correction of axis misalignment in the analysis of knee rotations
TRANSCRIPT
Human Movement Science 22 (2003) 285–296
www.elsevier.com/locate/humov
Correction of axis misalignment in the analysisof knee rotations
F. Marin *, H. Mannel, L. Claes, L. D€uurselen
Institute of Orthopaedic Research and Biomechanics, University of Ulm, Helmholtzstr. 14,
89081 Ulm, Germany
Abstract
The Cardanic or Eulerian description is the most commonly used method for the descrip-
tion of in vivo knee rotation. It is based on the determination of external anatomical land-
marks used for the decomposition of the position of the tibia relative to the femur by three
rotations about three pre-defined axes. However, the in vivo localisation of external anato-
mical landmarks is known to be difficult and subjective. Even a small mislocalisation may lead
to dramatic consequences: the Cardanic description may become irreproducible and angle val-
ues may be overestimated. This error is well documented in the literature and known as the
‘‘cross-talk effect’’. Therefore this study proposes an additional calibration step of the classic
Cardanic description by a reorientation procedure of rotation axes. The procedure is based on
biomechanical constraints of knee kinematics as they appear during a knee squat exercise
using the finite helical axis (FHA) method and is independent of anatomical landmark. The
method was validated with the help of a special set-up modelling a perfect knee. Furthermore,
an inter-session reliability study was performed involving tests on two healthy subjects during
knee squat exercises. We found that the reorientation procedure was more reproducible than
the classic Cardanic description. We observed a maximum inter-session difference of 37.1� forthe adduction angle obtained with the classic Cardanic description. In contrast, the maximum
angle difference obtained with the reorientation procedure was less than 10�.� 2003 Elsevier B.V. All rights reserved.
PsycINFO classification: 2330
Keywords: Knee kinematics; Knee axis; Cross-talk effect; Finite helical axis; Reorientation
* Corresponding author. Tel.: +49-90-731-502-3492; fax: +49-0-731-502-3498.
E-mail address: [email protected] (F. Marin).
0167-9457/$ - see front matter � 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0167-9457(03)00036-8
286 F. Marin et al. / Human Movement Science 22 (2003) 285–296
1. Introduction
The most commonly used method to describe rotations of the knee joint is the
Eulerian or Cardanic description, which decomposes the position of the tibia with
regard to the femur (or vice versa) into three sequential rotations along three pre-defined axes named flexion/extension, abduction/adduction and internal/external
axial rotation (Grood & Suntay, 1983). These three pre-defined axes are also
known as anatomical axes and are defined by palpation of external anatomical
landmarks (Cappozzo, Catani, Croce, & Leardini, 1995). The flexion/extension
angle is identified as the rotation about the transepicondylar axis of the femur,
the abduction/adduction as the rotation about the floating axis, and the internal/
external axial rotation as the rotation about the long axis of the tibia. Thus, from
an anatomical and clinical point of view the attention focuses on a description ofthe rotations of the human knee joint. The Cardanic convention, although criti-
cised for its biomechanical conception of knee rotations (Woltring, 1994), is
nevertheless the one supported by the International Society of Biomechanics
(Wu & Cavanagh, 1995) and thus became the standard for the knee rotation de-
scription.
Criticism arose because the localisation of the anatomical points is difficult
(Blankevoort, Huiskes, & de Lange, 1988). Della Croce, Cappozzo, and Kerrigan
(1999) reported precision errors with six examiners for two subjects from 13 to25 mm for anatomical landmarks localisation. The consequences are misorientation
of the anatomical axes from 5� up to 13� (Blankevoort et al., 1988; Della Croce
et al., 1999; Piazza & Cavanagh, 2000; Ramakrishnan & Kadaba, 1991). But the
effects of such misorientation are dramatic as they lead to overestimation of the
abduction/adduction angle and internal/external rotation angle (Baker, Finney, &
Orr, 1999; Blankevoort et al., 1988; Ramakrishnan & Kadaba, 1991; Woltring,
1994). They can even scramble the interpretation of the screw home mechanism
(Hollister, Jatana, Singh, Sullivan, & Lupichuk, 1993; Piazza & Cavanagh, 2000).The misorientation of the axes is defined as the cross-talk effect (Chao, 1980; Ram-
sey & Wretenberg, 1999). It affects the kinematical analysis of motions in joints
that articulate principally with only one major component of motion. In the case
of the knee joint this means that if the axis of flexion/extension is not accurate
(Hollister et al., 1993) or not the same for more than one examiner, a flexion/ex-
tension motion will be cross-talked into the ad/abduction and the internal/external
rotations.
Knowing the clinical reality with different examiners and successive patient ex-aminations, this is a serious limitation for the use of the Cardanic representation,
Thus from a clinical point of view the use of this representation must be ques-
tioned because until now no effective solution has been proposed to counter this
error.
The aim of this study is to propose a procedure to obtain reliable knee axes with-
out anatomical landmarks by a reorientation of the anatomical coordinate system
based on biomechanical constraints.
F. Marin et al. / Human Movement Science 22 (2003) 285–296 287
2. Method
2.1. Reorientation procedure algorithm
For our method the reorientation was computed in three steps: first the compu-tation of the classic Cardanic representation with axes obtained from anatomical
landmarks, then the determination of the new flexion axis, and finally the reorienta-
tion.
2.1.1. Step 1: Cardanic representation with axes obtained from anatomical landmarks
As we wanted to find a method to counter the cross-talk effect, we first obtained a
Cardanic representation of the knee motions in order to have data to refer to. For
this in vivo knee movement investigation that we wanted to be non-invasive, we usedthe optoelectronic system Optotrak 3020 (Northern Digital, Toronto, Canada)
which permits markers comprising four LEDs to be tracked. One marker was at-
tached to the thigh permitting tracking of marker femur coordinate frame (mf). A
second marker on the shank was used to locate marker tibia coordinate frame (mt)
(Fig. 1).
Considering that for in vivo experiments in clinical environment, the hardest
constraint is time we consequently chose an ergonomic protocol. For this reason
Fig. 1. Marker localisation on the thigh (mf) and the shank (mt).
288 F. Marin et al. / Human Movement Science 22 (2003) 285–296
we used only three anatomical landmarks for defining the femur and the tibia
reference frame. Trochanter major, Epicondylus lateralis and medialis were used
to characterise the anatomical femur coordinate frame (af) and we defined the an-
atomical tibia reference co-ordinate (at) locally equal to the af at full knee exten-
sion (Marin, Allain, Diop, Maurel, & Lavaste, 1999). The orientation of the axesfollowed the recommendations of the ISB (Wu & Cavanagh, 1995), the Z-axis be-
ing the medio-lateral axis, the Y -axis the distal–proximal axis and the X -axis the
postero-anterior one. For the computing the hypothesis was as follows: Let Xfa,
Yfa and Zf
a be the axes of af . During the knee motion, af being supposed to be em-
bedded in the mf coordinate system, and at in mt. The relative position of at to af
was computed (Allard, Cappozzo, Lundberg, & Vaughan, 1998). The Cardanic
angle was deduced (Allard et al., 1998) and we obtained the angle ua for the
rotation about the Z-axis of af identified as flexion/extension, the angle wa forthe rotation about the X floating axis identified as ad/abduction, and the angle
ha for the rotation about the Y -axis of at identified as internal/external rotation.
2.1.2. Step 2: Determination of the new flexion axis
The reorientation procedure was performed using the method of the Finite Helical
Axes (FHA). Therefore with the relative position of at in af at each time, orientations
of the FHA of the displacement from the at in af were calculated. We used the pa-
rameters of Rodrigues (Bisshopp, 1969) extracted with the help of the Caley trans-formation (Mladenova, 1991). Then we determinated the mean orientation of the
FHAs between 40� and 80� while the subject was performing the squat movement
during its ascent phase and identified it as the mean flexion/extension axis (Mannel,
D€uurselen, Gillner, & Claes, 2000).
2.1.3. Step 3: Reorientation
Finally in step 3 we identified the precedent mean flexion/extension axis as the an-
atomical reoriented Z-axis of the femur (Zfr). For this purpose in a first step we de-
fined the anatomical initial reoriented reference coordinate of the femur (rf0) with
Xf0r ¼ ðYf
a � ZfrÞ=normðYf
a � ZfrÞ, Yf0
r ¼ Zfr � Xf0
r , and Zfr. Then rf0 was rotated with
an angle a along Zfr to define the intermediate reoriented reference coordinate of
the femur (rfa).
As in step 1, at full knee extension the intermediate reoriented reference coordi-
nate of the tibia (rta) has the same orientation as rfa. During the knee motion, rfa
is supposed to be embedded in the mf coordinate system, and rta in mt. The relative
position of rta to rfa is computed. The Cardanic angles obtained are angle ua (rota-tion along the Z-axis of f a), angle wa (rotation along the X floating axis), and ha (ro-tation along the Y -axis of ta).
The reoriented reference coordinates of the femur (rf) was then set equal to rfa
with the amplitude of wa minimal in the interval ua 2 ½0� 40�� and a 2 ½�30� 30��.Thus, ua ¼ ur is assimilated to the reoriented flexion/extension angle, wa ¼ wr to
the reoriented abduction/adduction angle, and ha ¼ hr to the reoriented internal/ex-
ternal rotation.
F. Marin et al. / Human Movement Science 22 (2003) 285–296 289
2.2. Experiments
In order to validate our results and to find out the inter-session variabilities of the
classic Cardanic angle in comparison to the reoriented one we performed the follow-
ing two experiments:
2.2.1. Validation
To validate our method by verifying the two reorientation criteria, we used a spe-
cial set-up which produced an accurate and well defined motion which was consid-
ered as our reference knee motion (Fig. 2).
For this set-up the mechanical linkage consisted of two segments connected by a
joint with two rotational degrees of freedom. The orthogonal rotation axes could be
exactly located in space each with two oppositely drilled chamferings. A special gearensured a coupling of axial rotation and flexion with a ratio of 15:40 during the first
40� of flexion and then only a pure hinge motion between 40� and 90� of flexion. TheCardanic angle obtained with the reference frame by palpation of drilled chamfer-
ings reflects the real motion of this mechanism. The reoriented Cardanic angles were
then computed and we calculated the maximum differences with the Cardanic angles
obtained.
2.2.2. Inter-session reliabilities
For the investigation on the inter-session reliability we had data of two healthy
subjects A and B (aged 28 and 31 years respectively). Subject A performed four
Fig. 2. Special set-up to validate the reorientation procedure.
290 F. Marin et al. / Human Movement Science 22 (2003) 285–296
sessions spaced over three days, and subject B three sessions spaced over six months.
Motions of the right knee during the knee squat exercise were recorded. The position
feet were in neutral position.
The Cardanic angles obtained on one side with the anatomical reference frame
with anatomical landmarks and those with the reoriented reference frame were com-pared. The maximal differences were computed between the four curves for the sub-
ject A and three curves for the subject B. The range of motion (ROM) of abduction/
adduction, internal/external rotation, reoriented abduction/adduction and reoriented
internal/external rotation was determined from 0� to 90� of flexion and reoriented
flexion respectively.
3. Results
3.1. Validation
Due to the design of the set-up, the positions of the axes of rotation were easy to
locate. Thus the Cardanic angle obtained using the landmark procedure described
exactly the motion of the mechanism (Fig. 3). Here we found a difference of less than
1� between the classic Cardanic angles and reoriented Cardanic angles.
Fig. 3. Cardanic angles (in degree) for the set-up. Graphs (a) and (c) are angle rotations deduced from
axes defined with localisation of axes of rotation. Graphs (b) and (d) are angle rotations deduced from
axes after the reorientation procedure.
F. Marin et al. / Human Movement Science 22 (2003) 285–296 291
3.2. Inter-session reliability
Concerning the two series for the inter-session reliability we found that for all re-
sults the flexion/extension angle and reoriented flexion/extension angle presented a
difference of only 1� to 2�. This suggests that the reorientation has the least effectson the flexion/extension quantity.
For subject A (Fig. 4), we observed a maximum difference of 37.1� for the abduc-tion/adduction, 17.6� for the internal/external rotation, but only a difference of 2.9�for the reoriented abduction/adduction and 9.5� for reoriented internal/external ro-
tation. Between the four sessions, the ROM of the abduction/adduction varied from
7.5� to 23.4� and the ROM of the internal/external rotation from 3.3� to 16.5�. On
the contrary, ROM of the reoriented Cardanic angle was more homogeneous with
the ROM of the reoriented abduction/adduction from 1.3� to 2.9� and ROMof the reoriented internal/external rotation from 2.1� to 3.1� (Table 1).
Fig. 4. Cardanic angles (in degree) for the subject A during knee squat exercise obtained from four ses-
sions. Graphs (a) and (c) are knee angle rotations deduced from axes defined themselves with anatomical
landmarks. Graphs (b) and (d) are knee angle rotations deduced from axes after the reorientation proce-
dure.
Table 1
Comparison between the Cardanic angles obtained with anatomical landmark and reoriented Cardanic
angles for the subject A and B
Subject A Subject B
Cardanic angle
with anatomical
landmarks
Reoriented
Cardanic
angle
Cardanic
angle with
anatomical
landmarks
Reoriented
Cardanic
angle
Maximal difference Ad/adduction 37.1� 2.9� 15.7� 3.3�Maximal difference internal/
external rotation
17.6� 9.5� 14� 4.3�
Maximum ROM Ad/adduction 23.4� 2.9� 9.5� 6.2�Minimum ROM Ad/adduction 7.5� 1.3� 2.3� 4.5�Maximum ROM internal/exter-
nal rotation
16.5� 3.1� 26.8� 2.2�
Minimum ROM internal/external
rotation
3.3� 2.1� 11.1� 1.6�
292 F. Marin et al. / Human Movement Science 22 (2003) 285–296
Moreover, the waveforms of the Cardanic angle curves obtained with the anato-
mical landmarks were radically different in between the four sessions. In contrast to
these results, the reoriented Cardanic angle curves showed the same characteristic for
all four sessions.
For subject B (Fig. 5), similar observations were made. We found that the max-
imal difference for the abduction/adduction angle is 15.7�, and 14� for the internal/
external rotation.
Differences for Cardanic angles obtained with the reoriented anatomical axis weresmaller with only 3.3� for the reoriented abduction/adduction angle and 4.3� for thereoriented internal/external rotation. Maximum and minimum ROM of abduction/
adduction was 9.5� and 2.3� respectively, and 26.8� and 11.1� for the ROM internal/
external rotation. The ROM of the reoriented Cardanic angles presented once again
less variation: between 4.5� and 6.2� for the reoriented abduction/adduction angle
and 1.6� and 2.2� for the reoriented internal/external rotation (Table 1).
4. Discussion
We found that the Cardanic angle obtained with anatomical landmarks was not
reproducible. Moreover, the value of ROM of the classical Cardanic angle obtained
using the landmark procedure gave data that were out of range from what can be
considered physiological values for healthy subjects (Woltring, 1994). These two ob-
servations are due to the misorientation of anatomical axes (cross-talk effect) which
may occur when there is more than one examiner or more than one examination.In contrast, the reorientation of the anatomical axes was more reproducible and
the values of the ROM of abduction/adduction or internal/external rotation never
exceeded 6.5� during the knee squat exercise. This result was the desired one
and was obtained by implementing of the reorientation procedure which relocates
Fig. 5. Cardanic angles for the subject B during a knee squat exercise obtained from three sessions.
Graphs (a) and (c) are knee angle rotations deduced from axes defined themselves with anatomical land-
marks. Graphs (b) and (d) are knee angle rotations deduced from axes after the reorientation procedure.
F. Marin et al. / Human Movement Science 22 (2003) 285–296 293
the position of the knee rotation axes. This method removes the errors introduced by
the mislocalisation of anatomical landmarks.
The idea of a reorientation of the anatomical reference coordinate has been pro-
posed by Woltring (1994), but it was only motivated by numerical considerations.Our reorientation method was motivated by biomechanical considerations resulting
from recent studies. We used our reorientation method during the ascent phase of
the squat exercise. We chose this motion as knee kinematics depend on the load con-
ditions (Asano, Akagi, Tanaka, Tamura, & Nakamura, 2001; Blankevoort, Huiskes,
& de Lange, 1990). In order to have a reproducible localisation of FHA, load con-
ditions must be reproducible. Therefore the knee squat is an excellent motion to
choose because of its ease and safety (Escamilla, 2001). It assures stability of the knee
(Jonsson & Karrholm, 1994) and reflects the active mechanisms of the knee joint.Moreover the ascent phase bounded the mechanical condition because the sub-
ject performed a motion against the gravity force, which can be considered as a con-
stant.
294 F. Marin et al. / Human Movement Science 22 (2003) 285–296
Secondly we defined the mean flexion axis the FHA between 40� and 80�. Recent
studies on the in vivo estimation of the contact point localisation between the femur
and the tibia during weight-bearing exercises (Asano et al., 2001; Hill et al., 2000)
showed that during 45� to 90� flexion the contact point on the tibia remains un-
changed. As the postural femoral condyles can be approximated with spheres ofthe same radius (Kurosawa, Walker, Abe, Garg, & Hunter, 1985), it can be assumed
that between 45� and 90� the in vivo loaded knee can be approximated as a hinge
joint. Consequently the motion is a pure flexion/extension. Moreover, an in vitro
study (Blankevoort et al., 1990; Wilson, Feikes, Zavatsky, & O�Connor, 2000) andthe in vivo study (Jonsson & Karrholm, 1994) using the FHA to characterise knee
kinematics have shown that FHAs are stable between 40� and 80�. In this range, ori-
entations of FHA are almost perpendicular to the sagittal plane. This was observed
in vivo for 13 subjects in the study of Jonsson and Karrholm (1994), in vitro for fourspecimens in the study of Blankevoort et al. (1990) and for 15 specimens in the pub-
lication of Wilson et al. (2000).
Finally we assumed that abduction/adduction quantity must be minimal during
the first 40� of flexion. Studies of Blankevoort et al. (1990) and Jonsson and Karrholm
(1994) have shown that between 0� and 40� of flexion the FHA tilts only in the frontal
plane and stays stable in the transverse plane of the femur. This indicates that the axis
of rotation has only components about the sagittal and the vertical axes of the femur
which can be interpreted as a flexion/extension and an internal/external rotation(Jonsson & Karrholm, 1994). This fact has indirectly been measured with the help
of the contact point investigation between the tibia and the femur obtained with
MRI (Hill et al., 2000) and with the biplanar image-matching technique (Asano et al.,
2001) under weightbearing conditions. Both studies showed that the contact point
in the lateral condyle moves backwards and in the medial condyle moves forwards
during the first 45� of flexion. The authors interpreted these motions as the combina-
tion of external rotation of the tibia and flexion. Indeed, if abduction/adduction was
present the contact point in the condyle could move laterally or medially as well.However, the reorientation method still presents two limitations due to practical
considerations. The first limitation concerns the problem of soft tissue movement
that we are always confronted with in the field of non invasive in vivo studies. We
must assume that the thigh marker and the shank marker provide an accurate image
of the motion of the tibia and the femur. We minimised the soft tissue error by put-
ting the thigh markers onto a large neoprene band which reduced the motion of the
soft tissues. The thigh markers are the most susceptible to affect the reliability of the
measurement. Moreover the subject was told to perform the knee squats withouthigh acceleration or deceleration.
The second limitation concerns the application of the reorientation concept on
pathological patients. The FHA stability and minimum abduction/adduction are cri-
teria based on analysis of healthy subjects only. Moreover, they are only useful to
describe a kinematical model of the squatting knee motion of a healthy subject.
To avoid a falsely adjusted application of the reorientation when dealing with patho-
logical subjects, we proposed to control two parameters: the mean deviation of
orientation of FHA between 40� and 80� and the existence of a minimum amplitude
F. Marin et al. / Human Movement Science 22 (2003) 285–296 295
of wa when a 2 ½�30� 30��. If the mean deviation of FHA is high, it indicates that the
motion between 40� and 80� of flexion is not pure flexion but coupled with other ro-
tation. If no minimum amplitude of wa is found, it suggests that the abduction/ad-
duction motion is not an artefact of the cross talk effect. In this case, we suggest
that specific criteria should be found to perform the reorientation on pathologicknees.
5. Conclusion
The reorientation of knee axes has been developed using a kinematical model of a
knee squat motion in order to correct axis misalignment (cross-talk effect) arising
with the classical landmarks method for obtaining the Cardanic angles. The reorien-tation criteria assume that the abduction/adduction is minimal during the first 40� offlexion, and that the flexion axis is identified between 40� and 80� of flexion during
the knee squat ascent phase. Thus the reorientation is independent of anatomical
landmarks. Inter-examiner and inter-session results confirm clearly that knee rota-
tion angles are reliable after reorientation.
Being independent of the examiner, the reorientation procedure offers the oppor-
tunity of objective knee kinematics follow-up measurements of patients with conser-
vative therapy or post-operative procedures with previous verification of two controlparameters. In the future specific reorientation criteria will be developed for specific
pathologies.
Acknowledgement
This work was supported by the German Research Council (Deutsche Fors-
chungsgemeinschaft, DFG DU254/2-2).
References
Allard, P., Cappozzo, A., Lundberg, A., & Vaughan, C. (1998). Three-dimensional analysis of human
locomotion. New York: John Wiley & Sons.
Asano, T., Akagi, M., Tanaka, K., Tamura, J., & Nakamura, T. (2001). In vivo three-dimensional knee
kinematics using a biplanar image-matching technique. Clinical Orthopaedics and Related Research,
388, 157–166.
Baker, R., Finney, L., & Orr, J. (1999). A new approach to determine the hip rotation profile from clinical
gait analysis data. Human Movement Science, 18, 655–667.
Bisshopp, K. E. (1969). Rodrigues� formula and the screw matrix. Journal of Engineering for Industry, 179–
185.
Blankevoort, L., Huiskes, R., & de Lange, A. (1988). The envelope of passive knee joint motion. Journal of
Biomechanics, 21, 705–720.
Blankevoort, L., Huiskes, R., & de Lange, A. (1990). Helical axes of passive knee joint motions. Journal of
Biomechanics, 23, 1219–1229.
Cappozzo, A., Catani, F., Croce, U. D., & Leardini, A. (1995). Position and orientation in space of bones
during movement: Anatomical frame definition and determination. Clinical Biomechanics (Bristol,
Avon), 10, 171–178.
296 F. Marin et al. / Human Movement Science 22 (2003) 285–296
Chao, E. Y. (1980). Justification of triaxial goniometer for the measurement of joint rotation. Journal of
Biomechanics, 13, 989–1006.
Della Croce, U., Cappozzo, A., & Kerrigan, D. C. (1999). Pelvis and lower limb anatomical landmark
calibration precision and its propagation to bone geometry and joint angles. Medical and Biological
Engineering and Computing, 37, 155–161.
Escamilla, R. F. (2001). Knee biomechanics of the dynamic squat exercise. Medicine and Science in Sports
and Exercise, 33, 127–141.
Grood, E. S., & Suntay, W. J. (1983). A joint coordinate system for the clinical description of three-
dimensional motions: Application to the knee. Journal of Biomechanical Engineering, 105, 136–144.
Hill, P. F., Vedi, V., Williams, A., Iwaki, H., Pinskerova, V., & Freeman, M. A. (2000). Tibiofemoral
movement 2: The loaded and unloaded living knee studied by MRI. Journal of Bone and Joint Surgery
(Br), 82, 1196–1198.
Hollister, A. M., Jatana, S., Singh, A. K., Sullivan, W. W., & Lupichuk, A. G. (1993). The axes of rotation
of the knee. Clinical Orthopaedics and Related Research, 290, 259–268.
Jonsson, H., & Karrholm, J. (1994). Three-dimensional knee joint movements during a step-up:
Evaluation after anterior cruciate ligament rupture. Journal of Orthopaedic Research, 12, 769–779.
Kurosawa, H., Walker, P. S., Abe, S., Garg, A., & Hunter, T. (1985). Geometry and motion of the knee
for implant and orthotic design. Journal of Biomechanics, 18, 487–499.
Mannel, H., D€uurselen, L., Gillner, S., & Claes, L. (2000). Invariant three dimensional analysis with regard
to human knee diagnostics. In Proceedings of 12th Conference of the European Society of
Biomechanics, Dublin.
Marin, F., Allain, J., Diop, A., Maurel, N., & Lavaste, F. (1999). On the estimation of the knee
kinematics. Human Movement Science, 18, 613–626.
Mladenova, C. (1991). Mathematical modelling and control of manipulator systems. Robotics &
Computer-Integrated Manufacturing, 8, 233–242.
Piazza, S. J., & Cavanagh, P. R. (2000). Measurement of the screw-home motion of the knee is sensitive to
errors in axis alignment. Journal of Biomechanics, 33, 1029–1034.
Ramakrishnan, H. K., & Kadaba, M. P. (1991). On the estimation of joint kinematics during gait. Journal
of Biomechanics, 24, 969–977.
Ramsey, D. K., & Wretenberg, P. F. (1999). Biomechanics of the knee: Methodological considerations in
the in vivo kinematic analysis of the tibiofemoral and patellofemoral joint. Clinical Biomechanics
(Bristol, Avon), 14, 595–611.
Wilson, D. R., Feikes, J. D., Zavatsky, A. B., & O�Connor, J. J. (2000). The components of passive knee
movement are coupled to flexion angle. Journal of Biomechanics, 33, 465–473.
Woltring, H. J. (1994). 3-D attitude representation of human joints: A standardization proposal. Journal
of Biomechanics, 27, 1399–1414.
Wu, G., & Cavanagh, P. R. (1995). ISB recommendations for standardization in the reporting of
kinematic data. Journal of Biomechanics, 28, 1257–1261.