ARTICLE IN PRESS
Journal of Biomechanics 36 (2003) 1391–1395
Short communication
A novel method for measuring medial compartment pressures within
the knee joint in-vivo
Iain A. Andersona,*, Andrew A. MacDiarmidb, M. Lance Harrisc, R. Mark Gilliesc,
Rhona Phelpsd, W.R. Walshc
a
Bioengineering Institute, University of Auckland, Floor 6, 70 Symonds Street, Auckland, New Zealand
b
Tauranga Hospital, Tauranga, New Zealand
c
Orthopaedic Research Laboratories, University of New South Wales, Prince of Wales Hospital, Sydney 2031, Australia
d
Industrial Research Ltd., Auckland, New Zealand
Accepted 31 March 2003
Abstract
A novel method for the measurement of knee joint forces in-vivo is described. A thin (0.2 mm) flexible electronic pressure sensor
was inserted through a narrow arthroscopic portal into the osteoarthritic medial compartment of the knee joint. The sensor partially
covered the load bearing area. The surgery was performed under local anaesthetic during normal arthroscopic examination
following patient consent. Results are presented for 11 patients. The method was used in a pilot study to assess the effects of four
valgus knee braces on medial compartment forces. An analysis of variance could not detect un-loading by any brace although there
were large variations in force output. These variations may be attributable to shifts in the sensor position. In-vivo measurement of
joint force is technically feasible.
r 2003 Elsevier Ltd. All rights reserved.
Keywords: Arthroscopy; Pressure sensing; In-vivo measurement; Knee joint; Knee brace; Osteoarthritis
1. Introduction
An electronic pressure sensor, implanted during
routine arthroscopy, was used for recording intraarticular forces within the medial compartment of the
knee. The sensor (Iscan 6900, Tekscan, Boston) consists
of a conductive ink grid comprising 121 sensing
elements, sandwiched between two layers of polyester
film. Other knee contact pressure studies have used Fuji
pressure sensitive film (Takahashi et al., 1997; Hayes
et al., 1993; Lee et al., 1997; Ihn et al., 1993) in ‘‘open’’
operations under general anaesthetic or cadaveric knees.
Fuji Film provides a single recording of the peak
pressure distribution. By contrast the Tekscan system
can provide a continuous output of pressure distribution, a desirable feature for a transducer to be implanted
in-vivo. Werner et al. (1995) suggest that errors in this
system could occur due to creep and temperature
*Corresponding author. Tel.: +64-9-3737599; fax: +64-9-3677157.
E-mail address: i.anderson@auckland.ac.nz (I.A. Anderson).
0021-9290/03/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0021-9290(03)00158-1
variation but indicates that its thinness (B0.1 mm) and
rapid response are valuable resources. The sensor is
reliable when used for recording joint pressures during
ligament balancing (Wallace et al., 1998) as well as
contact area and stress levels in total knee replacements
(Harris et al., 1999).
To evaluate this technique we have carried out a pilot
study to assess the influence of valgus knee braces on
medial compartment forces, to test the assertion that
braces unload this compartment. Other workers have
used imaging technologies to study brace efficacy.
Komistek et al. (1999) analysed fluoroscopic images on
patients wearing the Bledsoe Thruster brace and
measured an average change in condylar separation at
heel strike of 1.2 mm on 12 of 15 patients. Horlick and
Loomer (1993) could not see a significant change to
femoral-tibial angle or joint space on X-ray images of 39
patients with and without the Generation II brace fitted.
Matsuno et al. (1997) recorded a change of femoraltibial angle over 12 months from an average of 185.1 to
183.7 (1.4 ) on X-ray images from 12 out of 20 patients
ARTICLE IN PRESS
1392
I.A. Anderson et al. / Journal of Biomechanics 36 (2003) 1391–1395
wearing the Generation II brace. Load reduction has
been estimated indirectly using instrumented braces in
conjunction with gait analysis. Pollo et al. (2001)
calculated load reductions for the Generation II brace
of up to 17% on the medial side during the stance phase
of gait. Self et al. (2000) compared the moment of the
body about the knee with the counter-moment generated by the Donjoy Monarch brace to show that this
brace can significantly reduce the varus moment at the
knee at 20% and 25% of the stance phase. These
calculations cannot take into account the additional
influences of muscle action and ligamentous restraint.
Reliable in-vivo measurement of joint force would
overcome this problem and provide an important key to
our understanding of brace effectiveness. In the present
study our primary goal was to determine whether or not
the sensor will continue to work reliably after furling,
insertion through an arthroscopic portal and final
placement.
2. Methods
This study was subject to an independent ethics
committee and informed consent was obtained from all
patients. Measurements were performed under local
anaesthetic on three groups of patients during routine
arthroscopic surgery (Group I—patients 1–3, Group
II—patients 4–9 and Group III—patients 10–15). Joint
forces were measured using a single branch of an Iscan
6900 ‘‘quad’’ electronic pressure sensor of area 196 mm2
(Fig. 1). Cellulose leader tapes to aid in arthroscopic
placement were attached to top and bottom faces of
each sensor giving a total thickness of 0.2 mm. Sensors
were calibrated in compression within a Bionix 858
closed-loop servo-hydraulic mechanical testing system
Fig. 1. Tekscan transducer (surface area B2 cm2) held against a
femoral condyle. The medial condyle is on the right.
(MTS, Minneapolis, MN, USA). Each pressure sensor
and its leader tape were sterilised and sealed in a sterile
container. During surgery a mid-line viewing port was
established in the knee and the free end of the leader
tape was inserted through a portal on the medial side.
Forceps inserted through a lateral portal gripped the
tape. The surgeon pulled the tape through, drawing
the sensor into the joint space so that it lay across the
middle of the medial tibial plateau (Fig. 2). A probe was
used to assist in unfurling the sensor so that it lay flat. The
positioning procedure was monitored using an endoscope.
Each knee was sealed with a sterile plastic film.
For groups II and III the sensor signal response was
tested by manually loading and unloading the joint with
valgus and varus thrusts to the leg. Patients performed a
number of activities including double-leg stance and
single-leg stance (Fig. 3). Contact forces in the knee were
measured during unassisted stance and with up to 4
different commercial knee braces fitted to the affected
knee. The braces used were the Generation II Unloader
(Generation II, WA., USA), Donjoy Monarch (Smith
and Nephew, Ca., USA), Breg Tradition (Orthobionic
Inc., Ca., USA) and the V-Max (Bodyworks, Ca.,
USA). The braces were adjusted and fitted by a
professional orthotist. Ground reaction forces on the
affected side were measured using a stand-on load cell
(Kistler Instrumente AG, Winterthur, Switzerland,
model #9257B). Results are presented for 11 patients;
4 patients were excluded due to technical difficulties with
the sensor position (patients 11,13,15) or stand-on load
cell data (patient 1). Data was analysed using Tekscan
software (Tekscan, South Boston, USA) and Microsoft
Excel (Seattle, WA, USA).
The total force on the sensor, Fknee was calculated by
summing the response from all sensing elements for each
Fig. 2. The surgeon pulls the cellulose leader tape (left) out of the
lateral portal, thus drawing the Tekscan sensor into the medial portal.
ARTICLE IN PRESS
I.A. Anderson et al. / Journal of Biomechanics 36 (2003) 1391–1395
1393
each brace-on measurement (Fig. 4). Statistical analyses
were performed on Group III using an analysis of
variance (ANOVA) with a blocking structure to
incorporate the variability due to patient, brace-change,
and stance change influences on the sensor signal. The
experimental design was unbalanced for the ANOVA.
In particular, double-leg stance measurements were
taken either once (before single leg stance) or twice
(both before and after single leg stance) (refer Fig. 4). To
improve the experimental ‘‘balance’’ we used only the
first double-leg stance measurements for each no-brace
or brace-on measurement set.
We also performed a paired comparison t-test (single
sided Student’s) on adjacent brace-on and no-brace
measurements. The t-test difference was calculated by
averaging the two no-brace single or double leg stance
measurements that bracketed a brace-on measurement
and then subtracting the brace-on measurement. A small
value for the probability ‘p’ (o0.01) would support the
hypothesis, that significant unloading was taking place.
3. Results
Fig. 3. A single leg stance measurement is taken while the patient is
wearing a brace.
data frame and averaging data frames collected over
three 5 s intervals at a sample rate of at least 40 Hz. The
non-dimensional relative change in the signal Fstance
from single to double leg stance was calculated by
dividing the force for single leg stance (Fknee,sls) by the
average of adjacent double leg stance measurements
(Fknee,dls):
Fstance ¼ Fknee;sls =Fknee;dls :
The non-dimensional relative load (Frel) was the average
value of the force on the knee sensor with the brace on
(Fbrace) divided by the average value of the sensor force
for the no-brace condition (Fno brace):
Frel ¼ Fbrace =Fno
brace :
Before calculating Frel, both Fbrace and Fno brace were
normalised by dividing each by the total force through
the leg measured by the Kistler stand-on load cell. Frel
was expected to vary from Frel=0 representing complete
unloading by the brace to Frel=1 representing no
change in loading.
For Groups I and II no-brace stance measurements
were collected at the beginning of the experimental
procedure. For Group III no-brace stance measurements were collected at the start, end and in-between
Sensors continued to work after insertion for all
patients although signal levels were very low on three
patients (patients 11, 13, 15). The relative change in
sensor signal output from double to single leg stance
(Fstance), calculated for 64 measurements (no-brace and
brace-on for 11 patients), was 1.97 (s.d. 1.44). The range
for Fstance was 0.65 to 8.91. The relative load for the four
braces (Frel) ranged from 0.57–0.88 for single leg stance
and from 0.74 to 0.85 for double leg stance. Averaged
across the four braces, Frel was 0.78 (s.d. 0.29) and 0.70
(s.d. 0.36) for double and single-leg stance, respectively.
For Group III (#10, 12, and 14), the sensor signal during
no-brace stance activities displayed a significant degree
of variation through the course of an experiment (Figs. 4
and 5). Signals were highest at the start of an experiment
(Fig. 4). The ANOVA did not detect a significant unloading by any brace (p=0.389 for the F-ratio of 1.088
on 4 and 20 degrees of freedom) and demonstrated high
variation between patients, when changing between
braces and changes to stance (single to double). The
paired comparison results for double leg stance were
p=0.47, n=11 (all braces) and p=0.3, n=2 (best brace).
For single leg stance the results were p=0.24, n=11 (all
braces) and p=0.06, n=2 (best brace).
4. Discussion
Repeated no-brace measurements, on Group III,
demonstrated that loading conditions on the sensor
were subject to change (Figs. 4 and 5) probably linked to
shifts in sensor position, as it settled in, and small
ARTICLE IN PRESS
1394
I.A. Anderson et al. / Journal of Biomechanics 36 (2003) 1391–1395
Fig. 4. Tekscan data for patient (a) 10, (b) 12, (c) 14. Knee transducer force levels are presented for the initial valgus (surgeon opens joint) and varus
(surgeon closes joint) thrusts (panel 1), no brace double leg (dls) and single leg (sls) stance measurements (panels 2, 4, 6, 8, 10), and brace-on double
and single leg stance measurements for each of four separate braces (panels 3, 5, 7, 9).
changes to the flexion angle of the knee from measurement to measurement. We could not track sensor
position as the sensors could not be seen reliably on
X-ray. Sensor position would influence load measurement as the area of the sensor in the joint space was
approximately one third or less than the total contact
area in the medial compartment of the knee. This
estimate is based on measurements of medial knee
contact area by Ihn et al. (1993). While it is known how
sensor materials covering the full area of contact will
influence local contact stresses (Wu et al., 1998), it is not
clear how a sensor covering less than the full area of
contact will influence these stresses. Thus we limited our
study to an analysis of total force on the sensor, not
pressure distribution. Tekscan offers larger sensors but
the leads on these designs are too wide for fitting
through narrow arthroscopic portals. Reducing lead
width may aid in future efforts.
Osteoarthritic joint surfaces are characterised by
regions of healthy cartilage, damaged cartilage and
eburnated bone. We postulate that small changes to the
angle of knee flexion from measurement to measurement
will introduce big changes to contact conditions on the
sensor. An additional source of variation would have
been due to subjective effects such as muscle activity by
the patient but this was not addressed in the current
study.
The ANOVA on blocked data from Group III
and the paired comparison results failed to demonstrate
a significant brace effect. This appears to contradict
the Frel data which suggested that unloading was
of the order of 20–30%. No-brace measurements
at the beginning of an experiment were generally
high. These measurements were only collected at the
start of each experiment for Groups I and II. The
‘‘apparent’’ unloading seen in the Frel data could
therefore be the result of settling-in of the sensor
material.
The results of this study demonstrate that joint forces
can be measured in-vivo although a number of issues of
ARTICLE IN PRESS
I.A. Anderson et al. / Journal of Biomechanics 36 (2003) 1391–1395
1395
Acknowledgements
This research was supported by a grant from The New
Zealand War Pensions Medical Research Trust Fund.
The authors would also like to thank Dr. Murray Smith
(Univ. of Auckland) for his help with the statistical
analysis and Dr. David Pang (Industrial Research Ltd.)
for help with the experimental work.
References
Fig. 5. No-brace (a) double leg stance and (b) single leg stance results
collected on five occasions through the course of the experiment on
Group III patients. The signal levels have been normalised relative to
the measurements of largest magnitude. The largest magnitude single
leg stance measurements occurred at the beginning of the experiment.
detail need to be addressed to improve measurement
repeatability. These issues would include accurate
determination of sensor position, reliable sensor fixation
and monitoring of the flexure angle of the knee during
measurement.
This transducer system can also be used for dynamic
force/pressure measurements. For instance, osteoarthritic knee braces provide relief for patients during
normal dynamic activities such as walking. Thus to
comprehensively explore knee brace efficacy future
in-vivo studies should be extended to detect whether
or not unloading takes place during walking and other
activities.
Harris, M.L., Morberg, P., Bruce, W.J.M., Walsh, W.R., 1999. An
improved method for measuring tibiofemoral contact areas in total
knee arthroplasty: a comparison of K-scan sensor and Fuji film.
Journal of Biomechanics 32, 951–958.
Hayes, W.C., Lathi, V.K., Takeuchi, T.Y., Hipp, J.A., Myers, E.R.,
Dennis, D.A., 1993. Patello-femoral contact pressures exceed the
compressive yield strength of UHMWPE in total knee replacements. Proceedings of the 39th Annual Meeting, Orthopaedic
Research Society, 15–18 February, San Francisco, pp. 421.
Horlick, S.G., Loomer, R.L., 1993. Valgus knee bracing for medial
gonarthrosis. Clinical Journal of Sport Medicine 3, 251–255.
Ihn, J.C., Kim, S.J., Park, I.H., 1993. In vitro study of contact area
and pressure distribution in the human knee after partial and total
menisectomy. International Orthopaedics (SICOT) 17, 214–218.
Komistek, R.D., Dennis, D.A., Northcut, E.J., Wood, A., Parker,
A.W., Traina, S.M., 1999. An in vivo analysis of the effectiveness
of the osteoarthritic knee brace during heel strike of gait. Journal of
Arthroplasty 14 (6), 738–742.
Lee, T.Q., Gerken, A.P., Glaser, F.E., Kim, W.C., Anzel, S.H., 1997.
Patellofemoral joint kinematics and contact pressures in total knee
arthroplasty. Clinical Orthopaedics 340, 257–266.
Matsuno, H., Kadowaki, K.M., Tsuji, H., 1997. Generation II knee
bracing for severe medial compartment osteoarthritis of the knee.
Archives of Physical Medicine and Rehabilitation 78, 745–749.
Pollo, F.E., Otis J.C., Backus, S.I., Warren, R.F., Wickiewicz, T.L.,
2001. Reducing compartment loads with valgus bracing in the
osteoarthritic knee. Proceedings of the American Orthopaedic
Society for Sports Medicine, 28 June–1 July, Keystone, Colorado.
Self, B.P., Greenwald, R.M., Pflaster, D.S., 2000. A biomechanical
analysis of a medial unloading brace for osteoarthritis in the knee.
Arthritic Care and Research 13 (4), 191–197.
Takahashi, T., Wadi, Y., Yamamoto, H., 1997. Soft-tissue balancing
with pressure distribution during total knee arthroplasty. Journal
of Bone and Joint Surgery B 79 (2), 235–239.
Wallace, A.L., Harris, M.L., Walsh, W.R., Bruce, W.J., 1998.
Intraoperative assessment of tibiofemoral contact stresses in total
knee arthroplasty. Journal of Arthroplasty 13 (8), 923–927.
Werner, F.W., Green, J.K., Fortino, M.D., Mann, K.A., Short, W.H.,
1995. Evaluation of a dynamic intra-articular contact pressure
sensing system. Proceedings of the 41st Annual Meeting, Orthopaedic Research Society, Orlando.
Wu, J.Z., Herzog, H., Epstein, M., 1998. Effects of inserting a
presensor film into articular joints on the actual contact mechanics.
Journal of Biomechanical Engineering 120, 655–659.