You are on page 1of 31

Human Movement Science 3 (1984) 27-50

North-Holland

27

GAIT ANALYSIS METHODOLOGY *

Aurelio CAPPOZZO
Urtioersitir de@ Studi ‘La Sapienra’, Italy

Cappozzo, A., 1984. Gait analysis methodology. Human Movement


Science 3, 27-50.

This paper reports a discussion on the heuristic and applicative objectives of gait analysis and on
the experimental and analytical methods used in this context. In particular. the use of an effective
method of description of joint kinematics and kinetics i.e., joint function, as applied to the lower
limb joints during normal walking is reported. A methodological hypothesis for the evaluation of
gait, as an integrated phenomenon is presented and supported by experimental data concerning
normal and pathological walking and sportive gaits.

Introduction

Locomotion is the action with which the entire bulk of the animal’s
body moves through aerial, aquatic, or terrestrial space. Locomotion is
achieved by coordinated movements of the body segments employing
an interplay of internal and external forces (Cappozzo et al. 1976).

Although locomotion is a very complex phenomenon which can only


be thoroughly described through a multidisciplinary approach, the most
qualifying approach comes from the domain of classical mechanics,
that is to say that biomechanics has the greatest responsibility for
establishing the relevant scientific knowledge.

The quantitative description of all mechanical aspects of walking is


commonly referred to as gait analysis. This paper intends to elaborate
on this definition and on the scientific and applicative objectives of gait

* Some of the experiments referred to in this paper were carried out at Physical Support Italia
Ltd. Roma, where a VICON system was made available to the author. The collaboration of Dr.
Fabio Gazzani in the preparation of this manuscript is gratefully acknowledged.

Author’s address: A. Cappozzo, Laboratory of Biomechanics, Istituto di Fisiologia Umana,


Universita degli Studi ‘La Sapienza’, 00100 Roma, Italy.

0167-9457/84/$3.00 0 1984, Elsevier Science Publishers B.V. (North-Holland)

analysis. In addition to this, some crucial experimental and analytical


problems encountered in analyzing gait will be discussed.

What is gait analysis?

An effective way of outlining the contents of a discipline is to describe


its historical development and its heuristic and applicative objectives.

Borelli (1680) was the first to apply Galileo’s scientific method to the
study of locomotion. His treatise on walking has more than a mere
historical value. His analysis may be regarded as correct for walking at
a very slow cadence. However, lack of proper instrumentation and
incomplete awareness of the inertia principle probably mediated against
him abandoning what, today would be looked upon as a quasi-static
approach.

The leap from a static to a dynamic approach was facilitated by the


development of photography. Only then, could fast motions be stopped
and analyzed. Muybridge (1887) was the first to publish photographic
sequences of animal and human locomotor acts. These sequences were
obtained using a systematic method which permitted some detailed
observation of the phenomenon. However, measurements in the strict
sense could not be sufficiently accurate. Muybridge’s work awakened
interest and produced results of great relevance to the field of arts and
it also stimulated the use of photography in scientific research.

At about the same time, in fact, and after having become acquainted
with Muybridge’s work, Marey (1885) started to use photography in his
life movement studies. Although Marey’s experimental approach to the
problem was highly scientific and represented a fundamental enrich-
ment of man’s sensorial capacity, the observation of the phenomenon
however still remained to a large extent qualitative.

The history of modern biomechanics starts in 1895 when Braune and


Fisher begin publishing their work. Two fundamental developments
pioneered by them merit their receiving the title of fathers of modern
biomechanics. Firstly, they perfected the process of idealization of the
natural phenomenon being studied ~ basic to any scientific methodol-
ogy - which allowed the use to full capacity of the instruments of
classical mechanics. This permitted them to predict those aspects of the
phenomenon which were not observable or were difficult to observe
directly. Secondly, and also by virtue of the above-mentioned idealiza-

A. Cuppozro / Gait analysis methodolo~ 29

tion of the phenomenon, they could design experimental setups which


permitted proper measurements. In particular, Braune and Fisher were
the first to represent human body segments using a major stereotype of
classical mechanics: the rigid body. They also introduced into the study
of life movement the technique of stereometry, of stereophotogramme-
try in particular, which today is still the basic experimental technique in
biomechanics of human movement.

Stereometry permits the three-dimensional (3-D) reconstruction of


the instantaneous position of a moving point in a laboratory coordinate
system. If this is carried out for at least three noncollinear points fixed
in a body segment (supposed rigid) then the position of this segment is
thoroughly defined. Several techniques use the basic concepts of
stereometry: (1) stereophotogrammetry (Braune and Fisher 1895; Bern-
stein 1967; Eberhart et al. 1947; Winter et al. 1972; Jarrett et al. 1974;
Lindholm 1974; Macellari 1983) (2) light scanning (Mitchelson 1975),
(3) stereosonic systems (Brumbaugh et al. 1982).

Other techniques used for the measurement of body segment motion


are exoskeletal linkage and accelerographic methods (Chao 1980; Hayes
et al. 1983). The former method permits the measurement of the gross
relative motion between adjacent body segments. The use of accelero-
metry poses several technical and computational problems, and does
not therefore seem to be competitive with the methods discussed
previously. However, in those situations where body segments are
subjected to external impulsive forces, as when the heel strikes the
ground during gait, this technique plays a fundamental role.

The problem of measuring the 3-D movement of human body


segments was thus already solved at the end of the last century.
Subsequent developments have led to such measurements becoming,
largely, automated.

A further problem posed by gait analysis concerns the forces that


cause the observed movement and the assessment of their role within
the phenomenon being analyzed. The forces acting on the human body
during walking may be devided into two categories: internal and
external. The external forces represent all physical interactions between
the body and the environment. These are the gravitational and the
ground reaction forces. The internal forces are, among others, those
transmitted by body tissues as muscular forces, tension forces trans-
mitted by the ligaments and forces transmitted through joint contact
areas.

30 A. Capporro / Gait analym methoddog,

Generally, the only forces which can be measured are the ground
reactions. This simply requires the use of a dynamometric platform
which occupies a portion of the floor on which the locomotor act is
performed. The first dynamometer of this type, although rather primi-
tive, was built by Amar (1916). Only in 1947 was a dynamometric
platform able to measure the three forces and three moments necessary
to describe the interaction between foot and floor made available
(Eberhart et al. 1947).

The gravitational forces acting on each body segment can be


determined from the relevant mass and location of the centre of mass.
These quantities can be calculated, together with the segmental mass
moments of inertia, by means of prediction techniques from anthropo-
morphic dimensions. Current methods for their in uivo experimental
assessment are complex and some of them may be harmful to the
subject’s health. The prediction techniques can be divided into two
major methodological groups: (1) estimation through regression equa-
tions (Dempster 1955; Clauser et al. 1969; Chandler et al. 1975;
Zatsiorsky and Seluyanov 1983); (2) geometrical approximation
(Hanavan 1964; Jensen 1978; Hatze 1980b). The regression equations
have been determined through statistical analysis of data obtained from
sample populations of subjects. Most of the measurements have been
made using cadaver specimens. The geometrical approximation tech-
nique entails that the irregular shapes of the different body segments
are represented with standard geometric forms which admit simple
mathematicdl description.

The measurement of internal forces requires sophisticated techniques


which are often invasive, and which are difficult to use when dynamic
physical exercises, such as walking, are analyzed. For these reasons,
analytical procedures have been developed for estimating internal forces.
This approach requires data on the motions and the inertial properties
(mass, mass moments of inertia, and locations of the centres of gravity)
of the subject’s body segments, and on the ground reactions. The
analysis then leads to the calculation of a set of internal forces which
are consistent with the observed movement. Relevant analytical meth-
ods use classical mechanics and modelling approaches.

Using the equations of rigid body mechanics it is possible to de-


termine the intersegmental forces and couples, that is, the resultant
force actions transmitted across an imaginary surface which separates
two adjacent body segments (Bresler and Frankel 1950; Cappozzo

1983). These forces and couples represent all physical connections of


one body part with the other. They are assumed to maintain both body
parts in the same dynamical state they had before their imaginary
separation. From this knowledge and from soft tissue anatomical and
functional information the prediction of internal loads occurring during
physical exercise may be undertaken. This is achieved through modell-
ing techniques (Paul 1966; Seireg and Arvikar 1975: Hatze 1980a;
Cappozzo 1984).

In several instances an effective way to describe the process of


walking is achieved by analyzing the function of energy; variables such
as force and displacement can then be reduced and used in a more
compact form of work or energy variation (Cappozzo et al. 1976).
When a mechanical model of the human body is defined as a linkage of
physically-united rigid segments transmitting forces and couples, the
mechanical energy of these segments and the work done on them can be
calculated.

For mathematical tractability, the above-mentioned analytical ap-


proaches require the use of simplifying assumptions about the mechani-
cal structure and behaviour of the human body. Thus, the accuracy of
the analytical predictions depends not only on the quality of the input
data, but also on the validity of these simplifying assumptions. In
general it is necessary to validate the estimated quantities by comparing
them, or other related quantities, with empirical observations in a
number of selected cases. For this purpose physiological measurement
techniques such as electromyography, intra-abdominal and intra-discal
pressure measurements etc., are often used.

The ensemble of the analytical and experimental procedures which


lead to the determination of the above-mentioned quantities is com-
monly referred to as gait analysis.

Walking is accomplished through the summation of a certain number


of musculo-skeletal functions. An objective of gait analysis is the
quantitative assessment and description of these functions, that is, of
the joint motions which concur to form the locomotor act and the
forces that control these motions. This analysis permits recognition of
the “how” a given individual walks.

The way the above-mentioned functions are combined together with


the objectives that are pursued through their combination pertain to the
motion strategy. A strategy is chosen among those that are consistent
with the functional and structural constraints associated with the

32 A. Capporro / Gurt analysis merhodologr

locomotor system being analyzed. It is reasonable to assume that with


these different possible strategies different qualities of gait are associ-
ated. Based on these considerations, it may be concluded that the
quality of the gait performed by a subject depends on two major
factors. One is associated with the functional and structural constraints
imposed by the subject’s locomotor system and, given this locomotor
system, the second factor is associated with the subject’s ability to put
into action an effective motion strategy. Assigning a value to the quality
of gait is another objective of gait analysis, usually called gait evalua-
tion. The discrimination between the two above-mentioned factors
influencing the quality of gait is yet another, and the most ambitious,
objective of gait analysis.

Function assessment

Musculo-skeletal function assessment entails the quantitative descrip-


tion of joint kinematics and kinetics. Basically, this requires the de-
termination of the relative motion between the opposing bones and of
the relevant inersegmental load actions. When diarthrodial joints are
considered, the intersegmental couple may be calculated so as to be
assumed to equal the muscular moment acting about the joint therefore
yielding information about muscular function.

As was mentioned in a previous section, the experimental and


analytical determination of the above-mentioned mechanical quantities
represents no difficulty from a theoretical point of view. A free-body
diagram of the portion of the human body under investigation is
defined which is a linkage of physically-united rigid segments. With this
mechanical model, positional and inertial properties are associated.

The positional properties convey the information needed to locate, at


any instant in time, the position of any point in a segment relative to an
arbitrarily chosen reference observer. This is achieved by defining a
local coordinate system (o x y z) rigid with the segment and a global
coordinate system (0 X Y Z) rigid with the observer. Given the
position vector p of any point in the local coordinate system (I.c.s.), the
relevant position vector p in the global coordinate system (g.c.s.) is
given by:

A. Gait ana&sis nwthodolo~ 33

where A is the transformation matrix and PO is the position vector of


the 1.c.s. origin relative to the g.c.s. The columns of matrix A are the
direction cosines of the I.c.s. axes with respect to the g.c.s. Hence, the
definition of the instantaneous position of the moving segment relative
to the fixed observer reduces to determining the position vector p,, and
transformation matrix A, which therefore become the mechanical model
positional properties.

Stereometric techniques permit the determination of the instanta-


neous position of points fixed in a moving segment and, therefore,
allow simple vector calculations of the segment transformation matrix
and position vector. All stereometric techniques entail that the target
points be represented by convenient skin markers the physical realiza-
tion of which depends on the particular technique used. Consequently
the target points are selected according to the following practical
considerations:

(1) relative movement between markers and underlying bone due to


soft tissue deformation should be minimal in order to be consistent
with the assumption of rigidity of the body segment;
(2) the distance between markers should be sufficiently large so that
error propagation from measured marker coordinates to the transfor-
mation matrix is minimal;
(3) sufficient measurements should be available on the markers from at
least two stereometric detectors at anyone time (Woltring 1982).

The target points which meet the above requirements may not coincide
with anatomical landmarks. Consequently, the 1.c.s. determined using
these points may result in a thoroughly arbitrary geometric relationship
with respect to the relevant body segment. This poses some practical
problems which will be considered below.

The inertial properties associated with each model segment are


represented by the following parameters of the relevant body segment:

(1) mass,
(2) position vector of the centre of mass in a given l.c.s.,
(3) principal axes of inertia defined relative to a given 1.c.s. through a
transformation matrix and moments of inertia.

The position vector of the centre of mass and the principal axes of
inertia transformation matrix are given, in the previously cited litera-
ture, with respect to sets of axes passing through bony anatomical
landmarks. These “anatomical” axes are, in general, different from the
local axes derived from the markers mentioned above which in the
following will be referred to as technical markers. It follows. therefore,
that in order to associate the above-mentioned inertia properties with a
mode1 segment, information must be acquired that would permit the
determination of the former from the latter axes. This entails the
determination of the position vectors of selected anatomical landmarks
relative to the technical marker 1.c.s.

The above positional and inertial properties thoroughly define the


mechanical model. Thereafter, analytical mechanics permit the calcula-
tion of the relative motion between adjacent segments (i.e., joint
kinematics), and the intersegmental force and couple vectors. However,
it should be emphasized that these objectives require the solution of
some important practical problems.

One problem relates to the description of joint kinematics. The


crucial difficulty here is the overcoming of the ever-existing dichotomy
between the language of the physicist and that of the biologist and
physician. The former language should be made to converge towards
the latter, however, without losing its properties of quantitative exacti-
tude. This is not only to favour the use of this information in the
relevant applicative fields but also because the language of biology is
more likely to be the most effective in describing biological functions.

In general, the motion which occurs at a joint is described by six


independent parameters: three rotations and three translations. Trans-
lations, however, in most circumstances, are of small magnitude and
not accurately detectable using the experimental techniques typical of a
gait analysis laboratory. Therefore they will not be considered here. As
far as rotations are concerned, they require the definition of a reference
system of axes. The so-called gyroscopic system seems to match suc-
cessfully the requirements of both mechanics and functional anatomy.
This system is defined with one axis fixed to one segment, another axis
fixed to the second segment, and the third axis (the floating axis) is the
common perpendicular to the first two. This system of axes is not
necessarily orthogonal, but has the characteristic that joint angular
position is described by three angles (Bryant or Cardan angles) inde-
pendent of the temporal order in which the relevant rotations are

A. Capporzo / Gait analysts methodolow

Q
cu

thought to occur (Chao 1980; Grood and Suntay 1983).


An example of the application of the above procedure to the lower

limb is used in the author’s laboratory. An optoelectronic stereophoto-


grammetric system is used for the segment motion measurement. The
data presented here were obtained with a VICON system (Oxford
Dynamics Ltd) equipped with three TV cameras positioned as shown in
fig. la. The technical markers are placed on the subject’s pelvis, thigh,
shank and foot as shown in fig. lb. These positions satisfactorily meet
the practical requirements listed previously. The marker P3 is affixed to
a stick in line with the posterior superior iliac spines and is held in
place by an elastic belt. The thigh markers and the shank marker S, are
mounted on 5 cm sticks. In this way their relative distance is enlarged
and, as was said previously, measurement error propagation reduced. In
addition to this, marker visibility from the cameras is enhanced. The
markers S,, P,, P2, F,, and F2 are placed on the following anatomical
landmarks: head of the fibula, anterior superior iliac spines, fifth and
first metatarsal bones. In addition to the above-mentioned technical
markers, other markers - referred to as anatomical markers - are
placed on the following anatomical landmarks: lateral and medial
epicondyles of the femur, and lateral and medial malleoli.

The experimental protocol comprises three tests. During the first test,
the subject assumes a relaxed upright posture facing camera two so that
all markers are visible by at least two cameras. This test permits the
determination of the local position vectors of the C,, C,, M,, and M,
anatomical landmarks with respect to the axes defined by the technical
markers. The test data are also used for the determination of the joint
angular position during upright standing. After the above static test is
performed the anatomical landmark markers are removed. The second
test aims at the determination of the hip centre of rotation. The subject
performs a movement of abduction-adduction of the thigh followed by
a flexion-extension. Using the reconstructed 3-D trajectories of the
thigh markers it is possible, through a least-squares method to estimate
the location of the centre of rotation of the above movements, that is,
of the hip centre both in the pelvis and in the femur 1.c.s.‘~. At this
stage, information is available for the determination of the following
anatomical systems of reference (fig. lb). For the pelvis, the technical
marker system coincides with the anatomical system. For the thigh, the
x,-z, plane contains the anatomical landmarks C,, C,, and H, and the z,
axis joins the midpoint (K) between C, and C, and the hip centre. For

A. Cuppozro / Gait analysis methodology 37

the shank, the x,-z, plane is defined by the anatomical landmarks S,,
M,, and A4,. The z, axis is defined by the orthogonal intersection
between the x,-z, plane and the bundle of planes defined by S, and the
midpoint between the malleoli. The x, axis is made to pass through the
M, landmark so as to approximate the ankle flexion-extension axis
more closely. The foot anatomical system of reference was defined
using the markers F, and F2 and the point A shown in fig. lb. The
anatomical axes defined above approximately have the orientation of
axes of symmetry. As such they may be considered principal axes of
inertia, and the segmental centre of mass may be considered as lying on
the z axes. These simplications are justified by the magnitude of the
inaccuracy with which the above inertial parameters are predicted. The
joint coordinate systems and the relevant rotations are defined as in
table 1. The third test is the actual dynamic test. In fig. 2A an example
of lower limb joint kinematics is depicted for a normal adult male
subject walking at a free cadence.

As far as joint kinetics is concerned, the following general remarks


are worth of note. When calculating the intersegmental force and
couple vectors a point of application must be chosen. Selecting an
approximate point is important since both the numerical value and the
functional significance of the couple vector depend upon its location.
This point should be defined either as the point through which the
resultant articular surface contact force acts or, in the absence of
friction, could be taken as the joint centre of rotation. Thus, the
intersegmental couple is balanced by the moments of forces transmitted
only by the soft tissues around the joint, and in particular by muscles
and ligaments. The intersegmental couple vector will then be presented
in its scalar components which have functional significance. It is
thought that the components along the joint coordinate systems defined
above meet this requirement. The intersegmental couple calculated with

Table 1
Joint rotation axes (for symbols see fig. lb; f.a. = floating axis). For the foot internal and external
rotations are more commonly defined as eversion and inversion, respectively.

Hip
Knee
Ankle

Flex-ext.

xP
x,
3

Ad-abduc.

f.a.
f.a.
f.a.

RTO RHS RTO

0 50 100
TIME C% of cycle1

(Fl)

t
RTO RHS RTO

0 50 100
TIME C% of cycle1

(El

Fig. 2. (A) Lower limb joint kinematics during walking at 1.3 msec- ‘. Right limb of an adult male
subject (body mass = 63 kg, stature =1.73 m). Hip ( p), knee (-.P.-), ankle (- - - - -).
Abscisses indicate angles during up-right standing.
(B) Lower limb joint intersegmental couple components along the joint coordinate systems. Same
test and line types as above. Absiisses indicate zero moments.
(RHS. LHS = right and left heel strike; RTO, LTO = right and left toe off).

respect to the points H, K, and A shown in fig. lb, and resolved with
respect to the relevant joint coordinate systems, are given in fig. 2B as
obtained during the same walking test as the data in fig. 2A. The points
H, K and A approximate the location of the instantaneous rotation
pivots of the relevant joints. Similar kinetic data relative to the upper
trunk may be found in Cappozzo (1983).

Gait evaluation

Gait evaluation should be approached in terms of both assessment of


the resources that the subject employs to accomplish the locomotor act
and the quality of the result. In a previous section this was summarized
by the statement: “gait evaluation is the assessment of the ability of the
subject to move through space by ambulation”. It may be assumed that
this ability is directly related to the reliability of the locomotor act and
to its consistency with the relevant aesthetic canons (the word “reliabil-
ity” is used here in the engineering sense, that is, as the probability that
the function be performed satisfactorily under given circumstances).
This type of approach to gait evaluation carries with it an important
conceptual implication. Reference to normality, which is always some-
what implicit in gait evaluation, does not entail the comparison of the
normal and abnormal mechanical patterns per se, nor does it entail the
comparison between single functions as would be the case if the
quantities described in the previous section were used. The emphasis is
placed on the overall strategy of the movement making allowance for
the often objective impossibility that the patient achieve normal motion
trajectories since his neuro-muscular-skeletal system is, in some cases,
irreparably impaired. Nevertheless, the patient can achieve a high level
of ability in walking relative to his locomotor system.

Subsequent to the above considerations, a gait evaluation methodol-


ogy may be suggested that uses physical quantities capable of describ-
ing the following aspects of walking:

(1) symmetry and simplicity of the movement,


(2) maintenance of balance,
(3) mechanical load on tissues, and
(4) energy expenditure,

the first aspect being related to the aesthetics of walking, the others to
the reliability of the locomotor act.

Gait aesthetics is mostly associated with the upper part of the body
motion. The symmetry and simplicity of this motion is well depicted
through the representation of the relevant displacement components in
the frequency domain. This matter is discussed in full length in Cap-
pozzo et al. (1982), and Cappozzo (1981). As emphasized in these
papers, a “stereotype” pattern of movement of the upper body can be

40
RNTERO-POSTERIOR RXIS
I HRRM.

HEAD

II HRRM. III HRRM. IV HARM.

05 --I--
LRTERO-LATERFIL RXIS

I HARM. II HRRM. III HRRM. IV HRRM.

PELVIS

*---k-r--++

VERTICRL AXIS
I HRRM. II HARM. III HRRM. IV HARM.

PELVIS

*---i-+++

Fig. 3. Vector representation of the harmonic components of the linear displacements of the
longitudinal axis of the upper part of the body at head and pelvis level. The sectors are defined by
the phase and amplitude ranges of the harmonic component vectors relative to sixteen tests of five
normal subjects in the speed range 1.31-1.60 msec -‘. The harmonic component vectors relative to
three tests of an above-knee amputee walking in the same speed range are superimposed. Numbers
on the abscisses indicate mm/div.

A. Cappozro / Gart analysis methodologr 41

devised. This is characterized by a perfect symmetry of each elementary


displacement with respect to the anatomical planes. The stereotype
pattern of movement of the longitudinal axis of the trunk and head, for
instance, moves along the vertical (2) and antero-posterior (X) axes
with a cycle period equal to one step, and along the latero-lateral ( Y)
axis, with a cycle period equal to one stride and with mirror-image
symmetry with respect to the mid-sagittal plane. Thus, in the frequency
domain, the displacement components along the X and Z axes are
described by harmonics of even order only, while along the Y axis by
odd order harmonics only. These were termed intrinsic harmonics. The
other harmonics were termed extrinsic and were related to gait asymme-
tries. In a normal population the intrinsic harmonic coefficients were
found highly repeatable, while the extrinsic harmonic coefficients were
mostly of a random nature. In fig. 3 the significant intrinsic and
extrinsic harmonics are depicted, respectively. They are represented in
vectorial form. The sectors are defined by the phase range and ampli-
tude range of the harmonic vectors relative to a population of five
normal adult male subjects walking at a speed in the range 1.3-1.6
msec-‘. Superposed on these normative data, the harmonic vectors are
shown for three walking trials in the same speed-range of an adult male
above-knee amputee wearing a single-axis prosthesis. The magnitude of
the relevant extrinsic harmonics may be used as a measure of the
movement asymmetry. In this amputee’s gait, as opposed to normal
walking, the extrinsic harmonics are repeatable; this suggests that they
are associated with a specific locomotor system impairment and not
with random causes. The intrinsic harmonic vectors are often close to
the relevant normal range (fig. 3). Whether this circumstance would
apply to other types of patients or not, and the meaning which can be
associated with it, are matters that need further investigation. The
Lissajous’s plots presented in Cappozzo (1981) may also be an effective
way of representing the symmetry and simplicity of the movement. In
fig. 4A, B, and C these plots are given for a normal subject, the
above-knee amputee referred to above, and for an adolescent suffering
from cerebral palsy. These plots have a more qualitative character than
those in fig. 3, but they are more immediately readable.

The other aspect of walking which is of interest in the present


context is the effectiveness with which balance is maintained. The
Lissajous’s plots of fig. 4 may be able to supply some information
concerning this aspect as well. From those plots it would appear, for

(A
)

(B
)

P
E

L
V

IS

H
E

R
D

E
L

V
IS
YI
le

ft

Yl
le

ft

1
di
v=
l@
mm

1
di
v=
lQ
mm

(C
l

H
E

A
D

9P

E
L

V
IS

2T

U
P

id
ow
n

fd
ow

n
Yi

le
ft

YI

le
ft

ba
ck

fr
ig

ht

fr
ig

ht

fd
ow

n
fd
ow

1
di
v=
l@
mm

F
ig
.
4.

is
sa

Jo
u

s’
s

pl
ot

s
re

la
ti

ve

to
t

h
e

tr
aJ

ec
to

ri
es

re

fe
rr

ed

to
i
n
f

ig
.

3
as

r
ef

er
en

ce
d

to

a
s

ys
te

m
o

f
re

fe
re

n
ce

en

d
o

w
ed

it
h

tr
an

sl
at

io
n

al

m
o

ti
o

th
e

ve
lo

ci
ty

f
w

h
ic

eq
u

al
s

th
e
m
ea

ve
lo

ci
ty

f
th

e
re

le
va

n
t

p
o

in
ts

.
(A

)
N

or
m

al

su
bj

ec
t

w
al
ki
ng

at

1.

33

m
se

c
-‘

.
(B

)
am

pu
te

e
su

bj
ec

t
w

al
ki

ng

at

1.
38

se
c-

‘,
(C

)
ce

re
br

al

pa
ls

y
pa

tie
nt

al
ki

ng

at

1.
24

se
c-

‘.
(E

m
pt

y
ci

rc
le

s
an
d
sq

ua
re

s
in

di
ca

te

th
e

oc
cu

rr
en

ce

of

ri
gh

t
he

el

st
ri

ke

an
d

ri
gh

t
to

o
of
f.
re

sp
ec

tiv
el

y:

fu
ll

ay
m

bo
lh

re

la
te

to

t
h

e
le

ft

h
m

h
.)

--

-^
-^

-_
-_
__

-
cI

_
j_

A. Cappoiro / Gait anu~wrs methodologp 43

instance, that the amputee is able to control balance more effectively


than the cerebral palsy patient. His head, in fact, has a more regular
trajectory than his pelvis and this implies a compensatory movement of
the upper trunk. The opposite occurs in the cerebral palsy patient, who
thus appears unable to optimally control the upright posture of the
trunk. Another way of looking into the effectiveness of postural control

x RX15
HERD 0

PELVIS A

z2
0
H
c

A
A

A
AA A A

I4

.5 1.0 1.5 2.0 2.5


SPEED Cm s-l1

(A)
HERD 0

x RXIS
PELVIS A

r
6

v) 5
E

L.l

014
E
L

z3
0
H

A
A
AA

AA

0 “.r;’ o.
0

,
0 2.5 3.0 3.5 4.0

SPEED Cm se11
(B)

44 A. Copporzo / Gart anabsis methodolog

2 AXIS

Y
6

-F
cd 5
E

HERD 0

PELVIS A

AA
00

A
A
A

w 4- 0
E A Am A 0
L0

z 3-
.@

Al9
0AB

r
A

a2- l

w”

d l-
.c 4

I
0.5 1.0 1.5 2.0 2.5

SPEED Cm s -‘I
(Cl

HERD 0

PELVIS A

0
A

Q
4A

-. 0 2.5 3.0 3.5 4.0


SPEED Cm s -‘I

Fig. 5. Effective value of the linear acceleration of the upper body longitudinal axis at pelvis and
head level versus speed of progression. (A) Acceleration along the antero-posterior axis during
normal walking (empty symbols) and amputee walking (full symbols).
(B) Acceleration along the antero-posterior axis during running (empty symbols) and during
race-walking (full symbols).
(C) As in (A) but along the vertical axis.
(D) As in (B) but along the vertical axis.

A. Cuppoiro / Gail ana~vsis methodoloR) 45

is through the analysis of upper body accelerations both in the time and
frequency domains. The antero-posterior acceleration of the trunk and
head is bound to carry most of the relevant information. It is, in fact,
along this direction that, during locomotion, balance is “lost and
regained”; thus, control of this component is more critical. The follow-
ing experimental results help in appreciating the relevance of this
acceleration component. In fig. 5A, the effective value of antero-poste-
rior acceleration at pelvic and head levels is depicted, versus speed of
progression. This data refers to normal walking and amputee walking.
In fig. 5B the same data obtained during race-walking and running is
shown. From these plots it appears evident that the antero-posterior
component of head acceleration is well under control: whatever
locomotor act is performed normal or abnormal, the effective value of
this acceleration component has always approximately the same rela-
tively small value. This is intriguing and suggests that this acceleration
component must be viewed as a figure of merit for the effectiveness of
the control of balance. This opinion is reinforced by the analysis of this
acceleration component in the frequency domain. In Cappozzo (1982) it
was seen that in normal subjects the upper trunk movement with
respect to the pelvis causes a sort of filtering of the high-frequency
acceleration components present at the pelvic level. This was found to
occur in amputee-walking, in race-walking, and also in running. It is
thought that these phenomena are associated with the optimization of
the effectiveness of the labyrinth role in controlling balance, and
possibly with the requirements of vision as well. Of course, the hypothe-
sis that antero-posterior acceleration at the various levels of the upper
body may supply information on the effectiveness of postural control
remains to be validated. The analysis of neurological patients may be
helpful in this respect.

During walking, the mechanical loading of tissues should be main-


tained below a suitable threshold. This loading is essentially applied
along the crania-caudal direction. Horizontal forces are in fact ab-
sorbed by muscles through the yielding of some articulation. An overall
assessment of the mechanical loading of upper body tissues, which are
the most delicate, may be obtained through the analysis of the relevant
vertical acceleration. This data is reported in figs. 5C and D with
reference to the same test-trials as figs. 5A and B. As could be easily
foreseen, vertical acceleration increases with speed during normal
locomotion. The amputee, at his maximal speed of progression -

46 A. Capporro / Gaif malysrs methodolog:r

approximately 1.5 msec-’ - exhibits a vertical acceleration similar to


that of normal subjects at their maximal speeds of progression (fig. 5C).
After this observation, it may be asked whether the vertical acceleration
~ and the mechanical load associated with it ~ be one of the limits to
walking speed, both in normal and handicapped persons. However,
higher speeds of progression may be obtained by changing gait, as in
race-walking, for instance (fig. 5D). In this case, although speed of
progression increases with respect to normal walking, vertical accelera-
tion does not. Is it possible to infer, at this stage, that there exists a
limit over which the vertical acceleration’s effective value should not
go? In running, even higher speeds of progression are however accom-
plished. As shown in fig. 5D, the vertical acceleration during running is
remarkably greater than during walking and race-walking. But, if this
acceleration component is analyzed in the frequency domain, then the
following is found. The vertical acceleration amplitude spectrum for
walking is found to occupy the range l-5 Hz with a dip at approxi-
mately 3 Hz (Cappozzo 1982). These characteristics fit very well the
biodynamical properties of the body in that no resonances should occur
in relevant organs. During running, this acceleration spectrum remarka-
bly increases in amplitude but most of it is damped out below 3 Hz.
Thus, it appears as if it is permitted to increase the effective value of the
vertical acceleration over a certain threshold - as occurs in running -
but at the same time most of the acceleration spectrum must be
compressed within a low frequency range so as to stay away from those
frequencies where resonances of body parts occur.

The estimation of energy requirements is also relevant for the evalua-


tion of gait. Indirect calorimetry is the obvious technique to be used in
this case. Also heart rate is regarded as an indirect index of metabolic
energy expenditure during the execution of physical exercises. The use
of mechanical quantities in this context poses several interpretative
problems. This is basically due to the complex and differentiated
relationships existing between metabolic energy and kinetics of muscu-
lar contraction. In this context, the human body is often devided into
two sub-systems, the lower limbs and the upper body. The former
support and transport the latter. How this is done, in terms of mechani-
cal energy involved, is described by Saunders et al. (1953), Ralston and
Lukin (1969) and Cappozzo et al. (1976), among others. One basic
aspect in this context is the following. During walking, at the com-
monly adopted speed, the potential and kinetic energies of the upper

A. Capporzo / Gait una!vsis methodologv 4-l

body are almost in antiphase and the total energy variations are
relatively small. Thus, the upper body acts as a quasi-conservative
system. This is how muscular work is minimized and, thus, metabolic
energy expenditure is minimized. The total mechanical energy associ-
ated with the upper body during walking in the range of speeds of
progression was assessed in the author’s laboratory. This was carried
out through a 3-D analysis of the head, upper torso, and upper limbs
motion in five normal adult male subjects who each performed four
walk-trials. From the mechanical energy function the positive work
done by the lower limbs on the upper body during a walking cycle was
determined. This quantity was then normalized with respect to cycle
duration and body weight. In fig. 6 the relevant regression curve is
represented. This may be considered as normative data. Superimposed
on it, the normalized positive work found during four walking trials of
an above-knee amputee and two walking trials of a patient with knee
ankylosis are represented. In the amputee the positive work exchanged
between lower and upper body is larger than in normal walking. This
indicates that the lower limb muscles have to produce more mechanical
energy. The data of the other patients, on the contrary, is either within
or below normality. Although specific measurements were not made, it
is reasonable to exclude the hypothesis that this patient’s gait could be

SPEED Cm s -‘I

Fig. 6. Positive work done by the lower limbs on the upper body normalized with respect to body
weight and stride duration, versus speed of progression. The regression equation plus and minus
two standard deviations relative to twenty normal subject walk trials is represented together with
the results relative to four above-knee tests (squares) and two tests of a patient with knee ankylosis.

48 A. Cappozzo / Gait analysis me?hodolog~

more economical than normal gait. In fact, this patient tip-toed with
the healthy leg in order to gain clearance from the ground for the
affected limb. It is therefore likely that extra metabolic energy is
required by the quasi-isometric contraction of the sound limb ankle,
while the energy exchange between lower and upper body is kept close
to normality. The above examples confirm the difficulty of drawing
correct information concerning metabolic energy requirements from
mechanical energy data. However, mechanical energy analysis permits
the collection of information about energy flows from different muscu-
lar districts to the body segments, and between body segments (Winter
1979). This is opposed to metabolic energy expenditure assessment,
which is unavoidably a global assessment.

Conclusions

In this paper a brief account has been given of what gait analysis is and
what its objectives may be. Emphasis was placed on the distinction
between joint function assessment and gait evaluation. Experimental
and analytical methodologies have been described which permit the
accomplishment of the above objectives.

As far as the clinical application of gait analysis is concerned, the


following is noteworthy. Once the “numbers” that describe the mecha-
nics of the patient’s gait have been assessed, only the first step -
perhaps the least crucial step - has been accomplished. Following on
the analysis, based on the measurements and calculations referred to
previously, a synthesis of the data must be made which will supply the
clinically relevant information. Since the phenomenon being dealt with
is very complex and multifaceted, it is unlikely that the process of
synthesis can be completely carried out in numerical terms, that is, by
computer. Thus, the speculative intervention of the clinician in a rather
early stage of the synthesis can not be precluded. What message, both
in terms of content and form, should the computer provide the clini-
cian, and what methodology should the clinician employ in order to
deal with the message so as to accomplish an objective gait evaluation?
These are problems the solution of which biomechanicians have, as yet,
not perfected. In this respect some ideas have been put forward in this
paper. Implementation of such ideas is still at the experimental phase

A. Cupporzo / Gail cmalysis methodolou 49

and the validation of the clinical orientation of the methodology


presented here is still in the process of being substantiated.

References

Amar, J., 1916. Trottoir dynamographique. Comptes Rendus de I’Acadtmie des Sciences 163,
130-132.

Bernstein, N.A., 1967. The coordination and regulation of movements. Oxford: Pergamon Press
Ltd.

Borelli. G.A., 1680. De mom animalium. Roma.


Braune, C.W. and 0. Fisher, 1895. Der Gang des Menschen. Internationale Abhandlungen der

Mathematisch Physisch Koniglich Sachsischen Gesellschaft fir Wissenschaften 21, 151-324.


Bresler, B. and J.P. Frankel, 1950. The forces and moments in the leg during level walking.

Transactions American Society of Mechanical Engineers 72, 27-36, (paper No. 48-A-62).
Brumbaugh, R.B., R.D. Crowninshield, W.F. Blair and J.G. Andrews, 1982. An in-viva study of

normal wrist kinematics. Journal of Biomechanical Engineering 104, 176-181.


Cappozzo, A., 1981. Analysis of the linear displacement of the head and trunk during walking at

different speeds. Journal of Biomechanics 14, 411-425.


Cappozzo, A., 1982. Low frequency self-generated vibration during ambulation in normal men.

Journal of Biomechanics 15, 599-609.


Cappozzo, A., 1983. The forces and couples in the human trunk during level walking. Journal of

Biomechanics 16, 265-277.


Cappozzo, A., 1984. Compressive loads in the lumbar vertebral column during normal level

walking. Journal of Orthopaedic Research 1, 292-301.


Cappozzo, A., F. Figura, M. Marchetti and A. Pedotti, 1976. The interplay of muscular and

external forces in human ambulation. Journal of Biomechanics 9, 35-43.


Cappozzo, A., F. Figura, F. Gazzani, T. Leo and M. Marchetti, 1982. Angular displacements in the

upper body of AK amputees during level walking. Prosthetics and Orthotics International 6,
131-138.

Chandler, R.F., C.E. Clauser, J.T. McConville, H.M. Reynolds and J.W. Young, 1975. Investiga-
tion of inertial properties of the human body. Report No. AMRL-TR-74-137, Wright-Patterson
AFB, Ohio, NC.
Chao, E.Y., 1980. Justification of triaxial goniometer for the measurement of joint rotation.
Journal of Biomechanics 13, 98991006.

Clauser, C.E., J.T. McConville and J.W. Young, 1969. Weight, volume and centre of mass of
segments of the human body. Report No. AMRL-TR-69-70, Wright-Patterson AFB, Ohio, NC.

Dempster, W.T., 1955. Space requirements of the seated operator. WADC Technical Report
55-159, Wright-Patterson AFB, Ohio, NC.

Eberhart. H.D., V.T. Inman, J.B. de CM. Saunders, A.S. Levens, B. Bresler and T.D. McGowan,
1947. Fundamental studies of human locomotion and other information relating to design of
artificial limbs. Report to the National Research Council, Committee on artificial limbs.
Berkeley, University of California.

Grood, E.S. and W.J. Suntay, 1983. A joint coordinate system for the clinical description of three
dimensional motions: application to the knee. Journal of Biomechanical Engineering 105,
136-144.

Hanavan, E.P., 1964. A mathematical model of the human body. Report No. AMRL-TR-102,
Wright-Patterson AFB, Ohio, NC.

Hatze, H., 1980a. Neuromusculoskeletal control systems modelling a critical survey of recent
developments. IEEE Transactions on Automatic Control AC-25, 375-385.

Hatze, B.. 1980b. A mathematical model for the computational determination of parameter vjalues
of anthropometric segments. Journal of Biomechamcs 13. 833-843.

Hayes, W.C.. J.D. Gran, M.L. Nagurka, J.M. Feldman and C. Oatis. 1983. Leg motion analysis
during gait by multiaxial accelerometry: theoretical foundations and prehminary valtdation.
Journal of Biomedical Engineering 105. 283-289.

Jarrett, M.O., B.J. Andrews and J.P. Paul. 1974. Quantitative analysis of locomotton using
television. ISPO World Congress. Montreux. Switzerland.

Jensen, R.K., 1978. Estimation of the biomechanical properties of three body types using a
photogrammetric method. Journal of Biomechanics 11. 349-358.

Lindholm, L.E.. 1974. ‘An optoelectronic instrument for remote on-live movement monitoring’. In:
R.C. Nelson and C.A. Morehouse (eds.). Biomechamcs IV. Baltimore, MD: University Park
Press.

Macellari, V., 1983. CoSTEL: a computer peripheral remote sensing device for 3-dimensional
monitoring of human motion. Medical and Biological Engineering and Computing 21, 311-318.

Marey, I&J., 1885. Le methode graphique dans les sciences experimentales. 2nd edition with
supplement: le developpement de la methode graphique par I’emploi de la photographie. Paris:
G. Masson.

Mitchelson, D., 1975. ‘Recording of movement without photography’. In: D.W. Grieve, D. Miller,
D. Mitchelson, J.P. Paul and A.J. Smith (eds.), Techniques for the analysis of human
movement. London: Lepus Books.
Muybridge, E., 1887. Animal locomotion, an electro-photographic investigation of consecutive
phases of animal movements (11 volumes). Philadelphia, PA: J.B. Lippincott.

Paul, J.P.. 1966. Biomechanics of the hip joint and its clinical relevance. Proceedings of the Royal
Society of Medicine 59, 943-949.

Ralston, H.J. and L. Lukin, 1969. Energy levels of human body segments during level walking.
Ergonomics 12, 39-46.

Saunders, J.B. deC.M., V.T. Inman and H.D. Eberhardt, 1953. The major determinants in normal
and pathological gait. Journal of Bone and Joint Surgery 35A, 543-558.

Seireg, A. and R.J. Arvikar, 1975. The prediction of muscular load sharing and joint forces in the
lower extremities during walking. Journal of Biomechanics 8, 89-102.

Winter, D.A., 1979. A new definition of mechanical work done in human movement. Journal of
Applied Physiology: Respiration, Environment, and Exercise Physiology 46, 79-83.

Winter, D.A., R.K. Greenlaw and D.A. Hobson, 1972. Television-computer analysis of kinematics
of human gait. Computers and Biomedical Research 5, 498-504.

Woltring, H.J., 1982. ‘Estimation and precision of 3-D kinematics by analytical photogrammetry’.
In: J.P. Paul, M.M. Jordan; M.W. Ferguson-Pell and B.J. Andrews (eds.), Computing in
medicine. London: MacMillan Press.

Zatsiorsky, W.M. and V. Seluyanov, 1983. ‘The mass and inertia characteristics of the main
segments of the human body’. In: H. Matsui and K. Kobayashi (eds.), Biomechanics VIII-B.
Champaign, IL: Human Kinetics Publishers.

You might also like