Medical Engineering & Physics 26 (2004) 611–620

www.elsevier.com/locate/medengphy

Communication

Joint kinematics and spatial–temporal parameters of gait measured

by an ultrasound-based system
Rita M. Kiss a,
, László Kocsis b, Zsolt Knoll c
a
Academic Research Group of Structures, Bertalan L. u. 2., 1111 Budapest, Hungary
b
Biomechanical Laboratory, Department of Applied Mechanics, Budapest University of Technology and Economics,
Mu}egyetem rkp. 3., H-1521 Budapest, Hungary
c
MEDICaMENTOR Foundation, Pálos u. 3., H-1021 Budapest, Hungary

Received 24 February 2003; received in revised form 1 April 2004; accepted 20 April 2004

Abstract

Since measuring and recording techniques were developed, gait analysis has been frequently used in almost all fields of human
locomotion such as rehabilitation medicine, orthopaedics, sports science, and other related fields. The measuring range of usual
ultrasound-based devices is limited because the ultrasound sources must be always in visual contact with the microphones (mar-
kers). Our technique for the expansion of the measuring range is presented. Our approach is based on a mechanical axiom, which
states that the position and orientation of a segment of the human body is determined by an array of three points per segment.
The position of an invisible anatomical point of the segment could be determined by its position in relation to the fundamental
points, being in visible contact on the body segment. Before measurement, the position of investigated anatomical points in
relation to the fundamental points has to be given by an ultrasound-based pointer. The position of fundamental points of each
segment of the human body has to be measured during motion by the ultrasound-based device. A computer code calculates the
position of anatomical points from the above data on-line. This approach provides an opportunity to investigate a discretional
number and posture (lateral, medial, posterior and anterior) of anatomical points. Our model consists of 19 anatomical and
anthropometrical points. Based on the spatial coordinates of the anatomical points investigated, the spatial–temporal parameters
of gait and anatomical joint angles are estimated. No significant statistical difference was observed between the values presented
and those found in literature. Several clinical applications can be proposed such as monitoring of rehabilitation progress after
orthopaedic surgery and gait analysis in neurological diseases.
# 2004 IPEM. Published by Elsevier Ltd. All rights reserved.

Keywords: Gait analysis; Kinematics; Treadmill walking; Ultrasonic

1. Introduction length, cadence, velocity, etc.), forces and moments

Gait analysis can be described as a field of bio-
mechanical engineering dealing with the subject of
human locomotion. By means of different available
measuring techniques (for example, video recording),
the data of human gait are captured (i.e. the gait pat-
tern is recorded as a function of time) and further
analysis and calculation are done in order to obtain all
the data required for evaluating the quality of the
subject’s gait, including basic gait parameters (stride


Corresponding author. Tel.: +36-1-463-1738; fax: +36-1-463-1784.
E-mail address: kiss@vbt.bme.hu (R.M. Kiss).
1350-4533/$ - see front matter # 2004 IPEM. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.medengphy.2004.04.002
occurring in the joints, muscle activity during each gait
cycle, velocity and acceleration of each segment of the
limb, etc.
The mathematics used to achieve this goal are based
on Newtonian equations, whereas the human body is
considered to be a mechanical system of moving seg-
ments, driven by muscular, gravitational, inertial, and
reaction forces.
Since the measuring and recording techniques were
developed, gait analysis has been frequently used in
almost all fields of human locomotion [1].
Gait analysis can be successfully used in sport appli-
cations, post-injury assessment, disability evaluations,
612 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620

research analysis of injuries, industrial applications of

sports product design and improvement, etc.
The successful use of gait or motion analysis in the
diagnosis of patients with locomotor pathology and the
subsequent planning and assessment of treatment has
been limited because of its unreliability, particularly in
evaluating frontal and transverse plane components.
This is critical because in patients with pathological
gait, such as children with cerebral palsy, abnormalities
occur essentially in these planes [2]. In skin marker-
based gait analysis systems, skin movement artifacts
have been shown to affect the accuracy of calculated
joint kinematics much more in the frontal plane and
transverse planes than in the sagittal plane [3]. There-
fore, reduction of the effects of skin movement artifacts
in the two planes will improve the quality of gait analy- Fig. 1. The definition for calculation of the position of an optional
sis data for clinical purposes. point in the segment-embedded reference frame. xa1, ya1, za1, xa2, ya2,
The use of video or stereophotogrammetry in human za2, xa3, ya3, za3 are coordinates of active markers (known); x0, y0, z0
movement analysis requires determination of the pos- (during the calibration phase, x
0 , y
0 , z
0 ) are coordinates of an
optional point (unknown); q0 is the position vector of the optional
ition of active or passive markers on anatomical points
point in the segment-embedded reference plane (known); xS1, yS1, zS1,
before calculation of the kinematics and kinetics of xS2, yS2, zS2 are of active markers of pointer (known); e and f dis-
body segments. tances between the S1 and S2 active markers and between the S2
Our method is based on the following fundamental active marker and end of the pointer.
axioms:

1. The musculoskeletal system is generally modeled as

a multi-link chain with each body segment (pelvis,
thigh, shrank, foot) as a rigid link [4].
2. An array of three points per rigid body is needed
and is sufficient for the definition of a body-embed-
ded local reference frame, which represents the pos-
ition and the orientation of the rigid body. The
three points are named the fundamental points of
the local reference frame. The position of an ana-
tomical point on the segment could be determined
by its position in relation to the fundamental points.
Before the dynamic measurement is undertaken, the
position of investigated anatomical points in relation
to the fundamental points has to be recorded
(Fig. 1).
In our method, the three fundamental points of the
rigid body are the three markers per segment, deter-
mining the segment-embedded reference frame. The
position vectors of anatomical points in the segment-
embedded reference frame are determined before the
measurement during the so-called calibration phase.
The use of this approach provides an opportunity to
attach markers anywhere on a visible part of the seg-
ment and to analyze not just the visible lateral anatom-
ical points but also the medial, anterior, and posterior
anatomical points.
In this study, we describe a method for the esti-
mation of spatial–temporal parameters (step length,
stride length, walking base, etc.) and joint kinematics
during walking.
2. Method
2.1. Apparatus
The actual positions of three active markers attached
on the segment during walking and the position vector
of investigated anatomical points in the segment-
embedded reference frame should be recorded. A com-
mercially available ultrasonic device used for this
purpose is zebris CMS-HS, manufactured by ZEBRIS
Medizintechnik GmbH [5] (Tübingen, Germany).
The device consists of
1. a measuring head with three transmitter sensors,
2. triplets of ultrasonic receivers attached to a base one
on each segment, and
3. an ultrasonic pointer with two ultrasonic receivers
for determining the position vectors of the investi-
gated anatomical points of the segment in a seg-
ment-embedded reference frame (in relation to
fundamental points).
The transmitter sends out a burst of ultrasound, and
the delay it takes for this burst in reaching the receiver
is recorded. From this delay, the distances between the
transmitter and the receiver can be calculated from
D ¼ t  vs ð1Þ
where D is the distance in meters; t is the propagation
delay in seconds; and vs is the velocity of sound waves
 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620 613

in air in meter per seconds. Sound velocity can be

approximated in the air by
pffiffiffiffiffiffiffiffiffiffi pffiffiffiffi
vs ¼ jRT  20:05 T ð2Þ
where j is the isentropic coefficient which is j ¼ 1:40
in air; R is general gas constant which is R ¼
287:14 m2 =s2 K; T is the air temperature in degrees
Kelvin.

2.2. Determination of the position of investigated

anatomical points

The position and the orientation of a segment of the

human body are determined by the position of three
points per segment, named fundamental points. The
position of an anatomical point of the same segment
could be determined by its position in relation to the
fundamental points. This means that before the
measurement, the position of investigated anatomical
points should be determined in relation to the funda-
mental points. The position of fundamental points of
each segment of the human body has to be measured
during motion by the ultrasound device. A computer
code calculates—on-line—the position of investigated
points from the position of fundamental points using
the position vector of investigated points in the seg-
ment-embedded reference frame.
The three fundamental points of a segment are the
three active markers which are attached to the segment
using a polyester shell distributing the effect of the
muscles’ motion and checked for proper fixation and
stability (Fig. 1).
The ultrasound device measures the delay of ultra-
sound, which takes place for a burst from the transmit-
ter to reach the receiver, and which is recorded. From
this delay, the distances between the transmitter and
the receiver can be calculated using Eqs. (1) and (2).
The distances between all the three transmitters and the
active marker receiver can be determined with the same
method (Fig. 2). The coordinates of an active marker
can be calculated from
 2  where en , eg and ef are the unit vectors of the segment-
D1  D22
ya ¼ ð3Þ
4h
 2    from
D3  0:5 D21 þ D22 þ h2  l 2
za ¼ ð4Þ
2l
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ffi
xa ¼ D21  ðya þ hÞ2  z2a ð5Þ
where xa, ya and za are the spatial coordinates of an
active marker; Di is the distance between the ith trans-
mitter sensor and an active marker determined by
Eqs. (1) and (2); h and l are distances between the
transmitter sensors (Fig. 2).
Fig. 2. The definition of triangulation for calculation of the position
of the active marker. xa, ya, za are the global, spatial coordinates of
an active marker (unknown); D1, D2, D3 are the distances between
three transmitter sensors and an active marker (known), h and l the
distances between the transmitter sensors, xyz is the reference system
determined by the three transmitter sensors.
The easy calculation method above is repeated for all
active markers and for each sampling time. The coordi-
nates and the position vector of all active markers in
the global reference system are available during
motion. The position vector of an active marker is
noted as
ra ¼ rðxa ;ya ;za Þ ð6Þ
The position of an optional anatomical point (Fig. 1)
can be calculated by
r0 ¼ 0:5  ðra1 þ ra2 Þ þ q
0 ð7Þ
where r0 ¼ rðx0 ;y0 ;z0 Þ is the position vector of the
optional point in a global reference frame; rai ¼
rðxai ;yai ;zai Þ is the position vector of the ith active mar-
ker determined by Eqs. (3)–(5); q0 is the position vector
of the optional point in the segment-embedded refer-
ence frame determined during the calibration phase
before the measurement. The position vector can be
calculated by
q0 ¼ n0 en þ g0 eg þ f0 ef ð8Þ
embedded reference frame, which can be calculated
ra2  ra1
en ¼ ð9Þ
jra2  ra1 j
ra3  0:5  ðra1 þ ra2 Þ
eg ¼ ð10Þ
jra3  0:5  ðra1 þ ra2 Þj
e1 ¼ en  eg ð11Þ
where rai ¼ rðxai ;yai ;zai Þ is the position vector of the ith
active marker determined by Eqs. (3)–(5); n0, g0 and f0
are the scalar coordinates of the optional point in the
614 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620

segment-embedded reference frame, which can be cal-

culated from
  
n0 ¼ r
0  0:5  r
a1 þ r
a2  e
n ð12Þ


 
g0 ¼ r0  0:5  ra1 þ r
a2  e
g ð13Þ



f0 ¼ r0  0:5  ra1 þ r
a2  e
f ð14Þ

where r
ai ¼ rðxai ;yai ;zai Þ is the position vector of the ith
active marker determined by Eqs. (3)–(5) during the
calibration phase; e
n , e
g and e
f are the unit vectors of
the segment-embedded reference frame determined by
Eqs. (9)–(11) during calibration; r
0 ¼ rðx
0 ;y
0 ;z
0 Þ is the
position vector of the optional point during calibration
and determined by an ultrasound-based pointer. The
position vector can be calculated (Fig. 1) by

ðe þ f Þ
r
0 ¼ rS1 þ  ðrS2  rS1 Þ ð15Þ
e
where rSi ¼ rðxSi ;ySi ;zSi Þ is the position vector of the
ith active marker of the ultrasound-based pointer as
determined by Eqs. (3)–(5); e and f are the distances
between the active markers of the pointer and between
the active marker and pin of the pointer (Fig. 1).
By the described technique, any number of anatom-
ical points can be positioned to a measured triplet’s
active markers. The position of anatomical points can Fig. 3. Position of the anatomical points. (1) Right medial mal-
be calculated on-line and presented on the screen while leolus; (2) right heel; (3) right lateral malleolus; (4) right tibial
tubercle; (5) right head of fibula; (6) right lateral femoral epicondyle;
measurement is taking place. (7) right medial femoral epicondyle; (8) right greater trochanter; (9)
right ASIS; (10) left medial malleolus; (11) left heel; (12) left lateral
2.3. Biomechanical model malleolus; (13) left tibial tubercle; (14) left head of fibula; (15) left lat-
eral femoral epicondyle; (16) left medial femoral epicondyle; (17) left
greater trochanter; (18) left ASIS; (19) sacrum.
We used a human body model to initialize the pro-
cess of limb identification and location. Our model
refers to the 3D model defined by Winter [4] and modi- 2.4. Subjects
fied by Vaughan et al. [6] and by Bulgheroni et al. [7].
The study population consisted of 31 males (mean
The models investigated only the lateral anatomical
age 28:17  7:69 years, mean height 178:42  7:20 cm,
points of the human body.
mean mass 77:89  11:80 kg) and 20 females (mean age
Our mechanical approach provides an opportunity
25:09  4:21 years, mean height 168:07  5:70 cm,
to investigate a discretional number and the posture
(lateral, medial, posterior and anterior) of anatomical
points. Our model is composed of 19 anatomical points
identified by the position of fundamental points on seg-
ments and distances between the fundamental points
and the anatomical points. The medial and lateral mal-
leolus, the heel, the head of fibula and the tibial
tubercle are linked to three fundamental points on the
calf, the medial and lateral femoral epicondyle and
greater trochanter to three fundamental points on the
thigh, the left and right anterior superior iliac spine to
three fundamental points on the sacrum (Fig. 3). The
model is simple, and is adjustable for each person
studied.
mean mass 59:86  6:38 kg). For inclusion, subjects
were not to have any pathology that would affect gait
and had to be unfamiliar with treadmill walking. Each
subject provided an informed consent before partici-
pation.
2.5. Procedure
Subjects wore shorts and no shoes to allow access to
anatomical points of the lateral and medial malleolus
and heel. The subjects walked on a motorized treadmill
(Bont Zwolle B.V. Austria); the walking area of the
treadmill belt was 330 mm  1430 mm. The treadmill
was set at a constant speed of 3.0 km/h. The
 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620
advantages of using a treadmill for gait analysis are that
it allows for a convenient application of monitoring
equipment and provides a controlled setting by which
multiple gait cycles can be analyzed. In our measure-
ments, each subject performed one successful trial,
including at least six gait cycles. Walking on the tread-
mill can initially be an unfamiliar experience. This in
turn can influence the parameters measured. Therefore,
the measurement starts after 6 min of familiarization
time as suggested by Alton et al. [8] and Matsas et al. [9].
The ultrasonic receivers—the three ultrasonic active
markers—are mounted at predefined distances from
each other on rigid plates. The rigid plate with the
three active markers—named a triplet—is attached to
the segment by a polyester shell (Fig. 1). Our
developed special polyester shell and a safe fixation
ensure no relative motion of the marker system on the
segment and resists muscle motion. The anatomical
points are fixed to the marker system by a pointer dur-
ing the calibration phase, the relative position vectors
are constant, which means that even the skin is moving
on the hypothetical anatomical point; the calculation
does not take this into consideration. The triplets and
the fixation together reduce the skin motion. It is
important to check whether the triplet is stable. During
the measurement, the software calculates the distances
between the greater trochanter and the lateral femoral
epicondyle, the lateral malleolus and the tibial tubercle,
and the medial and the lateral femoral epicondyle. If
those distances are constant during the motion, the tri-
plets are stable; if those distances are not constant, the
triplets are repositioned during the measurement, and
the whole procedure has to be repeated.
Fig. 4. The arrangement of the measurement.
615
The triplets are attached to the sacrum, left and right
thighs and left and right calves. The position of a trip-
let on a segment is optional; however, it has to be
placed in a good visible position. The measuring head
with three transmitter sensors is positioned behind the
person, so triplets are positioned on the posterior
aspect of the body (Fig. 4). This measurement arrange-
ment avoids possible disturbances caused by handle-
bars attached to the treadmill or simply by the
movement of hands during walking, that may hide
active markers from the measuring sensors.
The position of the anatomical points of a segment
in relation to the fundamental points was specified by
an ultrasound-based pointer during the calibration
phase before measurement. Therefore, the patient has
to stand on the treadmill within the reach of the sen-
sors of the measuring head. By pointing the end of the
pointer to each anatomical point (on the surface of the
skin) and pressing the button on the pointer, its
location in space (with respect to the defined global
coordinate system) is registered by scanning the signals
of the pointer’s two ultrasound receivers. The position
of the investigated anatomical point in the segment-
embedded reference frame could be determined by
Eq. (8). The order of points is fixed according to the
applied biomechanical model and has to be considered
when entering them to the program. A usual arrange-
ment of the anatomical points is presented in Fig. 3.
2.6. Assessment parameters
The measurement system was used to process the
ultrasound-based recordings of the lower limb active
616 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620

marker movements. The measurement system was pro-

grammed to calculate coordinates for each of the 19
anatomical points (Eq. (7)) from coordinates of the 15
markers (Eqs. (3)–(6)). The raw data (the coordinates
of each investigated anatomical point) were smoothed
and filtered using a fourth-order zero lag digital Butter-
worth with high frequency cut-off at 5 Hz.
The spatial coordinates of anatomical points of the
lower limb define a number of gait parameters, which
are commonly used for the description of gait and can
be easily obtained (Table 1).
Anatomical joint angles are important because the
ranges of movement are of interest to clinicians
(e.g., hip abduction and adduction, knee flexion and
extension). Anatomical joint angles show how one
segment is oriented relative to another. There has
been some debate as to the most appropriate
method of defining joint angles [10,11]. We added a
new knee angle definition nearer to reality as a flex- Fig. 5. Definition of the knee angle (a). The knee angle is defined as
ion and extension, which take place at about the the angle between a spatial vector joining the lateral malleolus to the
mediolateral axis of the proximal segment. This defi- head of fibula and a spatial vector joining the lateral femoral epi-
condyle to the greater trochanter.
nition complements the usual frontal and transverse
plane components. To describe the position of the
thigh, the usual plane angles (as flexion and exten-
sion, and adduction–abduction) are defined. Thigh distance joining the epicondylus femoris lateralis to
rotation takes place at about the longitudinal axis of the trochanter major.
the distal segment. The distance joining the malleolus lateralis to the
The new knee angle—defined as the angle between head of fibula can be determined by
the spatial vectors joining the lateral malleolus to the qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
head of fibula and joining the lateral femoral epi- L35 ¼ ðx5  x3 Þ2 þ ðy5  y3 Þ2 þ ðz5  z3 Þ2
condyle to the greater trochanter (Fig. 5)— is
The distance joining the epicondylus femoris lateralis
a ¼ cos1

to the trochanter major can be determined by
ðx8  x6 Þðx5  x3 Þ þ ðy8  y6 Þðy5  y3 Þ þ ðz8  z6 Þðz5  z3 Þ
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
L35 L68
L68 ¼ ðx8  x6 Þ2 þ ðy8  y6 Þ2 þ ðz8  z6 Þ2
where a is the knee angle in degrees; x3, y3, z3 are
This calculation method determines a real spatial
spatial coordinates of the malleolus lateralis; x5, y5,
knee angle. The value of this angle does not depend on
z5 are spatial coordinates of the head of fibula; x6, the spatial position of the lower limb, as the compo-
y6, z6 are spatial coordinates of the epicondylus nent angles do, only on the relative position of the
femoris lateralis; x8, y8, z8 are spatial coordinates of shank to the thigh.
the trochanter major; L35 is the distance joining the For each subject, the newly defined knee angles were
malleolus lateralis to the head of fibula; L68 is the calculated at four different positions of gait cycle. The
four knee angles were at the initial contact, the mid-
stance and minimum and maximum values (Fig. 6).
Table 1 The midstance was defined as the point when the knee
Characteristic parameters describing individual gait patterns
joint had attained maximum flexion after the initial
Parameter Definition contact.
Cadence Number of steps per unit time The modified motion analysis technique provides an
Walking velocity Distance walked per unit time opportunity to define the pelvic obliquity angle parallel
Step length Distance by which each foot is with the pelvic flexion–extension and rotation angles.
in front of the other one
Our calculation method is equivalent with the sugges-
Stride length Distance by which each foot moves
forward in one gait cycle tion of Grood and Suntay [11]:
Walking base Side-to side distance between the z18  z9
two feet c ¼ tan1
y18  y9
 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620
Fig. 6. The knee angle–time graph during a gait cycle.
where c is the pelvic obliquity in degrees; y9 and z9 are
the coordinates of the right anterior superior iliac spine
(ASIS); y18 and z18 are the coordinates of the left
anterior superior iliac spine (ASIS).
The motion analysis technique is used to study the
motion of the tibia with respect to the femur into the
direction of ligaments during walking, as the medial
and lateral anatomical points of the knee are investi-
gated. The motion could be described by the relative
ligament-movement parameter, which is the relative
maximum displacement between the two characterized
points of the knee. The two characterized points of the
knee were chosen so that the line between these two
points is closely parallel with the investigated ligament.
The relative anterior cruciate ligament (ACL) move-
ment parameter is defined as the relative displacement
between the tibial tubercle and the lateral femoral epi-
condyle (Fig. 7a) specified by the minimum distance
between those two points,
DL46
eACL ¼ 
L46
where eACL is the relative ACL-movement parameter
(ACL deflection); L  46 is the minimum value of dis-
tances between the tibial tubercle and the lateral fem-
oral epicondyle anatomical points during one gait
cycle; DL46 is the displacement between the tibial
tubercle and the lateral femoral epicondyle anatomical
points. The displacement is calculated from
DL46 ¼ maxðL46 Þ  minðL46 Þ
L46 is the distance between the tibial tubercle and the
lateral femoral epicondyle anatomical points. The dis-
617
tance can be calculated from
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
L46 ¼ ðx6  x4 Þ2 þ ðy6  y4 Þ2 þ ðz6  z4 Þ2
where x4, y4, z4 are spatial coordinates of the tibial
tubercle; x6, y6, z6 are spatial coordinates of the epi-
condylus femoris lateralis.
The posterior cruciate ligament (PCL) movement-
parameter is defined as the maximum relative displace-
ment between the head of fibula and the medial
femoral epicondyle (Fig. 7b), the lateral cruciate liga-
ment (LCL) movement parameter as the maximum
relative displacement between the head of fibula and
the lateral femoral epicondyle (Fig. 7c), the medial cru-
ciate ligament (MCL) movement parameter as the
maximum relative displacement between the medial
femoral epicondyle and the medial malleolus (Fig. 7d).
3. Results
The validation of our technique is shown on a few
selected spatial–temporal parameters and knee joint
kinematics, which are either very common or newly
defined.
Spatial–temporal variables, such as the stride time,
the stride and the step length and the walking base, are
derived from the temporal and spatial coordinates. For
each subject, the average and the standard deviation of
parameters were determined from six complete gait
cycles. Fig. 8 displays a graphical representation of one
subject’s step lengths. As can be seen, the step length
differs at each gait cycle analyzed for one subject. The
other parameters show similar differences, which are
not significant (p > 0:47). Table 2 summarizes the
618 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620

Fig. 7. Anatomical points for the characteristics of the relative ligament-movement parameters. (a) ACL; (b) PCL; (c) LCL; and (d) LCM.

average values and standard deviation of these quan-
tities for healthy female and male subjects. No signifi-
cant differences were found between the spatial–
temporal variables of the left and right sides of one
subject (p > 0:37) and between these variables of all
subjects (p > 0:41). However, on the basis of the
results, we can establish that the step length and the
walking base of the dominant step are greater (5–10%),
than those of the undominant one, and the step length,
the walking base and the stride length of female sub-
jects are smaller than those of males.
The newly defined knee angle presented in this study
is shown as a function of the gait cycle during one step
(Fig. 6). Fig. 9 shows a graphical representation of one
subject’s knee angles during six cycles. As can be seen,
the knee angles differ at each gait cycle in a similar
fashion to the spatial–temporal parameters. The aver-
age values and the standard deviation of the maximum
and minimum values of the 51 adults tested are sum-
marized in Table 2. No statistical differences (p > 0:47)
were found between the left and right sides of one sub-
ject (p > 0:55) and between these values of subjects
(p > 0:39).
The relative ligament-movement parameters are
nearly the same at each gait cycle. The average and the
standard deviation of all four ligament-movement
parameters are summarized in Table 3. On the basis of
the results, we can establish that the relative ligament-
movement parameters of male subjects are closely

Fig. 8. Step length for one subject at six gait cycles during treadmill walking.
 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620 619

Table 2
The average values and the standard deviation of temporal–distance parameters and knee joint kinematics determined during treadmill walking at
male and female subjects

Step length [mm] Walking Minimum value of the knee Maximum value of the knee
v v
base [mm] angle [ ] angle [ ]
Right Left Right Left Right Left
Male Average 513.3 510.3 41.9 5.5 5.4 52.3 51.2
Standard 26.6 28.8 8.2 0.98 1.05 1.32 1.74
deviation
Female Average 470.7 466.3 39.0 7.3 7.7 57.3 57.6
Standard 20.1 29.9 9.9 1.29 1.88 1.96 1.85
deviation
Range in literature [12] 450–980 40–77 0–14 44–87

equivalent to the values of females. No difference was
found between the values of the right and left sides.
4. Discussion
The motion analysis technique presented in this
study is capable of reliably measuring kinematic para-
meters in gait. The fundamental spatial–temporal para-
meters can be determined from the spatial coordinates
of investigated anatomical points. The spatial–temporal
parameters presented in this study compare favorably
with the values found in literature found [12]. The
respective relative ligament-movement parameter—that
is, the maximum relative displacement of anatomical
points characterizing the ligament—describes the con-
dition of ligaments. The relative ligament-movement
parameters were used to quantify the movement of the
femur with respect to the tibia in the direction of the
ligaments. Let us see the characteristics of the relative
ACL-movement parameter, which describes the con-
dition of the ACL. At heel strike with the knee at full
extension, the tibia is at its maximum anterior position
during the gait cycle. The next key event occurs at ter-
minal extension where the tibia is located posterior
again while the knee is near full extension. During the
swing phase with the limb unloaded, the tibia moves to
its maximum posterior position at maximum knee flex-
ion, and then moves forward rapidly and the knee
extends prior to heel strike. The overall range, the
maximum displacement between the tibial tubercle and
the lateral femoral epicondyle was used to quantify the
characteristics of dynamic stability and the differences
observed between normal subjects and patients with a
ACL deficient knee. The measured data represent that
the relative ligament-movement parameters do not
depend on gender and dominant side. The values of
these parameters depend only on the movement of the
femur with respect to the tibia and on the translation-
motion of the femur’s condylus, which only depends on
anatomical state.
In conclusion, the proposed method appears to be a
promising monitoring tool for several purposes. First,
it allows for measurements of gait features during a
long period of treadmill walking, and thus supplies the
stride-to-stride variability of gait. In addition, the pro-
posed method investigates not just the lateral but also
the medial anatomical points and therefore provides
more information of gait than other methods. When

Fig. 9. Knee angles for one subject at six gait cycles during treadmill walking.
620 R.M. Kiss et al. / Medical Engineering & Physics 26 (2004) 611–620

Table 3
The average values and the standard deviation of the relative ligament-movement parameters at male and female subjects

Relative ACL- Relative PCL- Relative LCL- Relative MCL-movement

movement parameter movement parameter movement parameter parameter
Right Left Right Left Right Left Right Left
Male Average 0.25 0.26 0.34 0.34 0.32 0.32 0.0062 0.0062
Standard deviation 0.020 0.018 0.017 0.016 0.032 0.032 0.00050 0.00044
Female Average 0.25 0.26 0.33 0.33 0.35 0.36 0.0067 0.0066
Standard deviation 0.016 0.017 0.014 0.017 0.030 0.035 0.00046 0.00042

compared to other methods, our technique presents and by the Hungarian Academy of Sciences Research
some advantages: Group of Structures.

1. With our measuring arrangement, the motion of the

upper limbs does not hide the markers, and the posi-
tioning of the fundamental points has no effect on
the results.
2. The number of anatomical points can be modified
according to the needs of the investigation and the
19-point model is able to determine the newly
defined ACL movement parameters.
3. The movement analysis can be achieved not only
during few gait cycles but also during several gait
cycles, providing additional information.
4. The method uses a simple 3D body model easy to
adapt to each subject. Our method, however, pre-
sents a limit, which is due the limited range. Our
device has a maximum range of about 3 m. The
ultrasonic device used (zebris CMS-HS) could be
extended by the use of electromyography and by a
force plate for the analysis of the kinetics of the
human body.
The presented motion analysis technique can be used
in many clinical applications such as outcome evalu-
ation after knee and hip replacement or external pros-
thesis adjustment for amputees. This technique can
also be proposed as a monitoring tool to assess pro-
gress through rehabilitation.
Acknowledgements
This work was supported in part by Hungarian
Scientific Fund T034150 and T037272, by Stipendium
István Széchenyi (Rita M. Kiss and László Kocsis),
References
[1] Gage JR. Gait analysis in cerebral palsy. London (England):
MacKeith Press; 1991.
[2] Cappozzo A, Catani F, Leardini A, Benedetti MG, Della Croce
U. Position and orientation in space of bones during movement:
experimental artifacts. Clinical Biomechanics 1996;11:90–100.
[3] Andrews JG. Euler’s and Lagrange’s equations for linked rigid-
body models of three-dimensional human motion. In: Allard P,
Stokes IAF, Blanchi JP, editors. Three-dimensional analysis of
human movement. Chanoaign: Human Kinetics; 1995, p.
145–175.
[4] Winter DA. Biomechanics and motor control of human
movement. New York: Wiley; 1979, p. 175–87.
[5] CMS-HS ultrasound measuring system. Technical parameters.
Available from: http://www.zebris.de.
[6] Vaughan CL, Davis BL, O’Connor JC. Dynamics of human
gait. Cape Town (South Africa): Kiboho Publisher; 1999.
[7] Bulgheroni P, Bulgheroni MV, Andrini L, Guffanti P,
Giughello A. Gait patterns after anterior cruciate ligament
reconstruction. Knee Surgery, Sports Traumatology, Arthro-
scopy 1997;5:14–21.
[8] Alton F, Baldey L, Caplan S, Morrissey MC. A kinematic com-
parison of over ground and treadmill walking. Clinical Bio-
mechanics 1998;13:434–40.
[9] Matsas A, Taylor N, McBurney H. Knee joint kinematics from
familiarized treadmill walking can be generalized to over ground
walking in young unimpaired subjects. Gait and Posture
2000;11:46–53.
[10] Chao EY. Justification of triaxial goniometry for the measure-
ment of joint rotation. Journal of Biomechanics 1980;13:989–
1006.
[11] Grood ES, Suntay WJ. A joint coordinate system for the clinical
description of three-dimensional motions. Application to the
knee. Journal of Biomechanical Engineering 1983;105:136–44.
[12] Whittle M. Gait analysis—an introduction. Oxford: Butter-
worth-Heinemann; 1991, p. 218–21.