Tweet
Some interesting comments were made by Dwight Meglan on a Technical
Report from the US army, describing regression models for estimation
of inertial parameters of body segments, I'll certainly try to get
hold of that report (thanks for the reference, Dwight), and in the mean-
time want to add some of my own thoughts.
First, I couldn't agree more with Dwight's remark about the coordi-
nate system used. When I do biomechanical modelling, I want to have
a matrix representation of the inertia tensor in the same coordinate
system that I use to specify joint centers and muscle attachment
points. This is an anatomical coordinate system, so why not develop
regression models for all six components of this symmetric matrix?
Principal moments of inertia simply are not sufficient.
The same applies of course to the location of the center of mass.
Secondly, if I understand it correctly, stereophotometric methods
were used to 'measure' the inertial parameters that were then corre-
lated to anthropometric measurements. I wonder if the photometric
methods have ever been validated by direct measurement of the
parameters in cadaver segments. No such study has been done yet,
as far as I know. This certainly not a pleasant task, but in my
opinion the only method to justify the photometric methods. Also,
this would solve the problem of unknown mass density distribution.
In a study I did on inertial parameters of ponies, a photometric method
was used to estimate segmental parameters (volume, center of volume
and volume moment of inertia), and the inertial parameters were then
measured post mortem in the separated and frozen segments of the same
subjects. I then defined for each parameter a multiplication factor,
a kind of 'effective mass density', for each volume parameter to
produce an optimal estimate for the corresponding inertial parameter.
Optimal here means a minimal difference with the directly measured
values, using a least-squares fit on the group of five subjects.
These factors then take care of inhomogeneous mass distri-
bution, but also of systematic errors caused by misrepresentation
of the segment shape in the photometric reconstruction model
(Jensen's elliptic zone model was used).
In my case the correlation between photometric estimates and direct
measurements was poor. This was mainly due to some flaws in the
experimental setup used for the photometric measurements, but could
also indicate a more fundamental shortcoming of the method.
Possibly someone having more experience with this technique
can comment on this.
Ton van den Bogert, Dept. of Veterinary Anatomy,
University of Utrecht, The Netherlands.
Report from the US army, describing regression models for estimation
of inertial parameters of body segments, I'll certainly try to get
hold of that report (thanks for the reference, Dwight), and in the mean-
time want to add some of my own thoughts.
First, I couldn't agree more with Dwight's remark about the coordi-
nate system used. When I do biomechanical modelling, I want to have
a matrix representation of the inertia tensor in the same coordinate
system that I use to specify joint centers and muscle attachment
points. This is an anatomical coordinate system, so why not develop
regression models for all six components of this symmetric matrix?
Principal moments of inertia simply are not sufficient.
The same applies of course to the location of the center of mass.
Secondly, if I understand it correctly, stereophotometric methods
were used to 'measure' the inertial parameters that were then corre-
lated to anthropometric measurements. I wonder if the photometric
methods have ever been validated by direct measurement of the
parameters in cadaver segments. No such study has been done yet,
as far as I know. This certainly not a pleasant task, but in my
opinion the only method to justify the photometric methods. Also,
this would solve the problem of unknown mass density distribution.
In a study I did on inertial parameters of ponies, a photometric method
was used to estimate segmental parameters (volume, center of volume
and volume moment of inertia), and the inertial parameters were then
measured post mortem in the separated and frozen segments of the same
subjects. I then defined for each parameter a multiplication factor,
a kind of 'effective mass density', for each volume parameter to
produce an optimal estimate for the corresponding inertial parameter.
Optimal here means a minimal difference with the directly measured
values, using a least-squares fit on the group of five subjects.
These factors then take care of inhomogeneous mass distri-
bution, but also of systematic errors caused by misrepresentation
of the segment shape in the photometric reconstruction model
(Jensen's elliptic zone model was used).
In my case the correlation between photometric estimates and direct
measurements was poor. This was mainly due to some flaws in the
experimental setup used for the photometric measurements, but could
also indicate a more fundamental shortcoming of the method.
Possibly someone having more experience with this technique
can comment on this.
Ton van den Bogert, Dept. of Veterinary Anatomy,
University of Utrecht, The Netherlands.