Tweet
Dear BIOMCH-L subscribers,
************************************************** ********
* INERTIAL BODY SEGMENT PARAMETERS: OSTEOKINETICS' CORP. *
* COMMERCIAL E-MAIL OF MAY 11. *
************************************************** ********
In a commercial BIOMCH-L message of May 11, 1994, OsteoKinetics
Corporation of Newton, MA, USA, advertised a new product, termed
Inertiator, for determining the inertial properties of body
segments. Allegedly "THE FIRST AND ONLY commercially available
system that provides accurate, subject-specific and non-invasive
estimates" of inertial segment properties, the system achieves,
according to the announcement, an accuracy in the order of
2% of the true values of all segmental parameters (such as, for
instance, the principal moments of inertia).
Not only must the assertion regarding the system accuracy be
disputed, but the advertised system is also neither "THE FIRST"
nor the "ONLY". In actual fact, we introduced a video-based
technique for the determination of inertial body segment
parameters and the associated calibration and data acquisition
system already in July 1992 at SPIE's 1992 International
Symposium on Optical Applied Science and Engineering in San
Diego, and published it in all detail in Hatze, H. and Baca, A.
(1992): "Contact-free determination of human body segment
parameters by means of videometric image processing of an
anthropomorphic body model." In: Applications of Digital Image
Processing XV, SPIE-P-Vol. 1771, pp. 536-546. Our preliminary
reports on this method have appeared as early as 1991 (Hatze, H.
and Baca, A. in Application of Digital Image Processing XIV,
SPIE-P-Vol. 1567, pp. 264-276).
We are aware of a recently published Technical Note in the
Journal of Biomechanics, Vol. 26, No. 8, pp. 1011-1016 (1993) by
O. Sarfaty and Z. Ladin of Boston University, MA, USA ("A
video-based system for the estimation of the inertial properties
of body segments") in which essentially the same technique as
ours is described, albeit without mentioning any of the above or
even older, but pertinent, references relating to our work. To
put things into the correct perspective and to clarify the issue
concerning the original authorship, we have sent a "Letter to
the Editor" of the Journal of Biomechanics which letter has been
accepted and will appear in one of the next issues of this
journal.
In this connection it is interesting to observe that
OsteoKinetics Corporation joined BIOMCH-L and placed its
commercial announcement just after the editor of the Journal of
Biomechanics asked the authors of the above Technical Note for
an explanatory comment. This, however, is presumably pure
coincidence.
At present, we are perfecting our system for the video-based,
contact-free determination of all body segment parameters by
using a highly sophisticated image processing procedure
employing sub-pixel resolution, an optimized procedure for
automatic threshold determination, elimination of nonlinear
image distortions and parallaxe errors resulting from
varying segment-boundary-to-camera distances, and other features
for error reduction that are all unique to our system and are
not found elsewhere. In addition, our anthropomorphic segment
models fully account for changing density distributions across
and along the segments, and the geometrical model elements
selected are in accordance with actual segment morphologies,
which results in further improvments in accuracy, in contrast to
models used by others. Even with this sophisticated technique,
the overall accuracy that can be achieved today is about 3%,
with maximum errors of about 7% for some of the moment of
inertia values of smaller segments. The correctness of an error
bound of 2% for all inertial parameter values, as announced by
OsteoKinetics Corporation, is therefore more than doubtful.
Finally, our system works with normal video-cameras, the output
of which is processed by an appropriate frame grabber board
installed in a PC. The calibration frame and the dark background
can easily be built by the user himself, based on our detailed
specifications. Therefore, no special and hence expensive
hardware components need to be bought by the user.
We do not want to elaborate further on our system since BIOMCH-L
should definitely not be misused for posting commercial
messages. This has already been emphasized some time ago, and
quite correctly so. On the other hand, it was obviously
necessary under the given circumstances to react to
OsteoKinetics' Corp. commercial announcement in a corrective
manner.
Sincerely
Herbert Hatze and Arnold Baca
Department and Laboratory of Biomechanics, IfS,
University of Vienna
Auf der Schmelz 6, A-1150 Vienna, AUSTRIA
************************************************** ********
* INERTIAL BODY SEGMENT PARAMETERS: OSTEOKINETICS' CORP. *
* COMMERCIAL E-MAIL OF MAY 11. *
************************************************** ********
In a commercial BIOMCH-L message of May 11, 1994, OsteoKinetics
Corporation of Newton, MA, USA, advertised a new product, termed
Inertiator, for determining the inertial properties of body
segments. Allegedly "THE FIRST AND ONLY commercially available
system that provides accurate, subject-specific and non-invasive
estimates" of inertial segment properties, the system achieves,
according to the announcement, an accuracy in the order of
2% of the true values of all segmental parameters (such as, for
instance, the principal moments of inertia).
Not only must the assertion regarding the system accuracy be
disputed, but the advertised system is also neither "THE FIRST"
nor the "ONLY". In actual fact, we introduced a video-based
technique for the determination of inertial body segment
parameters and the associated calibration and data acquisition
system already in July 1992 at SPIE's 1992 International
Symposium on Optical Applied Science and Engineering in San
Diego, and published it in all detail in Hatze, H. and Baca, A.
(1992): "Contact-free determination of human body segment
parameters by means of videometric image processing of an
anthropomorphic body model." In: Applications of Digital Image
Processing XV, SPIE-P-Vol. 1771, pp. 536-546. Our preliminary
reports on this method have appeared as early as 1991 (Hatze, H.
and Baca, A. in Application of Digital Image Processing XIV,
SPIE-P-Vol. 1567, pp. 264-276).
We are aware of a recently published Technical Note in the
Journal of Biomechanics, Vol. 26, No. 8, pp. 1011-1016 (1993) by
O. Sarfaty and Z. Ladin of Boston University, MA, USA ("A
video-based system for the estimation of the inertial properties
of body segments") in which essentially the same technique as
ours is described, albeit without mentioning any of the above or
even older, but pertinent, references relating to our work. To
put things into the correct perspective and to clarify the issue
concerning the original authorship, we have sent a "Letter to
the Editor" of the Journal of Biomechanics which letter has been
accepted and will appear in one of the next issues of this
journal.
In this connection it is interesting to observe that
OsteoKinetics Corporation joined BIOMCH-L and placed its
commercial announcement just after the editor of the Journal of
Biomechanics asked the authors of the above Technical Note for
an explanatory comment. This, however, is presumably pure
coincidence.
At present, we are perfecting our system for the video-based,
contact-free determination of all body segment parameters by
using a highly sophisticated image processing procedure
employing sub-pixel resolution, an optimized procedure for
automatic threshold determination, elimination of nonlinear
image distortions and parallaxe errors resulting from
varying segment-boundary-to-camera distances, and other features
for error reduction that are all unique to our system and are
not found elsewhere. In addition, our anthropomorphic segment
models fully account for changing density distributions across
and along the segments, and the geometrical model elements
selected are in accordance with actual segment morphologies,
which results in further improvments in accuracy, in contrast to
models used by others. Even with this sophisticated technique,
the overall accuracy that can be achieved today is about 3%,
with maximum errors of about 7% for some of the moment of
inertia values of smaller segments. The correctness of an error
bound of 2% for all inertial parameter values, as announced by
OsteoKinetics Corporation, is therefore more than doubtful.
Finally, our system works with normal video-cameras, the output
of which is processed by an appropriate frame grabber board
installed in a PC. The calibration frame and the dark background
can easily be built by the user himself, based on our detailed
specifications. Therefore, no special and hence expensive
hardware components need to be bought by the user.
We do not want to elaborate further on our system since BIOMCH-L
should definitely not be misused for posting commercial
messages. This has already been emphasized some time ago, and
quite correctly so. On the other hand, it was obviously
necessary under the given circumstances to react to
OsteoKinetics' Corp. commercial announcement in a corrective
manner.
Sincerely
Herbert Hatze and Arnold Baca
Department and Laboratory of Biomechanics, IfS,
University of Vienna
Auf der Schmelz 6, A-1150 Vienna, AUSTRIA