View Full Version : Summary: Roentgen Stereophotogrammetric Analysis

Mark Watts
08-26-2002, 09:24 AM
This is the summary of replies to the question about RSA techniques that I posted on Friday 9th August 2002. The
question is first and the replied follow. Thanks to all those to who replied, I hope I can repay the favour soon.

Original Question:

Hi all,

I work with a group of Orthopaedic Surgeons who would like to investigate whether the mobile bearing of a total
knee replacement moves in vivo during dynamic activities. It is envisaged that the patients would be performing
flexion and extension of the knee while weight bearing.

We have looked through the available literature but are still unsure of what technique we should be using to collect
the data that we require. We are considering the use of Roentgen Stereophotogrammetric Analysis (RSA). Can RSA
be used to investigate dynamic activities such as knee flexion and extension during weightbearing? If not is there a
better technique eg video fluroscopy?

Any replies will be collated and posted for your viewing pleasure.

Thank you in advance.


Mike Murphy

I would suggest that you contact Scott Banks (see attached abstract from World Congress of Biomechanics).

3D Knee Kinematics from Pictures: How’s it work, Why’s it useful, and Where’s it going?
The Biomotion Foundation, 1411 N. Flagler Drive, Suite 9800, West Palm Beach, FL 33401, { HYPERLINK mailto:banks@alum.mit.edu }banks@alum.mit.edu

Mark Gillies

A guy who may be able to help is Scott Banks, a US surgeon, who has developed a method to monitor implant
movement using plane x-rays.

He should have some ideas to help out.

His email:- "Scott A. Banks, PhD"

Antony Hodgson

RSA is certainly a reasonable choice for the study you propose. To achieve good accuracy, you need to obtain
images from a calibration phantom when your cameras are in position. It also requires that you be able to obtain
simultaneous X-ray images from two or more cameras and that the exposure times are short relative to the
characteristic times of the motions you're investigating. Depending on how much movement you're trying to detect
and how accurately you need to be able to define the 3D position of the prosthesis during movement, video
fluoroscopy may also be acceptable.

Niclas Borlin

RSA has been used to study knee implant motion in vivo. See e.g.Karrholm et al. "Kinematics of successful knee
prosthesis during weight-bearing" Knee Surgery Sports Traumatology Arthoscopy 1994;2:50-9 or Nilsson et al.
"Active knee motion in total knee arthroplasty. A roentgen stereophotogrammetric analysis of the kinematics of the
Tricon-M knee prosthesis." Clin Orthop 1990;256:147-161. These studies were performed in "pseudo-dynamic" mode,
but there are some studies coming out of prof. Karrholm's lab in Gothenburg where he has an RSA film exchanger

I know the guys involved but have little first-hand knowledge of the studies, so I suggest you contact either of the
authors directly - Kjell G. Nilsson, {HYPERLINK "mailto:kjell.g.nilsson@orthop.umu.se"}kjell.g.nilsson@orthop.umu.se, or Johan Karrholm, {HYPERLINK "mailto:eva.fermen@orthop.gu.se"}eva.fermen@orthop.gu.se (his
secretary's email). If you want a contact closer to home (:-) you could try mailing prof. Bo Nivbrant in Perth at
{HYPERLINK "mailto:Nivbrant@cyllene.uwa.edu.au"}Nivbrant@cyllene.uwa.edu.au (beware though - his primarily a hip guy :-). All three have loads of RSA experience
and could probably tell you what can be done and what cannot.

You could also check out the extensive RSA reference list at {HYPERLINK "http://www.umrsa.com"}www.umrsa.com (click on method, then references).

Michelle L. Zec
{HYPERLINK "mailto:mlzec@ucalgary.ca"}mlzec@ucalgary.ca

I have attached two abstracts from the World Congress of Biomechanics held here last week. This is not my area of
expertise, but Anderst and Tashman are doing some really impressive stuff with in vivo dynamic measurements of
the knee utilising RSA. The authors on the other abstract (Thistlethwaite, Ronsky and Gill) are involved in a study
utilising RSA to evaluate different fixation techniques in TKA.

A Fast Circle Finding Technique for Digital RSA.
P. Thistlethwaite1, J.L. Ronsky1, H.S. Gill2
1Department of Mechanical and Manufacturing Engineering, University of Calgary
2University of Oxford, Nuffield Department of Orthopaedic Surgery/OOEC

A Unique Method to Establish Dynamic In Vivo Articular Surface Interaction
William J. Anderst and Scott Tashman
Motion Analysis Lab, Bone and Joint Center, Henry Ford Hospital, Detroit, MI
Email: { HYPERLINK mailto:anderst@bjc.hfh.edu }anderst@bjc.hfh.edu

Eberhard Hierholzer

I read your request on RSA of total knee implants. A number of years ago, I worked on loosening of hip prostheses
(Judet) using RSA (only tentative studies, no publications). This technique seems to be the only suitable one for this
type of examination, however it its difficult.

Several preconditions must be met:
Characteristiclandmarks on the prostheses must exist, which can be detected accurately from different
projections. At least three landmarks on each moving part must be visible in each record.
If movements relative to the bones must be measured, then artificial landmarks such as little balls or pieces of
wire must be implanted in each relevant bone (again at least 3 landmarks - visible from any direction – are
Generallyit might be desirable to classify movements with respect to "anatomical coordinates", e.g. longitudinal,
sagittal, lateral etc. This is particularly difficult due to the absence of clear bony landmarks in most cases.
Becausemovements may be small high measurement accuracy is essential which is difficult to achieve under the
above conditions (accurate SRA calibration necessary).

This is not to deter you from your project, but to mention some pitfalls (there may be some more). You probably
know the publications of Selvik who did a lot of work on RSA in the past. Otherwise I would encourage you to study
his papers and other literature cited there.

Jan Stolk
{HYPERLINK "mailto:jan@orth0044.azn.nl"}jan@orth0044.azn.nl

In our lab we have a lot of experience with RSA measurements. RSA is very well suited to measure implant
migration during loading. It has a high precision. But unfortunately you can only use it for static, or quasi-static
measurements. The bodies have to come to a stand still during the exposure-time while shooting the radiographs. So I
don't think RSA is the way to go for you.

We are currently working on a fluoroscopic system. But take a look at work from Banks (West Palm Beach, FL,
USA) or Komistek/Dennis (Denver, CO, USA). They have used fluoroscopic imaging-techniques to monitor implant

An other system for monitoring implant motion used ultrasound. Take a look at the paper I attached to this message.

Stukenborg-Colsman, C., Ostermeier, S., Wenger, KH. & Wirth, CJ.(2002). Relative motion of a mobile bearing inlay
after total knee arthroplasty – dynamic in vitro study. Clinicial Biomechanics. 17, 49-55.

Edward Valstar

Although RSA is a very useful tool for assessing micromotion of prostheses with respect to the surrounding bone, it
might be rather difficult to use it for studying the movement of a mobile bearing during a flexion extension motion of
the knee. Because the thickness of the bearing is rather small, and thus the distance between the tibial tray and the
femoral component is also small, it is very difficult to visualise the tantalum markers in the bearing: these markers are
easily overprojected by the metal of these two components. Another thing is that the markers will have to be
projected in two simultaneously taken X-rays - with the roentgen tubes directed at 20-degrees relative to these X-
rays. If a marker cannot be seen in one of these two X-rays, it cannot be used for analysis.

We have a rather extensive experience with RSA, but after some phantom tests with a mobile bearing knee, we have
decided to switch to fluoroscopy for studying mobile bearing knee kinematics. We first take a set of RSA X-rays to
assess the three-dimensional distribution of the markers in the bearing, and to create a three-dimensional marker-
model. Thereafter, fluoroscopic images of the knee are taken and the three-dimensional marker model is used to
assess the three-dimensional position and orientation of the bearing markers. A calibration object is used to
accurately assess the position of the X-ray source. Matching of a CAD model of the tibial component is used to
assess the position and orientation of this component, and subsequently the motion of the bearing relative to the tibial
component can be assessed. Because of the single X-ray source that is used it is much simpler to visualise the
markers than would have been possible with RSA

However, there are some problems related to fluoroscopy too. First, the rather poor spatial resolution and rather poor
quality of the images. There is also distortion of the image and blurring of the markers and contours when performing
motion. An alternative might be to perform a quasi-static motion of the knee and use conventional X-rays in stead of
fluoroscopy. The drawback of that approach, however, is that quasi-static motion differs from real kinematics of the

Well, this is a very short reply to your question. I hope that your research will be successful. If you have any more
questions, please don't hesitate to contact us, and if you happen to be in Europe - and I don't expect that that will
happen very often - you are more than welcome to visit our group.

Eric Charri
{HYPERLINK "mailto:Eric.Charriere@emea.strykercorp.com"}Eric.Charriere@emea.strykercorp.com

You should have a look at a paper presented at the World Congress of Biomechanics 2002 last week. I have
attached the abstract to this e-mail. They have used fluoroscopic method associated with 3D algorithms. And they
have found that the mobile bearing moves (but surely that will depend on the design). You should contact Dr.
Komistek for further information.
(See attached file: Mobile bearing.pdf)

In Vivio Determination of Polyethylene Bearing Motion Relative to the Tibia and the Femur.
Richard D. Kromistek1, Brian D. Haas1, Douglas Kilgus2, Adam Smith2, Scott A. Walker1
1Rocky Mountain Musculoskeletal Research Laboratory, Denver, Colorado, { HYPERLINK mailto:rkromistek@rmmrl.org }rkromistek@rmmrl.org
2Wake Forest Baptist Medical Center, Wilston-Salem, North Carolina

Mark Watts
BScAppHMS(Ex-Sc) Hons
Master of Philosophy Candidate
School of Human Movement Studies
The University of Queensland
St Lucia, QLD 4072

Tel: +61 7 3365 6313
Fax: +61 7 3365 6877
Email: mwatts@hms.uq.edu.au

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