, than the
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 11 of 17

Fig. 7 The distribution of FRI and its variability between left and right femurs during sideways fall at superior
aspects (a) and inferior aspects (b)

inferior one. The average variation of FRI during single-leg stance and sideways fall con-
figurations for left and right femora has been shown in Fig. 9. For the single-leg stance,
the average variations of FRI between the left and right femurs are 11% at the inferior
surface and 16% at the superior surface, whereas the FRIs vary between 17 and 27% for
the same during the sideways fall. However, the variation of FRI between the femurs of
an individual patient can be considerably high and may not be truly reflected by mean
variation. In our study, the maximum individual variation of FRI between the left and
right femurs has been found up to 91% at the superior surface and 50% at the inferior
surface. Since the lesser bone loss makes the inferior cortex thicker, the variation of
FRI is less in this location. On the contrary, the thinner superior cortex is perceptibly
responsible for higher variation of FRIs at the superior surfaces of the neck. The varia-
tion between the left and right femurs may be originated from the non-uniform osteope-
nia or osteoporosis. Obviously, both femora may not experience similar degree of bone
loss or not suffer osteoporosis in a similar manner. Moreover, any structural differences
in femur length, shape and size may also be responsible for such variation between the
femurs.
The FRIs between the left and right femur of men and women have also been com-
pared in Fig. 10. The FRIs at the superior aspects of the left and right femora of male
vary by approximately 15 and 11% during fall and one-legged stance, respectively. For
female subjects, these variations are approximately 23 and 11% for the respective load-
ing conditions. Since the inferior cortex is less likely to suffer bone loss, the mean FRIs
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 12 of 17

Fig. 8 Comparison of distributions of ηC at the superior surface of left (a) and right (b) femurs, and distribu-
tion of ηT at the inferior surface of left (c) and right (d) femurs using Anderson–Darling normality test

Table 1 Average variation of FRIs (ηT , ηC ) at the superior and inferior aspects of left
and right femora during single-leg stance and sideways fall configurations
Femur type Single-legged stance Sideways fall
Superior surface Inferior surface Superior surface Inferior surface

Left femur 1.83 ± 0.56 0.41 ± 0.13 2.26 ± 0.56 0.50 ± 0.21
Right femur 1.57 ± 0.29 0.46 ± 0.12 1.78 ± 0.42 0.60 ± 0.15

Fig. 9 Comparison of fracture risk indicators between left and right femur during single-stance and sideways
fall configurations at the superior and inferior surfaces of femoral neck region
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 13 of 17

Fig. 10 Comparison of FRI at the superior and inferior surfaces of femoral neck of the left and right femora of
male and female during single-stance and sideways fall configurations

are nearly equivalent and varied by less than 4% in this region. It is clearly evident that
the degeneration of bone between the femurs are not uniform that causes variations in
FRIs between the hips. Hence, the clinical practice of densiometric screening of only
one leg, considering the good correlation of BMD between the left and right femurs are
judgmental. The difference in FRI between male and female subjects during fall not only
helps us speculating the severity of the osteoporotic hip fracture in women health, but
the variation also corroborates the fact that women are more likely to suffer osteoporotic
hip fracture than men owing to age and physiological factors such as menopause. The
hormonal imbalance and loss of calcium may also affect the BMD loss in a nonuniform
manner more severely to women than men and consequently the hip fracture risk.
In addition to the distribution of FRI at left and right femurs (Figs. 7, 8), the strength
of association between the FRIs of the left and right femora due to fall has been statisti-
cally compared using paired sample t-Test. This test typically provides a measure of the
separation of data in any two sets and a measure of confidence if those data reflect real
(significant) differences or not. The p value of two sample paired t-Test of FRI at the
superior aspects of left (N = 30) and right (N = 30) femurs is 0.02 and p ≤ 0.05; and the
t-value is 2.43. Therefore, there is a statistically significant difference between FRIs of left
and right femora at the superior aspects, left femur (mean = 2.26, SD = 1.03) and right
femur (mean = 1.83, SD = 0.41), p ≤ 0.05. Therefore, we reject the null hypothesis that
there is no difference in FRI distribution between left and right femurs. On the contrary,
for the inferior aspects of the left and right femurs, we have found t value is 1.43 and p
value is 0.16, i.e. p ≥ 0.05. Therefore, we fail to reject the null hypothesis that there is no
difference in FRI at the inferior aspects between left and right femurs. As noted earlier,
the superior aspect is the most vulnerable location, where the fracture typically initiated
due to fall, and we are more concerned on assessing hip fracture at this location. A larger
population based study with specific age and sex group can give a better insight into the
justification of bilateral hip scanning.
The tension and compression of superior and inferior surfaces are also affected by
other factors such as the orientation of trabecular and femur, diameter and length of the
femoral neck, the angle of impact, and impact site, which also require further investiga-
tion to get an insight into the difference of FRI and its cause in left and right femora, if
any. Nevertheless, the current work on assessing hip fracture risk has been conducted
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 14 of 17

on continuum level, which is unable to give insight into the effect of bone microstruc-
ture that leads to bone yielding, damage, and fracture at the organ level [52]. Due to
aging, disease process and the influence of therapeutic agents, bone loss occurs. A more
rod-like structure of trabecular bone is, therefore, observed in the femoral neck, which
thereby is likely to be more susceptible to bending and buckling failure modes. Con-
nectivity density (trabecular connectivity) also decreases and influence the hip fracture
as well [53]. With ages, cortical pores fuse together to form giant pores [54, 55], which
could be considered a pivotal loss of cortical thickness and strength and leads to fracture
triggered by fall. However, the effect of bone microstructure in organ level fracture can
be captured by multiscale modeling, which is beyond the scope of current work. Moreo-
ver, it is not known if the subjects of the dataset considered in this study are healthy or
diagnosed with osteoporosis based on existing assessment criteria. A similar study with
known dataset of osteoporotic patients can provide us a more conclusive result of the
variability of the FRIs between the left and right hips and will be conducted in future.
The goal of the current work is to study the difference in FRI between the left and right
femur by investigating the effect of fall, comparing both single-stance and sideways fall
configurations, to justify the need for bilateral hip scanning for predicting hip fracture
risk. Comparing to BMD measurement time of 20 min in early 1980s with dual-pho-
ton absorptiometry (DPA), today’s imaging modalities take few minutes for a single hip
scanning in automated sequential bilateral hip mode and adds a minute or so for the
second hip. Hwang et al. found a considerable difference in femoral neck BMD between
the right and left sides [17], which would be inaccurate if the measurement was done
unilaterally. The left–right discordance has been found up to 14% at the femoral neck
of 28% of 537 subjects examined by Cole and Larson [56]. The left–right difference in
BMD exceeded the Smallest Detectable Difference (SDD) of 0.024 g/cm2 for 52.1% par-
ticipants of a total of 3481 patients at femoral neck [57]. The variation of left–right FRI
can be originated from lifestyle, physical activities, alcohol consumption and can be
influenced by the use of medication. Hence, the clinical significance of bilateral assess-
ment of predicting hip fracture is obvious. In general, the radiation dose (even this is
low now a days) as well as the cost restrict the bilateral hip scanning done clinically, but
the cost would be minimal as repositioning is often not required for imaging the second
femur. Moreover, the computational analysis will be an effective tool for assessing bilat-
eral hip fracture risk in terms of cost and time and will lead to a more efficient and clini-
cally accurate assessment. This preliminary study, however, shows the variability of FRIs
between the left and right femurs and justifies that bilateral hip scanning in predicting
hip fracture is necessary for more accurate prediction of hip fracture risk.

Conclusions
The QCT-based finite element modeling demonstrates an important route to assess hip
fracture to elderly people, who specially suffer from osteoporosis. A priory assessment of
hip fracture is a prerequisite to prevent the fracture to not only lower the treatment cost
but also to lessen the sufferings. The patient specific maximum principal stress based
fracture risk indices, ηC and ηT , have been evaluated for both left and right femurs to
investigate the correlation and/or difference between the FRIs of the femurs. The vari-
ability of FRI distribution and lower p value of two paired sample t-Test indicate strong
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 15 of 17

difference between the FRI of the two femurs during fall. However, the variation of FRI
of an individual patient could be even significantly large. Therefore, a bilateral hip scan-
ning is required to prevent, detect and to treat the hip fracture risk more accurately.
Authors’ contributions
TF analyzed and interpreted the patient data and performed the computational modeling. He also took part in the inter-
pretation of data and drafting of the manuscript. TF and YL contributed substantially to numerical modeling approach
and manuscript. YL revised the manuscript critically for important intellectual content. Both authors read and approved
the final manuscript.

Author details
1
Department of Physical Medicine and Rehabilitation, Feinberg School of Medicine, Northwestern University, Chicago, IL
60611, USA. 2 Legs + Walking Lab, Shirley Ryan AbilityLab, Chicago, IL 60610, USA. 3 Department of Mechanical Engineer-
ing, Faculty of Engineering, University of Manitoba, Winnipeg, MB R3T 2N2, Canada.

Acknowledgements
The authors would like to gratefully acknowledge the Great-West Life PET/CT Centre located at the Health Science
Centre, Winnipeg, Canada, for providing the CT dataset.

Competing interests
The authors declare that they have no competing interests.

Availability of data and materials

The data that support the findings of this study are available from Great-West Life PET/CT Centre but restrictions apply to
the availability of these data, which were used under license for the current study, and so are not publicly available. Data
are however available from the authors upon reasonable request and with permission of Great-West Life PET/CT Centre.

Consent for publication

Not applicable.

Ethics approval and consent to participate

All personal information was removed from the CT dataset as required under human research ethics approval, from the
Great-West Life PET/CT Centre located at the Health Science Centre, Winnipeg, Canada.

Funding
The authors acknowledge the funding supports from NSERC and Research Manitoba for the reported research.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Received: 6 March 2017 Accepted: 22 September 2017

References
1. Albertsson D, Mellström D, Petersson C, Thulesius H, Eggertsen R. Hip and fragility fracture prediction by 4-item clini-
cal risk score and mobile heel BMD: a women cohort study. BMC Musculoskelet Dis. 2010;11:55.
2. Marshall D, Johnell O, Wedel H. Meta-analysis of how well measures of bone mineral density predict occurrence of
osteoporotic fractures. BMJ. 1996;312:1254–9.
3. Adams JE. Advances in bone imaging for osteoporosis. Nat Rev Endocrinol. 2012;9:28–42.
4. Faulkner KG. Improving femoral bone density measurements. J Clin Densitom. 2004;6:353–8.
5. Franck H, Munz M, Scherrer M. Bone mineral density of opposing hips using dual energy X-ray absorptiometry in
single-beam and fan-beam design. Calcif Tissue Int. 1997;61:445–7.
6. Yang R-S, Tsai K-S, Chieng P-U, Liu T-K. Symmetry of bone mineral density at the proximal femur with emphasis on
the effect of side dominance. Calcif Tissue Int. 1997;61:189–91.
7. Huddleston AL, Rockwell D, Kulund DN, Harrison RB. Bone mass in lifetime tennis athletes. JAMA. 1980;244:1107–9.
8. Rao AD, Reddy S, Rao DS. Is there a difference between right and left femoral bone density? J Clin Densitom.
2000;3:57–61.
9. Xu H, Gong J, Chen J-X, Zhang T-M, Wu Q-L. Bilateral femoral bone mineral density measurements in Chinese
women and men. J Clin Densitom. 2007;10:165–9.
10. Keyak J, Sigurdsson S, Karlsdottir G, Oskarsdottir D, Sigmarsdottir A, Zhao S, Kornak J, Harris T, Sigurdsson G, Jonsson
B. Male–female differences in the association between incident hip fracture and proximal femoral strength: a finite
element analysis study. Bone. 2011;48:1239–45.
11. Schwarz P, Jørgensen NR, Jensen L, Vestergaard P. Bone mineral density difference between right and left hip during
ageing. Eur Geriatr Med. 2011;2:82–6.
12. Bonnick S, Nichols D, Sanborn C, Payne S, Moen S, Heiss C. Right and left proximal femur analyses: is there a need to
do both? Calcif Tissue Int. 1996;58:307–10.
13. Lopes JB, Danilevicius CF, Caparbo VF, Takayama L, Carvalho JF, Pereira RM. Effect of the bilateral hip bone density
measurement on clinical practice in elderly subjects. Maturitas. 2009;63:257–60.
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 16 of 17

14. Hall M, Heavens J, Ell P. Variation between femurs as measured by dual energy X-ray absorptiometry (DEXA). Eur J
Nucl Med. 1991;18:38–40.
15. Faulkner K, Genant H, McClung M. Bilateral comparison of femoral bone density and hip axis length from single and
fan beam DXA scans. Calcif Tissue Int. 1995;56:26–31.
16. Lilley J, Walters B, Heath D, Drolc Z. Comparison and investigation of bone mineral density in opposing femora by
dual-energy X-ray absorptiometry. Osteoporos Int. 1992;2:274–8.
17. Hwang HJ, Park SY, Lee SH, Han SB, Ro KH. Differences in bone mineral density between the right and left hips in
postmenopausal women. J Korean Med Sci. 2012;27:686–90.
18. Seeman E, Delmas PD. Bone quality—the material and structural basis of bone strength and fragility. N Engl J Med.
2006;354:2250–61.
19. Tsouknidas A, Anagnostidis K, Maliaris G, Michailidis N. Fracture risk in the femoral hip region: a finite element analy-
sis supported experimental approach. J Biomech. 2012;45:1959–64.
20. López E, Ibarz E, Herrera A, Mateo J, Lobo-Escolar A, Puértolas S, Gracia L. A mechanical model for predicting the
probability of osteoporotic hip fractures based in DXA measurements and finite element simulation. Biomed Eng
Online. 2012;11:84.
21. Keyak JH, Falkinstein Y. Comparison of in situ and in vitro CT scan-based finite element model predictions of proxi-
mal femoral fracture load. Med Eng Phys. 2003;25:781–7.
22. Keyak JH, Meagher JM, Skinner HB, Mote CD Jr. Automated three-dimensional finite element modelling of bone: a
new method. J Biomed Eng. 1990;12:389–97.
23. Lotz JC, Gerhart TN, Hayes WC. Mechanical properties of trabecular bone from the proximal femur: a quantitative CT
study. J Comput Assist Tomogr. 1990;14:107–14.
24. Zienkiewicz OC, Taylor RL. The finite element method for solid and structural mechanics. Oxford: Butterworth-heine-
mann; 2005.
25. Reilly DT, Burstein AH. The elastic and ultimate properties of compact bone tissue. J Biomech. 1975;8:393–405.
26. Ashman R, Van Buskirk W. The elastic properties of a human mandible. Adv Dental Res. 1987;1:64–7.
27. Rho JY, Ashman RB, Turner CH. Young’s modulus of trabecular and cortical bone material: ultrasonic and microten-
sile measurements. J Biomech. 1993;26:111–9.
28. Rho J, Hobatho M, Ashman R. Relations of mechanical properties to density and CT numbers in human bone. Med
Eng Phys. 1995;17:347–55.
29. Taddei F, Cristofolini L, Martelli S, Gill HS, Viceconti M. Subject-specific finite element models of long bones: an
in vitro evaluation of the overall accuracy. J Biomech. 2006;39:2457–67.
30. Trabelsi N, Yosibash Z, Milgrom C. Validation of subject-specific automated p-FE analysis of the proximal femur. J
Biomech. 2009;42:234–41.
31. Keyak J, Lee I, Skinner H. Correlations between orthogonal mechanical properties and density of trabecular bone:
use of different densitometric measures. J Biomed Mater Res. 1994;28:1329–36.
32. Keller TS. Predicting the compressive mechanical behavior of bone. J Biomech. 1994;27:1159–68.
33. Dalstra M, Huiskes R, Odgaard A, Van Erning L. Mechanical and textural properties of pelvic trabecular bone. J
Biomech. 1993;26:523–35.
34. Keyak JH, Rossi SA, Jones KA, Skinner HB. Prediction of femoral fracture load using automated finite element mod-
eling. J Biomech. 1998;31:125–33.
35. Dragomir-Daescu D, Den Buijs JO, McEligot S, Dai Y, Entwistle RC, Salas C, Melton LJ III, Bennet KE, Khosla S, Amin S.
Robust QCT/FEA models of proximal femur stiffness and fracture load during a sideways fall on the hip. Ann Biomed
Eng. 2011;39:742–55.
36. Snyder SM, Schneider E. Estimation of mechanical properties of cortical bone by computed tomography. J Orthop
Res. 1991;9:422–31.
37. Yosibash Z, Tal D, Trabelsi N. Predicting the yield of the proximal femur using high-order finite-element analysis with
inhomogeneous orthotropic material properties. Philos Trans A Math Phys Eng Sci. 2010;368:2707–23.
38. Lim TH, Hong JH. Poroelastic properties of bovine vertebral trabecular bone. J Orthop Res. 2000;18:671–7.
39. Faisal TR, Luo Y. Study of stress variations in single-stance and sideways fall using image-based finite element analy-
sis. Bio-Med Mater Eng. 2016;27:1–14.
40. Taddei F, Schileo E, Helgason B, Cristofolini L, Viceconti M. The material mapping strategy influences the accuracy
of CT-based finite element models of bones: an evaluation against experimental measurements. Med Eng Phys.
2007;29:973–9.
41. Helgason B, Taddei F, Pálsson H, Schileo E, Cristofolini L, Viceconti M, Brynjólfsson S. A modified method for assigning
material properties to FE models of bones. Med Eng Phys. 2008;30:444–53.
42. Chen G, Schmutz B, Epari D, Rathnayaka K, Ibrahim S, Schuetz MA, Pearcy MJ. A new approach for assigning bone
material properties from CT images into finite element models. J Biomech. 2010;43:1011–5.
43. Koivumäki JE, Thevenot J, Pulkkinen P, Kuhn V, Link TM, Eckstein F, Jämsä T. Ct-based finite element models can be
used to estimate experimentally measured failure loads in the proximal femur. Bone. 2012;50:824–9.
44. Nishiyama KK, Gilchrist S, Guy P, Cripton P, Boyd SK. Proximal femur bone strength estimated by a computationally
fast finite element analysis in a sideways fall configuration. J Biomech. 2013;46:1231–6.
45. Grassi L, Schileo E, Taddei F, Zani L, Juszczyk M, Cristofolini L, Viceconti M. Accuracy of finite element predictions in
sideways load configurations for the proximal human femur. J Biomech. 2012;45:394–9.
46. Robinovitch S, Hayes W, McMahon T. Prediction of femoral impact forces in falls on the hip. J Biomech Eng.
1991;113:366–74.
47. Yoshikawa T, Turner C, Peacock M, Slemenda C, Weaver C, Teegarden D, Markwardt P, Burr D. Geometric structure of
the femoral neck measured using dual-energy X-ray absorptiometry. J Bone Miner Res. 1994;9:1053–64.
48. de Bakker PM, Manske SL, Ebacher V, Oxland TR, Cripton PA, Guy P. During sideways falls proximal femur frac-
tures initiate in the superolateral cortex: evidence from high-speed video of simulated fractures. J Biomech.
2009;42:1917–25.
Faisal and Luo BioMed Eng OnLine (2017) 16:116 Page 17 of 17

49. Ota T, Yamamoto I, Morita R. Fracture simulation of the femoral bone using the finite-element method: how a
fracture initiates and proceeds. J Bone Miner Metab. 1999;17:108–12.
50. Courtney AC, Wachtel EF, Myers ER, Hayes WC. Age-related reductions in the strength of the femur tested in a fall-
loading configuration. J Bone Joint Surg. 1995;77:387–95.
51. Beck TJ, Ruff CB, Warden KE, Scott WW Jr, Rao GU. Predicting femoral neck strength from bone mineral data: a struc-
tural approach. Invest Radiol. 1990;25:6–18.
52. Marinozzi F, Marinozzi A, Bini F, Zuppante F, Pecci R, Bedini R. Variability of morphometric parameters of human
trabecular tissue from coxo-arthritis and osteoporotic samples. Annali dell’Istituto superiore di sanita. 2012;48:19–25.
53. Johnell O, Kanis J. An estimate of the worldwide prevalence, mortality and disability associated with hip fracture.
Osteoporos Int. 2004;15:897–902.
54. Chen H, Zhou X, Shoumura S, Emura S, Bunai Y. Age-and gender-dependent changes in three-dimensional micro-
structure of cortical and trabecular bone at the human femoral neck. Osteoporos Int. 2010;21:627–36.
55. Mayhew PM, Thomas CD, Clement JG, Loveridge N, Beck TJ, Bonfield W, Burgoyne CJ, Reeve J. Relation between age,
femoral neck cortical stability, and hip fracture risk. Lancet. 2005;366:129–35.
56. Cole R, Larson J. The effect of measurement of the contralateral hip if the spine is not included in the bone mineral
density analysis. J Clin Densitom. 2006;9:210–6.
57. Mounach A, Rezqi A, Ghozlani I, Achemlal L, Bezza A, El Maghraoui A. Prevalence and risk factors of discordance
between left- and right-hip bone mineral density using DXA. ISRN Rheumatol. 2012;2012:6.

Submit your next manuscript to BioMed Central

and we will help you at every step:
• We accept pre-submission inquiries
• Our selector tool helps you to find the most relevant journal
• We provide round the clock customer support
• Convenient online submission
• Thorough peer review
• Inclusion in PubMed and all major indexing services
• Maximum visibility for your research

Submit your manuscript at

www.biomedcentral.com/submit