PDA

View Full Version : replies to motion analysis on the Mac



Robinovitch, Steve
01-18-1995, 12:46 PM
Dear BIOMCH-L readers: a couple of months ago, I asked some questions regarding
motion analysis equipment for the MAC. I received several replies, which I have
finally got around to compiling. Thanks very much to all who replied.

__________________________________________________ ___________________________

My original posting was:


Dear BIOMCH-L readers:

We are starting a new orthopaedic biomechanics research laboratory at San
Francisco General Hospital. One equipment item we are considering to purchase is

a cinegraphic motion analysis system for either 2-D or 3-D studies. I wanted to
ask BIOMCH readers two questions related to this:

1) do you know of any motion analysis systems which can be hooked up to a
Macintosh computer (with a PowerPC chip and National Instruments A/D board)
for data acquisition and control? The only one I have heard of is from a
company called MacReflex. Do any BIOMCH readers have experience with this
system?

2) how many cameras are required for acquisition of 3-D kinematics? I have
heard responses which vary from four cameras to more than six.

I will post all responses. Thanks.

Steve Robinovitch, Ph.D.
Asst. Professor, Dept. of Ortho. Surgery (UCSF)
Director, Biomechanics Laboratory
San Francisco General Hospital
tel: (415)206-6864
FAX: (415)206-3288

__________________________________________________ _________
Hi Steve:

About two years ago I looked into purchasing a motion analysis system for
our lab (UCSF/UCB Ergonomics lab). At that time, and I don't think much
has changed since then, the only one is the MacReflex system. Lawrance
Livermore Labs have one that they purchased about two years ago. We
really wanted to go with a mac based system but the MacReflex wasn't the
best deal. We went with a Selspot system working off a 486 machine. We
recently went to the new intel pc chip (pant...). Ron Tal, of our lab,
investigated connecting the selspot system to the Mac and realized it was
a very large scale effort, but doable. We stayed with the pc system,
though. You can purchase a A/D board to go with the selspot, but it cost more
than an NI board.

To do 3-d analysis, you need a minimum of two cameras. The requirement
for more camera usually is dependent on the range of motion and the
marker placement you require. For example Bob Mann's lab at MIT only
used 2 camera system to analyze gait, however, people at SF shiner's
hospital use 6 for their gait experiments.

To get more information about both MacReflex and Selspot systems you can
contact Manfred and Victoria Berger of Innovision systems in Michigan
(they've probably already sent you e-mail - they subscribe to
biomech-l). We have worked with them over the past two years and find
them very helpful. Their e-mail address are
manfred@innovision.win.net
vicki@innovision.win.net

I don't have the name and andress of the person at LNL with me. I'll
e-mail you tomorrow with it.

If you are interested in coming to our lab to talk about the selspot
system we have please let me know. We're about a 30 minute drive from
SFGH. We are using a six camera system to measure wrist position.
We use a 2 camera system to measure finger motion during tapping.

Good luck.
Jack
510 231-9405.
////
Jack Tigh Dennerlein ~0-0~
http://mote.berkeley.edu/~jax/jax.html \- /


__________________________________________________ ________________________
Hi!

The number of cameras that you need will depend on the pathologies
you are investigating as well as the type of movement.

In our case we wanted to do gait analysis of Spinal cord injured
patients using FES-assisted walking. In turned out that for a
bilateral analysis we needed at least 6 cameras because many markers
were blocked from the different views as a result of all of the
rotations of the segments.

Good luck.
Michel Ladouceur, PhD(c)
McGill University
School of P.& O.T.
3630 Drummond, Room 105
Montreal (Quebec)
H3G 1Y5

Tel: 514.398.4519
Fax: 514.398.8193
e-mail: gsml@physocc.lan.mcgill.ca

..it is true that my tactic is to make sweeping categorical
statements. Whether or not this is a fault ... is debatable. My own
feeling is that it leads more quickly to the solution of scientific
problems than a cautious sitting on the fence.

E. Mayr, _The Growth of Biological Thought_, Harvard University
Press, 1982, p. 9


__________________________________________________ ________________________

Hi, Dr. Robinovitch,
I don't really know the answer to your first question. About your second
question, theoretically speaking, you need at least two cameras to acquisit 3d
kinematics. The purpose to use more than two cameras in 3D kinematics
acquisition is to improve the quality of the kinematic data (minimize the
need for interpolation and extrapolation and the constraints to the data
aquisition procedure). The exact number of cameras needed really depends on
the motion to be analyzed. Generally speaking, six cameras should be
sufficient for most of the human body motions analyzed in an orthopedic
biomechanics laboratory.

Bing Yu, Ph.D.
Orthopedic Bimechanics Laboratory
Mayo Clinic
Rochester, MN 55905


__________________________________________________ ________________________

Steve Re: the above query.

We have a Lab set up which has Peak for low speed, and Motion Analysis Inc
for high speed and auto analysis of video. About two years ago we wanted a
qualitative anaysis system similar to the Peak Video Illustrator and decided
to develop a Mac based system. This uses the Radius VidoeVision capture card
and software in the Hypercard environment (The PowerMac 8100 series is the
recommended Mac configuration). A version for qualitative analysis (VQAS),
with limited quantitative capability, has been marketed for about the last 6
months in NZ.

A quantitative version of VQAS is under development to do "full" kinematic
and inverse dynamic analyses. In the first instance this is most likely to
be for 2D planar motion!

I assume from your system description (ie use of the National Instruments
A/D board), that you also use National Instruments Labview3 as an aquisition
and software analysis package. We started with Labview2 on the Mac and are
now upgrading to Labview3 (.1? runs "native") on the PowerPC. Teaching labs
have 6100 series Powermacs. If you are not using Labview I recommend taking
a look time at its application.

Email if you need further info.

Regards

Barry Wilson

__________________________________________________ ________________________

The Mac reflex system is, like nearly all Mac software, easy to drive. Its
results look good. However, it is a direct recording system. The image on
screen is only the marker points. If there are obscured points, it may be
impossible to decide where the points should be. Also, if there is a
problem in recording, the MOTION must be performed again. This may not be
possible.

A Video based system eg PEAK allows re-digitising of obscured points, using
manual estimation from the video image, allowing judgemment to be made on
probable position. Reflections from the background can be also allowed
for.
However, this is not a real time system and the digitising process is very
tedious and takes a long time, especially if there are problems with the
time coding or other video hardware.

>
> 2) how many cameras are required for acquisition of 3-D kinematics? I have
> heard responses which vary from four cameras to more than six.

This depends on the motion and how much obscuring of marker points takes
place. Minimum of 2 cameras BUT this assumes all points are visible to both
cameras. Using more cameras to show points that go missing on other
cameras is possible BUT ensure that the software will allow the collection
of dis-continuous data. The amount of digitising time required, especially
when lots of manual estimation of missing points occurs, can be very large.

You may also like to read my posting of "Practical PEAK suggestions", of 24
October 1994.
I can send this if you can't find it.

John Yelland

__________________________________________________ ________________________
Steve,

I am currently using a MacReflex 2D system with one camera in my
biomechanics/ergonomics laboratory. It seems to work quite well for my 2D
needs. I am currently seeking funding to purchase a 3D system and am
shooting for a total of four cameras for the system, so I share your second
question - how many cameras are required?? I suspect part of the answer
has to do with the type of activities people will be doing, e.g., how fast
will they be moving, potential for well hidden markers, size of person's
excursion area, etc. Have you talked to Bill Hand, (707) 579-2596?? I'll
be interested to hear what you find out since I share some of your
questions. If you have any specific questions I can answer about the 2D
system, let me know.

Carter Kerk


Carter J. Kerk, PhD, PE, CSP 236E Zachry Engineering Center
Assistant Professor (409) 862-4149 Office
Industrial Engineering Department (409) 847-9005 Fax
Texas A&M University kerk@zeus.tamu.edu
College Station, TX 77843-3131 USA

__________________________________________________ _______________________

Hi Steve,

I'm marginally familiar with the MacReflex
system. My understanding is that it is average performance and priced right. I
believe you will have more flexibility with a PC platform so you may wish to
explore that option as well particularly since the cost of the computer itself
is down in the noise when you consider cost of the whole system.

Regards,

Will

__________________________________________________ _______________________


Hi from Vienna-Austria

We are working in a gaitlab using a videometry sytem with six cameras. The
number of cameras strongly depends on your application. Theoretically it is
sufficient to do 3-D Analysis with only two cameras. This is a very unstable
situation as you loos markers whenever one is covered by a bodysegment
or merging with another marker in the field of view of one of this cameras.
To have a robust system for analysis of human movement you need three
cameras for each side. For full body analysis this sums up to six cameras.

For more information see also:

Kollmitzer J., Errors and practical tests for 3-D gait analysis system
setups with videometry and forceplates, Proc. Third International Symposium
on 3-D Analysis of Human Movement,1994 Stockholm

Kindest Regards

Josef Kollmitzer
Univ.Clinic PMR
Waehringer Guertel 18-20
A-1090 Vienna Austria

__________________________________________________ ______________________

My colleague in the dept. of Psychology here have a Mac Reflec. You can
contact Jan Piek on PIEK@psychology.curtin.edu.au

I have just recently bought a Litek Advanced Systems V-scope motion
monitor, which works on ultrasound. It promises to be very useful (and very
cheap at US$5000). At present it only works with the IBM PC, but they're
developing a Mac version, fax Joseph Eliav Tel Aviv Israel 3-497719

You need a minimum of 2 cameras for 3D kinematics, but often need more
because the markers are not seen by one of them when the subject moves.
Typically, we use 3 for gait - 4 for sporting actions. It is possible to do
limited 3D analysis with ONE camera - a paper was published a few years ago
on this.

Best wishes - I'm afraid movement is not the easiest thing to measure!

Chris

__________________________________________________ __________________
Dr. Chris Kirtley MB ChB, PhD c.kirtley@info.curtin.edu.au
^
Lecturer, Bio-engineering --_ / \
/ \
School of Physiotherapy, Perth #_.---._/
Curtin University of Technology, V
GPO Box U1987,
Perth 6001, Tel +61 9 351 3649
Western Australia. Fax +61 9 351 3636
__________________________________________________ __________________

Steve: Regarding your posting, I don't know about question 1 (although I
can bet it will be substantially more expensive). Regarding #2, you
theoretically only need two cameras. This is what we use with a
commercial Watsmart system using ired's. However, this assumes you will
always be able to see the markers.


Dan Gottlieb
Dep't of Ortho
The University of Vermont