Use of embedded strain gages for the in-vitro study of proximal tibial cancellous bone deformation during knee flexion-extension movement: development, reproducibility and preliminary results of feasibility after frontal low femoral osteotomy.

Abstract

This paper reports the development of an in-vitro technique allowing quantification of relative (not absolute) deformations measured at the level of the cancellous bone of the tibial proximal epiphysis (CB(TPE)) during knee flexion-extension. This method has been developed to allow a future study of the effects of low femoral osteotomies consequence on the CB(TPE).
Six strain gages were encapsulated in an epoxy resin solution to form, after resin polymerisation, six measurement elements (ME). The latter were inserted into the CB(TPE) of six unembalmed specimens, just below the tibial plateau. Knee motion data were collected by three-dimensional (3D) electrogoniometry during several cycles of knee flexion-extension. Intra- and inter-observer reproducibility was estimated on one specimen for all MEs. Intra-specimen repeatability was calculated to determine specimen's variability and the error of measurement. A varum and valgum chirurgical procedure was realised on another specimen to observed CB(TPE) deformation after these kind of procedure.
Average intra-observer variation of the deformation ranged from 8% to 9% (mean coefficient of variation, MCV) respectively for extension and flexion movement. The coefficient of multiple correlations (CMC) ranged from 0.93 to 0.96 for flexion and extension. No phase shift of maximum strain peaks was observed. Inter-observer MCV averaged 23% and 28% for flexion and extension. The CMC were 0.82 and 0.87 respectively for extension and flexion. For the intra-specimen repeatability, the average of mean RMS difference and the mean ICC were calculated only for flexion movement. The mean RMS variability ranged from 7 to 10% and the mean ICC was 0.98 (0.95-0.99). A Pearson's correlation coefficient was calculated showing that RMS was independent of signal intensity. For the chirurgical procedure, valgum and varum deviation seems be in agree with the frontal misalignment theory.
Results show that the methodology is reproducible within a range of 10%. This method has been developed to allow analysis the indirect reflect of deformation variations in CB(TPE) before and after distal femoral osteotomies. The first results of the valgum and varum deformation show that our methodology allows this kind of measurement and are encourageant for latter studies. It will therefore allow quantification and enhance the understanding of the effects of this kind of surgery on the CB(TPE) loading.

Full text

Page 1

TECHNICAL NOTEOpen Access
Use of embedded strain gages for the in-vitro
study of proximal tibial cancellous bone
deformation during knee flexion-extension
movement: development, reproducibility and
preliminary results of feasibility after frontal low
femoral osteotomy
Stéphane Sobczak1*, Patrick Salvia1, Pierre-Michel Dugailly2, Philippe Lefèvre2, Véronique Feipel2,
Serge Van Sint Jan1, Marcel Rooze1,2,3
Abstract
Background: This paper reports the development of an in-vitro technique allowing quantification of relative (not
absolute) deformations measured at the level of the cancellous bone of the tibial proximal epiphysis (CBTPE) during
knee flexion-extension. This method has been developed to allow a future study of the effects of low femoral
osteotomies consequence on the CBTPE.
Methods: Six strain gages were encapsulated in an epoxy resin solution to form, after resin polymerisation, six
measurement elements (ME). The latter were inserted into the CBTPEof six unembalmed specimens, just below the
tibial plateau. Knee motion data were collected by three-dimensional (3D) electrogoniometry during several cycles
of knee flexion-extension. Intra- and inter-observer reproducibility was estimated on one specimen for all MEs.
Intra-specimen repeatability was calculated to determine specimen’s variability and the error of measurement.
A varum and valgum chirurgical procedure was realised on another specimen to observed CBTPEdeformation after
these kind of procedure.
Results: Average intra-observer variation of the deformation ranged from 8% to 9% (mean coefficient of variation,
MCV) respectively for extension and flexion movement. The coefficient of multiple correlations (CMC) ranged from
0.93 to 0.96 for flexion and extension. No phase shift of maximum strain peaks was observed. Inter-observer MCV
averaged 23% and 28% for flexion and extension. The CMC were 0.82 and 0.87 respectively for extension and flexion.
For the intra-specimen repeatability, the average of mean RMS difference and the mean ICC were calculated only for
flexion movement. The mean RMS variability ranged from 7 to 10% and the mean ICC was 0.98 (0.95 - 0.99).
A Pearson’s correlation coefficient was calculated showing that RMS was independent of signal intensity. For the
chirurgical procedure, valgum and varum deviation seems be in agree with the frontal misalignment theory.
Conclusions: Results show that the methodology is reproducible within a range of 10%. This method has been
developed to allow analysis the indirect reflect of deformation variations in CBTPEbefore and after distal femoral
osteotomies. The first results of the valgum and varum deformation show that our methodology allows this kind of
measurement and are encourageant for latter studies. It will therefore allow quantification and enhance the
understanding of the effects of this kind of surgery on the CBTPEloading.
* Correspondence: ssobczak@ulb.ac.be
1Laboratory of Anatomy, Biomechanics and Organogenesis (LABO) (CP 619),
Faculty of Medicine, Université Libre de Bruxelles (ULB), Bruxelles, Belgium
Full list of author information is available at the end of the article
Sobczak et al. Journal of Orthopaedic Surgery and Research 2011, 6:12
http://www.josr-online.com/content/6/1/12
© 2011 Sobczak et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.

Page 2

Background
Valgus deformity of the knee is a well-known factor in
the aetiology of lateralized gonarthrosis. Following Pau-
wel’s theory [1], Maquet proposed to use frontal lower
limb realignment techniques [2-4]. However, in 25 to
30% of the cases, pain persists after treatment [5]. For
other authors [6], 3D correction of knee joint deforma-
tion could improve the outcome of the realignment
treatment. Only one experimental study on the effects
of high tibial osteotomies on knee joint kinematics and
muscular moment arms of thigh muscles has been pub-
lished [7]. Several studies [8-11] reported joint con-
straint distribution and contact area distribution
for various joints, various static positions and loading
conditions. The methods utilised do not allow tracking
of in-vivo femoro-tibial articular deformations during
continuous knee flexion-extension motion despite the
usefulness of these studies. In-vitro and in-vivo analysis
of dynamic joint deformation patterns is also still a chal-
lenge. The mechanical stress in bones cannot be mea-
sured in a living subject without the use of invasive
surgical procedures, due to obvious ethical concerns.
Indirect alternatives can be found in finite element mod-
elling generated from medical imaging [12]. Previous
studies [13-15] reported results on deformations of the
cancellous bone of the tibial proximal epiphysis (CBTPE)
in static conditions evaluating the mechanical character-
istics of different extracted bone portion in compression,
traction and/or torsion. This paper proposes an in-vitro
method to study the CBTPE deformations using
embedded strain gages during continuous knee motions.
Embedded strain gages were used previously by some
authors to measure hip prosthesis cement deformation
[16] or investigate the stress in the cement layer under-
lying tibial plateau [17]. The utilisation of embedded
strain gages is complex and difficult.
This study did not aim at analyzing the force transmis-
sion in the femoro-tibial joint compartment and did not
allow absolute CBTPE. The aim was the quantification the
indirect reflect in the CBTPEof tibial plateau loading
reported by the presented measurement method before
and after various kinds of osteotomies. With such data, it
will be possible to analyze the relative variations of the
local deformation and to increase our understanding of
the relationship between constraint patterns and overall
joint motion. Indeed, the real 3D impact of osteotomies
on tibial plateau loading is poorly reported in the litera-
ture [7]. The direct method presented here should pro-
vide innovative data on that particular topic.
Methods
Specimens and setting
Six fresh-frozen lower limbs (average age: 84 ± 9 years;
4 males, 2 females) were obtained from the ULB Body
Donation program. Thawing occurred at room tempera-
ture 24 hours before specimen preparation and experi-
ment. Each specimen included a full lower limb with its
hemi-pelvis. The pelvis and femur were rigidly mounted
on the experimental jig in an anatomical position
(Figure 1). The distal tendon of 8 muscles of interest
was carefully dissected and cut at their distal musculos-
keletal junction. Distal muscle attachments were left
intact. Muscles dissected were the rectus femoris (RF),
vastus lateralis (VL), vastus intermedius (VI), vastus
medialis (VM), biceps femoris (BF), semitendinosus
(ST), semimembranosus (SM), gracilis (Grac) and tensor
fasciae latae (TFL). All other muscles were kept intact.
Special care was given to respect the integrity of hip and
knee joint capsule and ligaments. One fishing wire (Sur-
flon®, Nylon coated, American Fishing Wire, 90 Lb.,
USA) was attached to each dissected tendon (Figure 1D)
by an orthopaedic surgeon according to Bull’s method
[18,19]. Each fishing wire ran proximally through tun-
nels drilled into the bone at the level of the related mus-
cle origin to allow joint loading following the
physiological muscle lines of action. Total loading was
300 N (RF + VM = 80 N; VL and VI = 60 N each; BF,
ST, SM, Grac, and TFL = 20 N each). Muscle loading
was selected to respect the forces that each muscle
could generate; this was determined from muscle
volume and muscle pennation angle [20]. The fishing
wire of selected muscles (RF, BF, ST, SM, Grac and
TFL) were attached to the mobile axis of six Linear
Variable Displacement Transductors (LVDT, Solartron
Metrology®, West Sussex, UK) to measure tendon
excursion during flexion-extension movement (Figure
1A). Two electrogoniometers [21-23] were used to col-
lect continuous femoro-patellar and femoro-tibial 3D
kinematics (Figure 1B). The results of tendon excursion
and full joint kinematics are not presented here.
Measurement Element (ME)
Six MEs were used in this study. Each ME included two
components: a strain gage (SG) and an epoxy resin cylin-
der. The strain gage was made from cupronickel alloy (60/
40) (Rosette unidirectional®, FLA-1-17, length: 3 mm, 120
Ω, TML, Tokyo, Japan) and was embedded in an epoxy
resin solution (LX 112®, Ladd Research Industries, Willis-
ton, USA) (Figure 2A). The resin was selected for its low
shrinkage properties during polymerisation to avoid
damage of the SG. Small polypropylene tubes (Ø: 4.7 mm)
were used as moulds to encapsulate the SG in the resin.
After polymerisation (1 day at 30°C and 1 day at 60°C), the
MEs were cooled down at room temperature.
Determination of epoxy resin Young’s modulus
Young’s modulus of the epoxy resin was determined to
define the influence of the epoxy resin on further
Sobczak et al. Journal of Orthopaedic Surgery and Research 2011, 6:12
http://www.josr-online.com/content/6/1/12
Page 2 of 10

Page 3

Figure 2 Representation and location of measurement element. Schematic view and location of MEs. A1 and A2: Schematic view of the ME.
A3: ME used in this study including a Strain Gage. B and C: 3D representation (obtained from medical imaging) of ME locations in the proximal
epiphysis of the tibia. ME1 and ME6: most anterior edge of the medial and lateral condyles, respectively; ME2: most medial point of the medial
condyle, ME3 and ME4 = most posterior edge of the medial and lateral condyles, respectively. M5: most lateral point of the lateral condyle
(tunnels were drilled 10 mm below these landmarks, see text for further details).
Figure 1 Schematic view of the experimental setting. Schematic view of the experimental setting showing a specimen mounted rigidly on
the experimental jig in anatomical position. A: representation of the LVDT placed in the prolongation of the action line of different muscles. B:
tibial 3D electrogoniometer. C: illustration of the loading muscles, RF, VL and VM by cerclage with metallic wires. D: representation of the fixation
of the other muscles according to the bull’s method.
Sobczak et al. Journal of Orthopaedic Surgery and Research 2011, 6:12
http://www.josr-online.com/content/6/1/12
Page 3 of 10

Page 4

measurement. Ten resin cylinders (L: 12 mm, H: 30
mm) were processed. Cylinders were placed in a servo-
hydraulic test system (MTS 858 Mini Bionix, 10-kN
loading head, Eden Prairie, Minnesota, USA). Compres-
sion force was applied along the longitudinal axis of the
cylinders and was released at the yield point (i.e., corre-
sponding to the limit of the cylinder elastic range).
Young’s modulus was then calculated in the elastic part
of the cylinder deformation [24]. The epoxy resin
Young’s modulus (E) was obtained by equation 1. The
average value was 2.09 ± 0. 03 GPa.
E
F
L
S
L
=
( / )
/

()
0
(1)
Where:
E = Young’s modulus (Pa)
F = Force applied (N).
S = cross-sectional area of cylinder (m2).
ΔL = Length variation (m).
L0 = Initial length (m).
Data acquisition
A custom-made device including six amplifying modules
was developed to collect the output from the 6 SGs
(Figure 2B). Each module included a Wheatstone bridge
which allowed the measurement of tension imbalance in
the bridge. These modules were supplied in differential
current (DC) with a floating power supply of 9V. Each
Wheatstone bridge was adjusted using an offset correc-
tion and the parasite noise related to external interfer-
ences were eliminated (common mode rejection). The
ratio between the signal power and the parasite noise
power (i.e., S/N ratio) was equal to 66 db. All modules
were connected to an acquisition board (DAP3200a,
Microstar Laboratories). The acquisition frequency was
208 Hz.
Definition of the strain gage deformation
The deformation (ε) was obtained by equation 2 [25].
This equation shows the relation between the resistance
variation (ΔR/R0) and the gage factor (GF). The gage
factor defines the SG deformation in a well-defined
direction. GFwas given by the gage manufacturer and
was equal to 2.15 ± 1%. This value was constant and
independent from the load applied.
 =ΔR Ro
GF
/
(2)
Where:
ε = Deformation
ΔR = Resistance variation.
Ro = Nominal resistance of SG (120Ω).
GF= Gage factor
Calibration
A custom-made device was built to measure the SG
length variation (ΔL) and to perform SG calibration. A
SG was embedded into a LX 112 epoxy resin plate (L =
30 mm, l = 12 mm, d = 3 mm) linked to a mechanism
(Figure 2E). The latter mechanism allowed stretching of
both resin plate and SG in a controlled way. The elon-
gation range was from 1 μm to 50 μm. Deformation
results were expressed in resistance variation to obtain a
calibration equation (Eq. 3) after Analog-to-Digital Con-
version (ADC). The final deformation ε was given by
equation 4. Note that this calibration was dependent on
the above-determined epoxy resin Young’s modulus (see
Eq. 1).
f r x
Δ ( )
x
e
(.)
=−
−
3 0 0002
6
(3)
e =
−
−
(.) /3 0 0002
G
6
e
F
x Ro
(4)
Where:
ε = Deformation
x = ADC variations
Ro = Nominal resistance of SG (120Ω).
GF= Gage factor
Location of MEs
ME location was standardized using computed tomogra-
phy (CT, Siemens SOMATRON, helical mode, slice
thickness = 0.5 mm, inter-slice spacing = 0.5 mm, image
format = DICOM 2.0) on each specimen before con-
straint experiments took place. CT data segmentation
and 3D-model reconstruction of bones were performed
using a dedicated software interface (Amira®, Visage
Imaging, Inc., San Diego, USA). Six tunnels (Ø: 4.2 mm)
were drilled into the cancellous bone underlying the
tibial plateau at 10 mm below the joint line. Tunnel
depths were 13 mm and strain gauges were placed
10 mm from the cortical bone (Figure 2 A1). ME loca-
tions were standardized using strict definitions describ-
ing anatomical landmark locations on the available 3D
models (Figure 2C). Tunnel orientation was parallel to
the cartilaginous surface of the tibia. Tunnel diameter
size was slightly smaller than the ME diameter to ensure
a tight fit with and a maximal contact between the MEs
and the cancellous bone. One ME was introduced into
each tunnel (Figure 2D), and no glue was used.
Experimental protocol
For each specimen, three repetitions of two cycles of flex-
ion-extension movement were performed. Measurement
Sobczak et al. Journal of Orthopaedic Surgery and Research 2011, 6:12
http://www.josr-online.com/content/6/1/12
Page 4 of 10

Page 5

started with the specimen knee in full extension main-
tained by the muscle loading. Flexion was then per-
formed manually by pushing with an open hand on the
distal part of anterior face of the leg. Once knee flexion
was obtained, the manual pressure was released and the
knee passively moved back by the muscle loading of the
quadricipital tendon. All above-mentioned muscle load-
ing was kept in place during the entire measurement ses-
sion. Intra- and inter-observer reproducibility was
analysed on one specimen (specimen 1). To allow repro-
ducibility analysis, three operators independently per-
formed the above data collection (three trials of three
repetitions including two motion cycles) on the same
specimen within a three-hour interval. Valgum and
Varum procedure were realised on another specimen.
These angular corrections (6° and 12°) were applied using
a specially-built control system. Before processing each
repetition for further analysis, the extension of the first
cycle and the flexion of the second cycle were selected to
avoid experimental noise (i.e., system oscillations) usually
observable at the beginning and end of movement. Data
were then normalized to flexion-extension range of
motion.
Statistical analysis
Inter- and intra-observer reproducibility was analyzed by
coefficients of multiple correlation (CMC) and mean
coefficient of variation (MCV). Intra-observer reproduci-
bility was obtained by comparing the mean of the 3
repetitions of the three trials for one operator. Inter-
observer reproducibility was performed by comparison
of the mean of the 3 repetitions performed by each
operator at 3 different times. For intra-specimen repeat-
ability we calculated the mean RMS difference between
the three repetitions for each ME. An ICC (3.1 Two-
Way Mixed) was also calculated. These were then aver-
aged across MEs and subjects. Pearson’s correlation
coefficients were calculated to analyse if the RMS was
dependent of data amplitude during movement. For val-
gum and varum deviations comparison, a faithful analy-
sis of the different patterns of curve was carried out.
Results
Intra-observer reproducibility
Good superposition of different repetitions was obtained
for all MEs (Figure 3). The CMC showed a mean value of
0.93 for flexion (range: 0.83-0.99) and of 0.96 for extension
Figure 3 Intra-operator reproducibility. Intra-operator variability during flexion and extension motion. ME 1: antero-medial strain gage; ME 2:
medial strain gage; ME 3: postero-medial strain gage; ME 4: postero-lateral strain gage; ME 5: lateral strain gage; ME 6: antero-lateral strain gage.
Sobczak et al. Journal of Orthopaedic Surgery and Research 2011, 6:12
http://www.josr-online.com/content/6/1/12
Page 5 of 10