01-06-2002, 09:29 PM
Dear Colleagues,
as announced a few weeks ago on BIOMCH-L, the coordinators of the BIONET
project, the BIOMCH-L moderator, and Prof. Hatze agreed to conduct on
BIOMCH-L the discussion of contemporary issues in biomechanics on a
one-by-one basis in a mutually cooperative manner. The objective is to
identify on a world-wide scale those research areas in biomechanics which
need urgent attention and adequate funding. The first topic was identified
by Dr. Viceconti on 12 December 2001 and is still under discussion.

You are all cordially invited to participate in these discussions and
thereby provide a valuable contribution to the development of a new phase
of biomechanical research which may be termed Second Generation Biomechanics.

Following the suggestion of the list moderators, we recommend to all
colleagues willing to take a position on the matter to write directly to
the entire list. The present Topic #2 introduced below is more general,
intriguing, and should elicit a diverse spectrum of opinions.

Hoping for lively discussions,

Herbert Hatze,
Alberto Leardini on behalf of the BIONET Consortium

Here is the second topic for the discussion forum:
Proponent: A. Leardini)


Human motion is the visible result of a series of processes that take place
in the neuromusculoskeletal system. The interest in human motion analysis
arises from the fact that, hopefully, motion analysis will yield
information on these underlying (possibly pathological) processes which are
not directly observable except, perhaps, by using highly invasive techniques.

HUMAN MOTION is usually defined as the set of time functions of a SPECIFIC
NUMBER OF CONFIGURATIONAL COORDINATES, given on a certain time interval.
And this is where the problems begin. Configurational coordinates define
the configuration of a MODEL and NOT that of the human body as the real
biosystem to be investigated. Obviously, body models can range from very
simple with only a few degrees of freedom to fairly complex. In all cases,
however, they represent a simplified analog of the actual biosystem. This
simulation model of the human body, no matter how oversimplified, will
yield outputs (such as ground reaction forces) that are always consistent
with its dynamics. Whether or not these outputs are biologically realistic
is another matter.

In MOTION ANALYSIS (myoskeletal inverse dynamics) THE SITUATION IS
DIFFERENT: We are given a (generally inadequate) human body model, invert
its equations of motion, and then feed these with inputs (body segment
parameter values, the recorded and processed motion data and their time
derivatives, the recorded ground reaction forces, pressure distribution
functions, etc.) that have been determined experimentally with varying
degrees of accuracy. And this is THE FUNDAMENTAL PROBLEM OF MYOSKELETAL
of the human musculoskeletal system are produced from input data that
originate from complex REAL BIOSYSTEMS and therefore are, in principle,
INCOMPATIBLE with the models. This may lead to large errors and makes the
results of the analysis unreliable, as demonstrated in a paper appearing in
the January-2002-issue of the Journal of Biomechanics Nr. 35/1, pp.
109-115: The fundamental problem of myoskeletal inverse dynamics and its
implications by Herbert Hatze (already printable from the journal web site).

The IMPLICATIONS of this discrepancy are profound indeed. Except for
quasi-static motions with minor accelerative phases, inverse dynamically
computed quantities such as resultant joint moments and powers, shear and
compressive joint loads, muscular work contributions, etc. are incorrect to
varying degrees. The magnitudes of the errors are, if at all, hard to
assess because experimental observation of the computed quantities is
usually prohibited owing to the involvement of invasive techniques. But
even purely kinematic descriptions of the observed motions are unreliable
for reasons outlined below. This is a depressing situation, especially for
biomechanists, engineers, medical professionals, and technicians involved
in the clinical assessment of pathological motion looking at several
different motor tasks such as gait, stair climbing-descending, chair
raising-sitting, etc., or in investigations on the efficiency of work or
other motions.

I shall now formulate some postulates that should help identify the basic
problems relating to the current topic. I shall do this in a deliberately
provocative style, hoping to stimulate lively reactions.

1. HUMAN BODY MODELS currently used in biomechanics are unrealistic and,
for most investigations, inadequate. (The modeling issue will be discussed
in more detail later as a separate topic). The human limb system can not in
general be considered as, or adequately approximated by, an assemblage of
interlinked rigid body segments. Rather, it is a RIGIDO-VISCOELASTIC HYBRID
composed of numerous (nearly) rigid bony segments, interconnected by
(visco)elastic tissue structures (ligaments, cartilage, joint capsules,
etc.), and having attached to them (visco)elastic active (musculotendinous
units) and passive (e.g. inner organs, blood vessels, etc.) components,
which, in turn, are interconnected among themselves. In reality, there
exist no limb segment boundaries: joints are spanned by elastic structures
(mainly muscles) that belong simultaneously to two or more limbs. The
motions of the elastic body structures relative to the skeleton can be
highly significant, especially in motions with pronounced accelerative
phases, such as impacts (e. g. heel strike in gait). The consistency of
some of these elastic structures (muscles) changes with their degree of
contraction. The detailed three-dimensional structures (bones, ligaments,
muscles, tendons, soft tissue structures, etc.) of terminal segments
(hands, feet) have to be fully accounted for, which is hardly the case with
present-day models. Joints generally have six degrees of freedom (with the
possible exception of the hip joints) and are exceedingly complex
structures whose detailed functional behavior still awaits elucidation.

2. Current MOTION RECORDING TECHNIQUES rely almost exclusively on the
detection of the spatial motion of passive or active position or (and)
angular orientation sensors (markers, goniometers, magnetic field sensors,
etc.) These devices are usually attached to the skin or clothing of a
subject and are subject to shifts relative to the underlying tissue
(muscle, fat, bone). This creates the well known skin shifting problem
which introduces systematic low frequency errors by generating non-genuine
artificial motion patterns that are superimposed on the skeleton motion.
But, as has been mentioned under point 1, the motion of the skeleton
represents only the rigid body part of the total body motion, excluding the
motion of the many elastic structures that move relative to the skeleton.
Thus, there is no justification at all to define the motion of the skeleton
as "the motion of the body". In fact, this notion is incorrect in
principle. The oscillations of a goniometer, for instance, attached to the
thigh and lower leg may indeed represent genuine oscillatory motions of the
quadriceps muscle as elastic structure, and therefore real and possibly
significant submotions of the human body. Such low frequency submotions
will influence the recorded ground reaction forces of the real biosystem
but will not be represented by the inverse dynamics of inadequate body
models, thereby introducing an error source. Similar arguments hold true
also for the placement of other motion recording devices such as markers or
magnetic sensors, and the interpretation of their output signals. We would
be well advised to begin thinking about what our motion sensors actually

3. Closely related to point 2 above is the problem of appropriately
PROCESSING THE MOTION DATA (conditioning, filtering, and time derivative
computation of noisy data). Especially the second derivatives are extremely
sensitive to even small errors in the kinematic motion data. In fact,
computing second derivatives from noise-contaminated data sequences is a
so-called incorrectly posed problem for which adequate solutions (via
regularization techniques) are very difficult to achieve and are always
unreliable to a certain degree. But the computation of correct second
derivatives is a prerequisite for the correct inverse dynamical computation
of all kinetic quantities. Still an unsolved problem.

4. Most present-day motion analysis systems compute, if at all, only the
skeletal system characteristics (resultant joint moments, shear and
compressive joint loads, etc.). The REAL AND DIAGNOSTICALLY MOST
DYNAMICS. In other words, we would like to obtain not only the resultant
joint moments that generated an observed motion but also the histories of
all the muscle forces (and possibly even their neural control inputs) as
well as their individual torque contributions that produced the computed
resultant joint moments. These important problems are also largely unsolved
and will be the subject of two other topics ("Inadequate Muscle Models" and
the "Myoskeletal Indeterminancy Problem") to be discussed later on in the
present series.

5. Incorrect values of SEGMENT PARAMETERS (segmental lengths, masses,
volumes, principal moments of inertia, components of mass centroid
locations, etc.) also significantly influence the quality of the motion
analysis results. These parameter values are highly subject-specific and
must be determined for each subject individually. (To be discussed
separately in another topic in this series). It should be kept in mind that
some of these parameters will relate to rigid substructures such as bones
while others will characterize the properties of viscoelastic structures
such as muscles.

Herbert Hatze

The above formulation by Prof. Hatze is particularly exhaustive and clear.
I would suggest you all to contribute by:
1. identifying what you feel are the most critical issues in this context
among the ones mentioned;
2. pointing out the most significant solution proposals from the literature
in the past years;
3. proposing possible directions for future investigations.

Authoritative researchers will be invited to the discussion, but the voice
from any of you will be particularly important. A summary will be provided.

Alberto Leardini

************************************************** ************************
Alberto Leardini, DPhil
Movement Analysis Laboratory
Centro di Ricerca Codivilla-Putti
Istituti Ortopedici Rizzoli
Via di Barbiano 1/10, 40136 Bologna ITALY
tel: +39 051 6366522
fax: +39 051 6366561
email: leardini@ior.it

"Where is the Life we have lost in living,
Where is the wisdom we have lost in knowledge,
Where is the knowledge we have lost in information."
Thomas Stearns Eliot, Choruses from ''The Rock'' (1934)
************************************************** ************************

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l