View Full Version : Body segment parameters

unknown user
08-25-1989, 01:44 AM
Hello All,

Recently I had the occasion to talk to a number of people from Europe and
the US who have been or are currently involved in mathematical modelling
of human motion. In the course of the conversations we talked about the
estimation of body segment mass and inertia parameters and I found that
no one that I talked to knew about a US Air Force report from 1981 which
is fairly extensive in its techniques for estimating segment data. Thus,
here is the relavent information for those who may interested in obtaining
a copy as well as the abstract and a brief summary (i.e., a subjective
evaluation) of how it has worked out for my application, a large program
to analyze kinematics, kinetics, energetics, and muscle dynamics during
human motion.
I'm not sure where you can actually order this report (it was given to
me by Necip Berme and he is on sabbatical this year so I can't find out
how he got it). If someone does know, please post a message as to how it
is done.

Report #: AFAMRL-TR-80-119 (Approved for public release)
December 1980, p. 109 (Unlimited distribution)
Title: Anthropometric Relationships of Body and Body Segment Moments
of Inertia
Authors: *McConville, J.T., *Churchill, T.D., +Kaleps, I., +Clauser, C.E.,
~Cuzzi, J.
* Anthropology Research Project, Inc.
Yellow Springs, Ohio 45387 USA
+ Air Force Aerospace Medical Research Laboratory
Aerospace Medical Division
Air Force Systems Command
Wright-Patterson Air Force Base, Ohio 45433 USA
~ Texas Institute for Rehabilitation and Research
Houston, Texas 77030

This report documents the results of a study aimed at demonstrating that mass
distribution properties of the human body and its segments can be predicted
from anthropometric dimensions. Investigators combined stereophotometric
and anthropometric techniques to measure 31 male subjects. Bodies were
photographically segmented into 24 parts and their volume, centers of volume
and principal moments were measured about the three principal axes of inertia
which were established with reference to anatomical axis systems based on
easily located body landmarks.
Seventy-five body size variables were measured anthropometrically and an
additional 10 dimensions were derived from the measured variables. Multiple
regression equations were devised for the total body and for each of the
segments using the most highly correlated variables on each segment and
stature and weight for determining volume and principal moments of inertia.
Included is a brief review of the literature with emphasis on those studies
by authors which provide the rationale for the validity of the stereophoto-
metric method in determining mass distribution properties of living subjects.
The data analysis section provides tables illustrating each of the segments
and their axis systems, the segmental data established in this study, and
a series of regression equations estimating volume and principal moments
from anthropometric measurements

The most unique feature of the methods used here is the tranformation
from an anatomically derived coordinate system (CS) to the segment principal
inertia CS located at the segment center of volume. This allows the
determination of a CS from easy to find and measure anatomic points. This
feature, in my opinion, is flawed unfortunately. The authors have gone
to considerable trouble to derive regression equations for the segment
parameters but there are no such equations for the transformation from the
anatomic CS to the principal inertia CS. Instead the mean and standard
deviation of the location of the volume center from the anatomic CS center
for all subjects is given. There is no subject specificity available.
The same applies to the rotation from the anatomic CS to the principal inertia
CS, which is given in terms of yaw-pitch-roll euler angles. The regression
equations on the other hand are well done. For most segments there are
15 to 20 different combinations of anthropometric measurements whose resulting
equation coefficients are given along with the multiple correlation coefficient
and the standard error of estimate. Most of the standard error of estimates
are around 10% of the mean of the segment parameter. Since only the volume
and moment of inertia of the volume are generated it is necessary to assume
a homogenous distribution of mass within the segment and use this to calculate
the mass and mass moments of inertia. These densities are discussed in
the report. Lastly, this study has the problem that just every study I've
looked at that deals with estimating segment inertia parameters (about 10
to 15 - experimentally oriented studies, not the computational studies
such as those by Hatze and Vaughan) has - Namely, the data used was collected
from adults and primarily males (in this case all males). We deal with
children having neuromuscular disorders just as often as adults and it would
be nice to have proper data on their segment parameters (I remember reading
a brief summary which looked at some Japanese children in one of the ISB
conference abstracts from a while back).
I have programmed the estimation equations in their most simple form which
uses only the height and weight of the subject to calculate the segment
parameters and then have placed the principal inertia CS using the estimates
of Dempster. It was done this way because we are interested in doing
retrospective analyses of the gait data collected by the lab over the past 8
or so years (about 1000 subjects) and these studies contain only the height and
weight of the subjects in addition to all the marker data (we use Vicon),
force plate and EMG data. For new studies we are setting up the option to
measure some of the anatomic points (there are 77 points required to get the
anatomic CS's !), but it is considerably more difficult to use this protocol
in a clinical situation where you subjects who tire easily or are uncooperative
such as childern :-). We are only using thirteen body segments so I have
combined some of the study's segment data to estimate segment parameters
using the parallel axis theorem, etc... The sum of the estimated segment
masses is approximately 10% higher than the actual weight of the subject
in every case tried so far. The effect of this difference upon kinetic
calculations, such as the intersegmental joint resultant moments and force,
has not been studied here yet, but I'll get around to it. (I'm sure others
have done this already but I can't remember any specific examples right
now, anybody care to comment ?)
One last note: I am not passing judgement upon the various methods used
to estimate segment mass-inertia parameters. The above is intended to get
others to discuss what I think is a fundamental problem in biomechanics
which has not been answered to the degree of generality that I feel it should.
The recent paper in the Journal of Biomechanics (this past spring, I can't
remember which issue) using MRI to estimate segment mass-inertia parameters
seems to indicate a possible answer as to how to get data specific to an
individual subject. But, how bullet-proof is the software ? Can it be
developed to the point that it can used everyday ? How long does an analysis
take and how much operator supervision is needed ? Will this technique
be enhanced to the point where it is practical and economical to be used
for clinical human motion analyses ? I wish those who working on these
problems the best and I hope to use the fruits of their labor in the not
too distant future. Enough said (by me), for now... :-)

What is the world coming to that people are obligated to include disclaimers
with everthing ? In that spirit, anything expressed here was not said by
anybody. Period.

Dwight Meglan, Phd-in-training & Research Engineer
Ohio State University Hosipital
Gait Analysis Laboratory
1054 Dodd Hall
471 Dodd Drive
Columbus, OH 43210 USA (614)293-3808 Email: meglan%gait1@eng.ohio-state.edu