вход по аккаунту



код для вставкиСкачать
Computer Aided Surgery
4:335–351 (1999)
Abstracts from the Third Annual North American Program
on Computer Assisted Orthopaedic Surgery(CAOS/USA ‘99), held at
UPMC Shadyside, Pittsburgh, Pennsylvania, USA on June 17-19, 1999
The role of the CAOS program is to educate practicing orthopaedic surgeons on the rapidly
evolving field of computer assisted orthopaedic surgery. The course presents the most current
thinking on the impact of computer assisted surgical techniques on the clinical and surgical
routine in several orthopaedic subspecialties. Within the last few years, UPMC Shadyside has
hosted surgeons, researchers, and corporate representatives from all around the world.
This year’s conference continued to emphasize the innovative use of computer-based
technologies in areas such as the integration of medical imaging and computer vision in the
operating room, robotic assistive devices and intraoperative navigational systems, surgical
simulations and planning, and virtual reality in surgery.
The realization of these advancements in medicine will require continued collaboration
between physicians and the research community. We need to bring together and educate
clinicians in order to define the current status of these technologies and explore future directions
and clinical applications. Therefore, the goals of CAOS are to:
• Educate practicing orthopaedic surgeons on new technologies which have the potential to
impact clinical practice;
• Identify new tools and emerging technologies which will enhance patient care;
• Foster interdisciplinary clinical research using technologies not traditionally used by
orthopaedists, namely: robotics, computer vision, virtual and hybrid reality, and intraoperative image guided surgery.
We have been encouraged to see so much interest in these new and emerging technologies
which will undoubtedly influence the way surgeons plan, simulate, and execute surgery now
and in the future.
Anthony M. DiGioia, M.D.
Course Chairman
Abstracts from CAOS/USA ’99
Richard M. Satava, MD FACS
Yale University School of Medicine, New Haven, CT
The revolution in surgery has not abated. Earlier successes in arthroscopy, laparoscopy
and other minimally invasive techniques are actually transition technologies which will
spawn the next generation of advances
In order to keep pace with the transition, it is essential to understand the fundamental
concept involved in thinking in Information Age terms, rather than the traditional
Industrial Age mindset. This enables entirely new capabilities in the whole new generation
of medicine, which also includes the infrastructure of telemedicine, medical informatics,
3-D visualization and point of service healthcare with ubiquitous realtime access through
telecommunications. These applications are mediated through the computer and information networks and as such are the essence of the paradigm shift in the field of medicine.
On the horizon are attempts to incorporate biologically derived methodologies, such as
tissue engineering, biomimetic materials and nanotechnology.
There are a number of enabling technologies which will usher in and support the next
generation of medical interventions. These include real-time image acquisition, point of
care data acquisition, 3-D visualization, computer enhancement (through digital signal
processing, scaling, filtering, etc.), remote manipulation and telepresence and distributed
networking. In order to move beyond minimally invasive procedures and completely
benefit from the computer aided revolution, we must take the broadest interpretation and
integrate the full spectrum of activities. However, we must balance the technology with
human compassion and empathy and with these technologies we can provide an enhanced
quality of medical care for each and every patient.
Newer technologies of robotics, telepresence surgery, remote manipulation and dexterity enhanced surgery may also be on the pathway to wherever the future is taking us.
Just as important, the other components of surgery, such as pre-operative planning,
interoperative navigation, and surgical education and training, must evolve along with the
technical components of the operative procedures.
Anthony M. DiGioia, III, MD1.2, Branislav Jaramaz, PhD1,2 and Bruce D. Colgan, MS1
Center for Orthopaedic Research, Shadyside Hospital, Pittsburgh, PA.
Center for Medical Robotics and Computer Assisted Surgery, Carnegie Mellon University, Pittsburgh, PA.
Objectives: Image guided and surgical navigation systems, robotic assistive devices and
surgical simulators have begun to emerge from the laboratory and hold the potential to
improve current surgical practice and patients’ outcomes. The goals of these new clinically focused technologies are to develop interactive, patient-specific preoperative planners to optimize the performance of surgery and the postoperative biologic response, and
develop more precise and less invasive interactive smart tools and sensors to assist in the
accurate and precise performance of surgery.
Background: Recent advances in the fields of medical imaging, computer vision, and
robotics have provided the enabling technologies to permit computer aided surgery to
become an established area which can address clinical needs. Although these technologies
have been applied in industry for more than 20 years, the field of computer assisted
orthopaedic surgery is still in its infancy.
Results: Technologies are emerging that will influence the way in which orthopaedic
surgery is planned, simulated, and performed. Image guided and surgical navigation
systems, robotic assistive devices, and surgical simulators have begun to emerge from the
laboratory and hold the potential to improve current surgical practice and patients’
Conclusions: The medical community is beginning to see the benefit of these enabling
technologies which can be realized only through the collaboration and combined expertise
of engineers, roboticists, computer scientists, and surgeons.
Russell H. Taylor, PhD
NSF Engineering Research Center for Computer-Integrated Surgical Systems and Technology, Johns Hopkins University, Baltimore, MD.
Robotic systems are beginning to have an increasing role in orthopaedic surgery. In many
cases, the key advantages offered are speed and precision. Systems such as Robodoc® and
Caspar®, for example, are intended to permit very accurate preparation of the femur in
cementless THR procedures. These systems may be thought of as examples of surgical
CAD/CAM. In other cases, the robot works cooperatively with the surgeon, and may be
thought of as a very simple surgical assistant.
This talk will introduce basic concepts of surgical robotics and illustrate them with
examples drawn from current or emergent orthopaedic systems.
Abstracts from CAOS/USA ’99
M. Sati, R. Hofstetter, M.A. Slomczykowski, H. Wälti, L.-P. Nolte
M.E. Müller Institute for Biomechanics, University of Bern, Switzerland
André Bauer, MD
Marbella High Care, Marbella, Spain
Introduction/Objectives: Intra-operative fluoroscopy is a valuable tool for visualizing
underlying bone and surgical tool positions in orthopedics. Disadvantages of this technology include the need for continued radiation exposure for visual control and cumbersome alignment to obtain a desired view. The purpose of this paper is to highlight a new
concept of a computer assisted freehand navigation system that uses single intra-operative
acquired fluoroscopic images as a basis for real-time navigation of surgical tools.
Background: Robot-assisted orthopedic surgery is a relatively young technique. The
first generation of robots was in hospitals at sites in the US and Germany until 1996, when
the second generation took over. Now the third generation is installed in European sites.
Although the technology itself is quite mature and constantly progressing, the manmachine interaction and, especially, ergonomic aspects need further enhancement.
Critical zones:
Design/Methods: Optoelectronic markers are placed on surgical tools, a patient reference, and the fluoroscope to track their position in space. Projection properties of the
fluoroscope are acquired through an initial precalibration procedure using a tracked
radiopaque phantom grid. Corrections are applied to compensate for both the fluoroscope’s image intensifier distortions and mechanical bending of the C-arm frame. This
enables real-time simulation of surgical tool positions simultaneously in several single
shot fluoroscopic images. In addition, through opto-electronically tracked digitization of
a target view point, the fluoroscope can be numerically aligned at exact angles relative to
the patient without any X-ray exposure. The system also allows reconstruction of threedimensional (3D) landmarks from bi-planar images of a given anatomy. Landmarks such
as the long axis of the proximal and distal fragment of a femoral fracture, the femoral neck
axis and the posterior-condyle axis can be used to provide the surgeon with real-time
feedback on fragment alignment, leg length and anteversion (AV) correction during the
reduction procedure. Other applications include insertion of dynamic hip screws (DHS)
and spine stabilization procedures such as pedicle screw insertion with potential for
development of new minimal invasive procedures.
Results: A specially designed optoelectronic accuracy checking device was used to demonstrate that a navigation accuracy of below 1 mm was achievable over an entire range of
C-arm orientations used in the OR. It has also been shown that anteversion can be measured
with a maximal error of ⫾5 degrees with user definition of landmarks being the largest source
of error. Cadaver trials in the fracture fixation area have shown the system’s successful use for
the difficult task of distal locking of unreamed femoral nails (UFN) and unreamed tibial nails
(UTN), and femoral fracture reduction. First clinical trials in the OR setting have shown
encouraging results of the system’s use for locking procedures. Preliminary results for the
entire femoral fracture reduction procedure revealed that the system provided the surgeon with
valuable information, however the large increase in surgical time must be reduced before the
technology can be used in routine practice for this last procedure.
Discussion and Conclusions: The presented method is conceptually based on X-ray
fluoroscopic images, which can easily be obtained using standard C-arm equipment
available in the majority of hospitals. Since surgeons are familiar with these kinds of
images, the interpretation of anatomical features is straightforward. The navigation system
provides the link between fluoroscopic imaging and surgical action. With the use of
registered images no further image updates are necessary during the intervention. Based
on the results of the present study, the following advantages of the novel technique can be
identified: (a) Improved accuracy and safety for both tool navigation and 3D feedback on
fragment positioning; (b) Significant reduction of radiation exposure; (c) Added value to
existing hospital equipment; (d) Potential for future minimally invasive approaches such
as in the spine.
Acknowledgments: This study was supported in part by the Swiss National Science
Foundation and the AO/ASIF Foundation, Davos, CH.
Planning: This is the greatest achievement of computer-assisted orthopedic surgery. In
robot-assisted surgery the surgeon’s plan will be transformed with the utmost accuracy,
but the robot will also reproduce any error in the plan, and there is no counter-check.
Problems can occur due to inadequate implant selection, positioning, or selection of
anteversion. For many criteria the orthopedic community still has no definite answers,
e.g.: Shall a cementless implant have close contact to cortical bone or should there be a
rim of cancellous bone? Should an extreme anteversion encountered during planning be
corrected or left as it is? Should leg length always be equalized or are other criteria more
important? With no consensus on such issues, it is pointless creating software to help the
surgeon during planning, yet the computer asks the surgeon to create a definite plan with
an accuracy of 0.5 mm and 0.5°!
Registration: This is the most cumbersome and lengthy process in robot-assisted surgery. Initially, 3 pins had to be implanted in a separate operation before the CT scan. This
drew much criticism, but the process of finding pins during surgery or after bone motion
was a greater nuisance. Also, 2 different probes were necessary, requiring difficult
tool-changes. The 2-pin method was an improvement but still cumbersome. Pinless
DiGiMatch technology solved the pin problems and facilitates recovery after bone motion.
Continuous registration will eventually make the bone motion monitor superfluous. No
advance in this field was as dramatic and important as the pinless surgery.
Fixation and access: Both fixation and access to the bone require careful soft tissue
management. Soft tissue damage can compromise otherwise excellent results. Leg fixation
initially required fixing a large frame to the bone, which consumed valuable OR time. A
solution with only 1 proximal clamp and tape fixation of the lower leg serves the purpose,
but continuous registration will allow for a less rigid fixation. Access problems may arise
due to the obesity of the patient, and the approach to the bone in the preparation of straight
stems. These require a straight, often very lateral approach. In hand-broached surgery, the
surgeon often deviates from this path to protect the greater trochanter and gluteus. The
robot does not accept such a compromise, and this can lead to severe damage in patients
with a small or medially bowed greater trochanter, eventually causing permanent instability of the joint. Surgeons urged the manufacturer to integrate anatomic implants into the
implant library and develop cutting files that allow for oblique access for the reamer. The
new cutting paths of the ABG and Antega implants preserve the said structures and require
less effort by the surgeon. This example shows how crucial feedback from the surgeons
to the engineers is.
Ergonomic Aspects: In robotic surgery, the line of sight of the surgeon is constantly
changing between the operational field, the remote control, and the screen. In certain steps,
i.e., registration or when force-freezes occur during reaming, the surgeon has to interact
closely with the computer, which requires 15 or 20 commands in a very short period.
Current solutions for this are unsatisfactory, lead to uneasiness for the surgeon, and
prolong the procedure. Ergonomic aspects are not a primary concern during development
of a robotic system, but deficits in this area cause longer OR times, which add to the
well-known discrepancy in OR time versus hand-broached procedures.
Conclusions: From the first operations in 1992 up to today, robot-assisted surgery has
undergone dramatic changes. The system is now more reliable, with many small improvements allowing a faster, smoother procedure. Yet deficits remain to be addressed,
specifically in the ergonomic field.
Abstracts from CAOS/USA ’99
S. David Stulberg, MD1, Frederic Picard, MD2
Northwestern University Faculty Foundation, Chicago, IL, USA.
Hopital Sud University Hospital Grenoble, France.
Objectives: The function and longevity of a total knee arthroplasty are related to the
accuracy with which it is implanted. Technologies utilizing computer integrated systems
offer opportunities for performing reproducible and very accurate total knee replacements.
The purpose of this study was to develop and clinically evaluate a computer assisted total
knee system that makes possible the efficient and precise performance of total knee
Design/Methods: A computer assisted system (Orthopilot) was developed to guide the
extramedullary placement of standard total knee cutting jigs. No special preoperative
x-rays are required for the use of the system. The position of the extremity and cutting jigs
is monitored intraoperatively by means of an optical tracker. A prospective study was
performed to compare the accuracy of the system (group 1) with a standard TKR (group
2). 30 patients with end-stage osteoarthritis were randomly assigned to one of the groups.
The clinical and radiographic results were evaluated 6 weeks and 3 months following
surgery by two experienced surgeons unaware of the type of procedure performed.
F. Picard, MD1,3, D. Saragaglia, MD1,2, D. Stulberg, MD3, F. Leitner, PhD4,
O. Raoult, PhD4.
Hopital Sud University Hospital Grenoble, France; 2Hopital CHR Chambery, France;
Northwestern University, Chicago, IL, USA; 4Praxim Company, France.
Background: The technical difficulty in the Total Knee Instrumentation procedure is to
achieve cuts perpendicularly to the mechanical axes of the femur and the tibia. The
longevity of total knee arthroplasty is closely related to its intra-operative positioning.
This procedure offers an effective and novel positioning method depending on a computer
assisted orthopedic surgery system called ORTHOPILOT. We tested this system on a
series of 30 patients with primary osteoarthritis of the knee (15 patients were operated with
a classical procedure and 15 patients with the ORTHOPILOT procedure). The trial was a
prospective randomized parallel study carried out in Grenoble from January 13 to
December 1, 1998.
Design/Methods: From thirty patients between 55 and 89 years of age (mean 69),
fourteen right knees and sixteen left knees were included in the study after checking
eligibility (inclusion and exclusion criteria, signing the consent form). The same surgeon
operated on all the patients, using either the classical procedure (SEARCH AESCULAP
Knee) or the computer-assisted system. Two independent surgeons followed up the
patients six weeks after surgery. The review criteria included: radiological criteria (the
main criterion was the femoro-tibial mechanical angle on the long leg X-ray), complications, and surgical criteria (such as duration of the procedure, post-operative bleeding).
Results: Radiological outcomes (measurements have been done on long leg X-ray,
coronal, and sagittal views):
Results: Radiographs were evaluated to determine what percentage of patients had: 1)
femoral-tibial angles between 3 degrees of varus and 3 degrees of valgus (100% group 1,
66.6% group 2); a femoral implant angle of 90 degrees to the coronal mechanical axis
(46.6% group 1, 0.06% group 2); and 3) a tibial resection angle within 2 degrees of varus
or valgus (100% group 1, 86.6% group 2). The average duration of the procedures was 101
minutes in group 1 and 74 minutes in group 2. Blood loss in the two groups was the same.
There were no complications in group 1 and 3 complications in group 2 (2 dvt, 1 stiff
Femoro-tibial angle: from 3 degrees of varus to 3 degrees of valgus: Usual technique (UT)
10 patients (66.6%), computer technique (CT) 15 patients (100%). Mechanical femoral
angle: in neutral position (90 dg): UT 1 patient (0.06%), CT 7 patients (46.6%).
Conclusions: A computer assisted total knee replacement system has been developed
which is safe and which allows accurate, reproducible positioning of conventional TKR
cutting guides.
Clinical outcomes: Duration of surgical procedure: UT average 74 min (55-100), CT
average 101.6 min (80-130). Duration of tourniquet: UT average 73.3-min (60-90), CT
average 92.3 min (75-120). Post-operative bleeding: UT average 414.6 ml (100-890), CT
average 490 (100-790). Four patients in the UT group had post-operative bleeding
exceeding 790 ml, the highest level in the CT group. Complications: UT 3 cases with
complications (2 dvt and stiffness), CT no complications.
Mechanical tibial angle: in neutral position (90 dg), UT 4 patients (26.6 %), and CT 8
patients (53.3%). Anterior space (anterior cortical/prosthesis): prostheses inside the anterior cortical femur, UT 6 patients (40%), CT 1 patient (0.06%). Tibial slope: from 2
degrees in varus to 2 degrees in valgus: (UT) 13 patients (86.6%), CT 15 patients (100%).
Conclusions: These results have satisfied our requirements for accuracy, reproducibility, and safety.
Abstracts from CAOS/USA ’99
B. Davies1, S. Harris1, M. Jakopec1, and J. Cobb2
Dept. of Mechanical Engineering, Imperial College, London, England
Dept. of Orthopaedic Surgery, The Middlesex Hospital, London, England
Objective: To provide an accurate method of implanting TKR prostheses, with a robotic
system replacing the jigs and fixtures used for manual surgery.
Background: TKR requires precise bone cutting, otherwise pain, poor gait, and early
prosthetic failure may result. In manual surgery, jigs and fixtures guide the saw, but errors
may occur in their application, and a flexing saw-blade may produce curved surfaces. In
place of jigs, a robotic system has been developed that stores the geometry and uses a
rotary cutter to remove the correct amount of bone for a good prosthesis fit. An active
constraint system prevents the surgeon from moving the cutter beyond the planned surface
or from damaging soft tissue.
Preoperative planning: All planning was done preoperatively using a Pentium PC. CT
images of the patient’s leg are segmented and the bones are labeled and tracked by the
computer. To determine alignment, the surgeon marks the axes and required angles on
x-ray images derived from low-resolution CT of the whole leg. A high-resolution data set
is used around the knee region. The computer overlays CAD prosthesis models on the CT
images, allowing them to be finely matched to the bone geometry in both 2D and 3D.
When the prosthesis location is approved, the robot receives information on the position
of each cutting plane. Flat planes are currently used, but the projection algorithm can be
extended to curved geometries. Titanium fiducial marker screws in the CT scans provide
a simple registration system for phantom and cadaver trials, but it is anticipated that
anatomical feature matching will be used to provide a less invasive procedure.
The Operative System: This comprises an active constraint robot (Acrobot) and a
control computer. A gross positioning system is used to locate the robot in the correct
region then locked off. The Acrobot has a force-controlled handle near the cutter, allowing
the surgeon to guide the robot and feel the cutting forces required. The robot constrains
the surgeon to safe regions by opposing attempts to move outside the bone or cut too
deeply. It removes frictional forces and gravity, and provides servo assistance while he is
in a safe region. The surgeon can feel the reaction forces when cutting the bone. Thus the
system uses the robot’s accuracy and the surgeon’s knowledge simultaneously. The
Acrobot has 4 axes of motion, 3 of which (pitch, yaw, and cutter-extension) are under
active constraint control. The fourth is under position control and aligns the Acrobot
optimally for each cutting plane. Constraint and assistance forces are computed in
real-world coordinates and translated to motor axis coordinates at the last stage. Registration of the robot with CT images is achieved by inserting a ball-ended probe into the
fiducial screw heads, recording the position of the robot for each. A least-squares
error-minimization algorithm provides a transformation matrix that translates the plane
positions and orientations and constraint outlines from the CT-based coordinates of the
planner to the real world coordinates of the patient’s leg.
Results: The robot has been used to cut plastic phantoms and cadaveric bones. Initial
tests on phantoms indicated overcutting of the bone resulting in the anterior and posterior
cuts being too close together. This was found to be due to cantilever bending of the robot
and distortion to the gross positioner as large forces were applied to the handle. Hardware
and software enhancements have improved this to provide a very good fit. Tests on
cadaveric legs have shown a close fit between prosthesis and bone with a negligible gap
between them, good orientation between the tibial and femoral components, and a close
match between the pre-operative planning and operative procedure.
Conclusions: Acrobot can accurately provide cuts for TKR surgery, allowing a knee
prosthesis to be implanted in both plastic phantoms and cadaveric bone. The rotary cutter
creates flat planes for the current generation of TKR prostheses, and the robot can be
extended to non-planar geometries and uni-compartmental components. The registration
procedures worked well, but less invasive anatomical registration will be developed for
future clinical studies.
Tiburtius V.S. Klos, MD 1,2 Raymond J.E. Habets, MS1, Anne Z. Banks, MS2, Scott A.
Banks, PhD2, Roger J.J. Devilee, MD1, Frank F. Cook, MD2
Department of Orthopaedics & Traumatology, Catharina Hospital,
Eindhoven, The Netherlands; 2Orthopaedic Research Laboratory, Good Samaritan Medical Center, West Palm Beach, Florida, USA
Abstract: Accurate placement of grafts is considered one of the most important factors
in ACL surgery. However, reconstruction with contemporary guiding systems can still
result in unacceptable graft placement variability. In order to improve the reproducibility
of graft placement, intra-operative visual feedback was added to arthroscopic technique.
First, fluoro-scopic visualization was added to evaluate guide wire placement prior to
tunnel drilling. Second, computer graphic overlays were added to the fluoroscopic view.
Three groups of patients were treated: 29 cases arthroscopically, 53 cases with fluoroscopy
added, and 125 cases with computer overlays. Graft placement variability was significantly reduced with fluoroscopic visualization. Computer overlays resulted in further
significant reductions in graft placement variability. Simple visual enhancements appear
to be useful in improving the accuracy of arthroscopic ACL reconstruction.
Table 1: Tibial placement of graft measured using
Staubli’s criteria (Knee Surg Sports Traumatol Arthroscopy
1994; 2:138-146).
Standard Deviation
A ⫽ Arthroscopic (29 cases), AF ⫽ Arthroscopic and Fluoroscopy (53 cases)
AFC ⫽ Computer Assistance (125 cases)
Table 2: Femoral Placement of graft measured using
Harner’s criteria (Arthroscopy 1994:10:502-512).
Standard Deviation
Conclusion/Discussion: Large variability in graft placement has been recorded even for
experienced orthopaedic surgeons. Computer-guided ACL reconstruction based on consistently identifiable radiographic landmarks resulted in less variability of graft location when
compared to arthroscopic or arthroscopic plus fluoroscopic techniques. This was achieved with
inexpensive computer components and without an increase in operating room time. The
amount of radiation exposure from using fluoroscopy during ACL reconstruction is minimal,
since only a single lateral projection is used. The addition of the computer to grab fluoroscopic
images did not increase the exposure time and may actually decrease it. Since identifiable
landmarks are used to measure graft location intraoperatively, future studies will report on the
patient outcome with respect to graft placement. This technique improved the surgical quality
of ACL reconstruction by improving the accuracy of graft placement, providing virtual tunnel
matching and precise verification of graft placement.
Abstracts from CAOS/USA ’99
Tiburtius V.S. Klos , Melinda K. Harman, Raymond J.E. Habets , Roger J.J. Devilee ,
and Scott A. Banks2
Department 2of Orthopaedics & Traumatology, Catharina Hospital, Eindhoven, The
Netherlands; Orthopaedic Research Laboratory, ICHS, West Palm Beach Florida, USA.
Abstract: Graft positioning in anterior cruciate ligament (ACL) reconstruction is usually documented from lateral postoperative radiographs. The purpose of this study was to
compare three methods of measurement for femoral placement in fifty patients with ACL
reconstruction. In order to decrease problems with graft visibility we used intra-operative
images with drill guides in position at the point of final drill placement.
Three methods were used for measuring ACL graft position: 1) the method described by
Harner, using the intercondylar notch line; 2) the method of Aglietti, using the extension
of the intercondylar notch line; and 3) the method using the circle of Amis and relating
placement to its position in the circle. The obtained intra-operative images were digitized
for computer-assisted measurement and divided in two groups. First group showed good
projections, with overlapping condyles and the second group showed average projection,
with less optimal overlapping condyles. A custom computer program was developed to
allow the user to identify anatomic locations on digitized intra-operative radiographs using
the mouse. Four orthopaedic surgeons used this program to measure the ACL graft
position according to the three methods. Subsequently statistical analysis was performed
using repeated measures ANOVA and the Intra-Class Correlation Coefficient (ICC) in
order to compare multiple observations by multiple observers. From this data values for
ICC were calculated and found substantially reliable for the method using Amis circle. In
the use of perfect overlapping radiographs, the method as advocated by Harner is reliable,
but this method is only moderately reliable in less-perfect overlapping radiographs.
Overall findings are displayed in Table 1. Additional difficulties with graft visualization
in post-operative radiographs will further decrease reproducibility of these measurements.
We concluded that consistent measurements documenting graft placement in ACL reconstruction are difficult to obtain. In our study, only one of the measurement techniques,
using a circle as mentioned by Amis, showed substantially reliable data (ICC ⬎ 0.6)
Table 1: Femoral graft position measurements (mean ⴞ
standard deviation) for all images from Group 1 and
Group 2 (n ⴝ 50).
Observer 1
Observer 2
Observer 3
Observer 4
78 ⫾ 4%4
67 ⫾ 3%2
61 ⫾ 4%
78 ⫾ 6%4
70 ⫾ 5%1,3,4
61 ⫾ 5%
78 ⫾ 4%4
67 ⫾ 3%2
61 ⫾ 5%
76 ⫾ 5%1,2,3
68 ⫾ 4%2
61 ⫾ 4%
different from observer 1 (p ⬍ 0.05, Student–Newman–
different from observer 2 (p ⬍ 0.05, Student–Newman–
different from observer 3 (p ⬍ 0.05, Student–Newman–
different from observer 4 (p ⬍ 0.05, Student–Newman–
M. Sati*, H.-U. Stäubli**, Y. Bourquin*, M. Kunz*, S. Käsermann**, L.-P. Nolte*
*M.E. Müller Institute for Biomechanics, University of Bern, Switzerland.
**Orthopaedics and Traumatology Surgical Clinic, Tiefenau, Switzerland.
Introduction/Objectives: A recent consensus of the international knee society revealed
that approximately 40% of ACL grafts are being surgically misplaced in current clinical
practice [International ACL study group, 1998]. This alarming finding exposes a serious
problem in the quality of a procedure that, if not performed correctly, can result in
premature degeneration of knee structures that eventually requires total knee replacement.
This problem is particularly disturbing since ACL injury usually occurs in younger
athletic individuals that would like to remain active. To help solve this problem, a
computer-assisted system has been developed at the M.E. Müller Institute for Biomechanics to perform intra-operative planning of ACL replacement
Design/Methods: Dynamic reference bases are fixed on the femur and tibia to track the
knee’s movement. Like a previous CAS knee system (Dessenne et al., J. CAS, 1995), no
intra-operative imaging is required and potential ligament attachment sites can be directly
digitized using a computerized palpation hook in a minimally invasive fashion when used
in conjunction with arthroscopy using standard endoscopic tools. Unlike previous technologies, the proposed system allows surgeons to define freely and interactively the
anatomical structures and viewpoints of these structures they judge are important for the
proper placement. The present system also allows incorporation of pre-operative X-rays
intra-operatively to assist in ligament placement. The computer should help in situ
planning of ligament placement by providing the surgeon information on graft impingement and elongation for various simulated surgical insertions and graft sizes. After
planning, the computer helps guide the surgical drill to the planned insertion site. To study
intra-operative variance in ligament placement, 6 clinicians (with varying levels of
experience) from two different clinics where asked to digitize a planned graft placement
within the same cadaver specimen first using only endoscopic control then with help of the
CAS system. Planned ligament positions of all surgeons using both techniques were
recorded in the computer for later comparison.
Results: Intra-operator variance of ligament position using only the endoscopic technique was strikingly high compared to that with the CAS system. Ligaments planned by
the less experienced surgeons showed both incorrect anatomical positioning (up to 10 mm
variance in the femur) and significant impingement. All ligaments placed with the CAS
system were placed anatomically correctly (variance only of the order of millimeters) and
impingement free.
Discussion and Conclusions: This flexible approach provides valuable quantitative
information on ligament positioning, impingement and deformations that are normally not
measurable in standard procedures. Such information will be valuable for improving the
consistency of ACL placement and providing better documentation that should help
surgeons to better compare the large number of present replacement techniques.
Abstracts from CAOS/USA ’99
J. Petermann1, R. Kober2, P. Heinze2, P.F. Heeckt3, L. Gotzen1
Dept. of Trauma-Surgery, Philipps-University, Marburg, Germany
orto Maquet GmbH & Co KG, Rastatt, Germany
Surgical Academy, Maquet AG, Rastatt, Germany
Introduction: Rupture of the anterior cruciate ligament is a frequent trauma among physically active adolescents and adults. Conventional replacement of the torn ligament often fails
due to incorrect placement of the tibial and femoral tunnels with consequent malpositioning of
the ACL substitute. Revision surgery has to be performed in approximately 15-25% of all
cases. For improved preoperative planning and precise intraoperative execution of the plan a
surgical planning and robotic system (CASPAR) has been adapted for ACL surgery. The
CASPAR-System allows for accurate positioning of the ACL graft according to the preoperative plan based upon 3D CT data.
Methods: Under local anaesthesia, fiducial markers (two until August 1999, since then one)
are placed in the distal femur and proximal tibia. A helical CT of both knee regions is then
obtained. After transferring the CT data to the planning station the surgeon plans the procedure
using predefined templates or free navigation. The entire planning procedure is PC-based and
facilitated by easy-to-use tools. The planning tools allow the surgeon to determine the position
of the ACL graft based upon individual experience and preferences or recommended standards. Skeletal landmarks such as the “Blumensaat line” form the basis for 3D planning. Since
the injured knee frequently has a higher degree of hyperextension, the uninjured contralateral
knee is used as a physiologic reference for the positioning of the ACL graft. Later impingement of the graft or intraoperative collision of the drill with the femoral condyles or fiducial
pins can be virtually simulated and avoided. Planning data are stored on a PC-card and
transferred to the robot. The surgical procedure starts by conventional harvest of the graft. Any
graft type, including allografts or synthetic grafts, can be used. We prefer bone-patellar
tendon-bone grafts. After preparation of the graft, the knee is flexed (about 115°) and femur
and tibia are rigidly fixed by external clamps. The fixation device is connected to the robot.
Excessive movement can be detected by a bone-motion sensor which automatically stops the
drilling process. After successful registration of the tibial fiducials, the pins are removed and
the tibial tunnel is drilled. The femoral part is done accordingly. The fixation device is then
removed and the ACL graft is placed and secured by conventional technique.
Results: To date 38 patients have been included in a prospective clinical trial. No major
adverse effects have been noted. Overall operating time was increased by approximately 30
minutes, with a clear trend towards shorter times in the last cases. One patient could not be
operated by the robot due to a hardware defect. Another four patients were only partially
operated by the robot due to registration problems, three at the femoral insertion. In these cases
surgery was completed by conventional technique. The postoperative course was uneventful
except for one patient who experienced a superficial infection at a pin site which resolved
under conservative treatment. On clinical examination all patients had stable knees with full
range of motion. No signs of impingement were noted. Patients were subjected to an
accelerated rehabilitation program according to Shelbourne.
Conclusions: Our initial clinical experience with the CASPAR-System shows it to be safe,
effective, and efficient for accurate placement of tibial and femoral canals in ACL replacement
surgery. The degree of precision achievable is well within the range of 0.5 mm, which cannot
be met by manual technique in a reliable and repetitive manner. Current disadvantages such
as additional surgery for pin placement and longer operating times have to be weighed against
obvious benefits for the patient. We believe that significantly increased surgical precision and
better 3D preoperative planning will improve long-term results and possibly avoid secondary
Anthony M. DiGioia, III, MD1,2, Branislav Jaramaz, PhD1,2, Mike Blackwell, MS2, David
A. Simon, PhD1,2, Fritz Morgan, MS2, James E. Moody, MA1, Constantinos Nikou, BS2,
Bruce D. Colgan, MS1, Cheryl A. Aston, MA1, Richard S. LaBarca, BS2, Eric
Kischell, MS2, and Takeo Kanade, PhD2
Center for Orthopaedic Research, UPMC Shadyside, Pittsburgh, PA.
Center for Medical Robotics and Computer Assisted Surgery, Carnegie Mellon University, Pittsburgh, PA.
Objectives: There has been little clinical research to examine the effects of patient
positioning and pelvic motion on the alignment of the acetabular implant during total hip
replacement surgery. Until now, no tools were capable of accurately measuring these
variables during the actual procedure.
Background: As part of a broader program in medical robotics and computer assisted
surgery, a clinical system has been developed that includes several enabling technologies.
Design/Methods: The hip navigation system (HipNav) continuously and precisely
measures pelvic location and tracks relative implant alignment intraoperatively. HipNav
technology is used to gauge current clinical practice and provide intraoperative feedback
to surgeons with the goal of improving the precision and accuracy of acetabular alignment
during total hip replacement.
Results: This system provides surgeons with a new class of image guided measurement
tools and assist devices.
Conclusions. These tools were successfully introduced into the clinical practice of
surgery with results showing the following: (1) There exist unpredictable and large
variations in the initial position of patients’ pelves on the operating room table, and
significant pelvic movement during surgery and during intraoperative range of motion
testing; (2) current mechanical acetabular alignment guides do not account for these
variations, and result in variable and—in the majority of cases— unacceptable acetabular
alignment; and (3) press fitting oversized acetabular components influences the final cup
H. Skibbe, M. Börner, U. Wiesel, A. Lahmer.
Berufsgenossenschaftliche Unfallklinik, Frankfurt am Main, Germany
The ROBODOC system has now been in use for about five years. Since October 1998,
application of the NEW REVISION SOFTWARE has been possible and 50 cases have
been operated using the new program.
Even though the pinless method can be used for primary THR, in revision cases the
ordinary “two-pin procedure” is used. On the one hand, the neck of the femur, which is
necessary for the pinless technique for pair matching, no longer exists due to the
osteotomy during the primary implantation. On the other hand, the artefacts in the CT
scans enable planning of the pinless model. While the import of the CT data is similar in
both procedures, the identification of the pins can only be carried out after excluding all
other metal signals (even though metal artefacts caused by the enclosed prosthesis are
Thereafter the main program can be started. A special technique is used to enhance the CT
images so that all the existing bone cement in the cavity which surrounds the old
prosthesis is clearly visible and can be distinguished from bone structures. A cutting path
can be planned to remove all of the existing bone cement. The next step is the planning
of the new prosthesis. Any prosthesis available in the ORTHODOC’s library can be used.
The new prosthesis is planned above the existing cavity of the old implant. Just like in the
program for primary THR, the prosthesis can be adjusted in any direction until a
satisfactory position is reached. The data is transferred onto a tape and loaded in the
ROBODOC. Intraoperatively, after the pin-finding procedure, the robot mills out the
existing bone cement and creates a new cavity for the new implant. The advantages of
using ROBODOC for revision surgery are obvious. Optimized preoperative planning of
the procedure is possible, the anteversion can be corrected, and the fibrous membrane and
sclerosis zone in the former cavity are removed by the cutter. The duration of the operation
is greatly reduced compared to the traditional method of removing the cement manually.
There is no risk of an intraoperative fracture of the shaft. Normally, the patient can be
allowed full weight bearing directly after recovering from the operation. A loose uncemented prosthesis can be replaced by a new uncemented implant. A loose cemented
prosthesis can be replaced either by a cemented or uncemented implant. If a cemented
implant is used as a new implant, only the removal of the bone cement can be planned with
the ORTHODOC program.
Abstracts from CAOS/USA ’99
Ulrich Wiesel, Armin Lahmer, Martin Börner, Hagen Skibbe.
Berufsgenossenschaftliche Unfallklinik Frankfurt am Main (BGU), Frankfurt am Main,
Branislav Jaramaz, PhD1,2, Anthony M. DiGioia III, MD1,2, Mike Blackwell, MS2, and
Constantinos Nikou, BS2
Center for Orthopaedic Research, University of Pittsburgh Medical Center Shadyside,
Pittsburgh, PA
Center for Medical Robotics and Computer Assisted Surgery, Robotics Institute, Carnegie Mellon University, Pittsburgh, PA
Objectives: The presentation intends to show the development and progress of robotic
THR at BGU Frankfurt with a special emphasis on the introduction and clinical use of the
Pinless system.
Background: In 1994 the first successful robot-assisted total hip replacement using a 3
pin-based system was performed at BGU Frankfurt. With an average OR-time of 90
minutes, using the ROBODOC 2-pin system for THR has become a standard procedure
at BGU. The robot guarantees precise trans-formation of the preoperative plan during
surgery, and femur fractures and varus or valgus malpositions can be avoided. In April
1999 the 2000th operation using ROBODOC was performed successfully at BGU Frankfurt.
The results were supported by a dog study at the Small Animal Clinic in Auburn,
Alabama, in 1995. There were no fractures or nerve palsies found in the ROBODOC
group, the gait analysis was superior, and there was closer alignment of the prosthesis to
strong cortical bone.
Design/Methods: The ROBODOC PINLESS SYSTEM was introduced at BGU Frankfurt in the middle of 1998. The system does NOT require the preoperative insertion of
registration markers (pins). In the preoperative planning phase, using the 3D CT data, the
surgeon places a model of the implant into position and the ORTHODOC determines the
cut path for ROBODOC. The surgeon then creates a 3D surface model which is used
intra-operatively for Pinless registration.
In the OR, after the head of the femur has been surgically removed in the normal
manner, ROBODOC is moved into position and a digitizer is used to collect points both
proximally and distally on the femur. The distal data is collected percutaneously. Once the
bone surface data is collected, the registration process is started. The points are compared
to the surface model created on the ORTHODOC to register the robot to the femur
position. The registration is verified by viewing independently digitized points superimposed on cross-sectional CT images. Once all the data has been collected, ROBODOC
starts cutting the implant cavity.
Two “recovery markers” are installed in the bone proximally and distally prior to the
surface point collection process. Once bone motion occurs during the milling of the cavity,
it takes about 30 seconds to relocate those markers to be able to resume the operation.
Results: After using the pinless system for almost a year, we believe that 95 to 98 % of
all patients can be operated on using the pinless system. Exclusion criteria are severely
disfigured post-traumatic cases, revision cases, and cases with a non-titanium implant in
the opposite leg, as there are too many metal artifacts in the CT scan. We have seen no
problems using the pinless system on patients with a cementless titanium implant in the
opposite leg.
152 patients were operated on at our hospital using the pinless system on a clinical trial
basis prior to the official release in October 1998. In the period between October 28th,
1998 and May 12th, 1999, 423 patients were operated on using ROBODOC, 253 of them
with the pinless system. The number of pinless procedures has risen steadily since
October. From April 1999, after the arrival of two new pinless robots, virtually all patients
were done pinless with only a few exceptions. Postoperative knee pain was eliminated
almost completely, as compared to the pin cases. The Pinless procedure has become
everyday routine at BGU just like the pin-based procedure: At the end of April 1999 we
had performed a total of 2000 ROBODOC operations since the introduction of the system
in 1994.
Conclusions: The major advantages of the pinless procedure are that only ONE operation is needed, providing greater scheduling flexibility for the patient and surgeon.
Postoperative knee pain is eliminated in most cases, and the costs per surgery are reduced.
The surgical time and radiation exposure is the same as in the pin-based system, while the
system accuracy and reliability are equal to those of the pin-based system.
Objectives: Radiographs are the most common means of postoperative imaging and
assessment following total hip replacement (THR) surgery. In order to more precisely link
the incidence of dislocation to implant placement, it is important to precisely measure
acetabular cup alignment relative to the pelvis.
Background: Most clinical studies of dislocation are based on the evaluation of postoperative radiographs. Improved measurement techniques are necessary to eliminate
potential inaccuracies in such studies.
Design/Methods: A computer assisted reconstruction technique is used to precisely
measure the cup orientation by matching the synthetic projection of the geometric model
to the outline on the digitized anteroposterior radiograph. This simple interactive procedure allows relatively precise measurements of cup orientation with respect to the plane
of the x-ray film.
Results: Initial studies indicate that the cup orientation can be measured with relatively
high accuracy (with errors on the order of 1 degree in abduction and 2–3 degrees in
version). However, the unknown flexion/extension of the pelvis on the x-ray table may
significantly affect the measurements. Based on our measurements, individual variations
in pelvic rotation between subsequent radiographs can be on the order of 20 degrees.
Conclusions: Accurate measurements of the acetabular cup alignment from postoperative radiographs are essential to establish a better link between the cup placement and the
postoperative outcomes. Precise measurements of the cup orientation from the radiographs
are not sufficient and additional measurement of pelvic orientation at the time of x-ray
acquisition are necessary. By using a single CT scan as a reference and by employing
2D–3D registration techniques, it is possible to obtain the measurements of pelvic
orientation and therefore perform complete measurements of cup orientation.
John A. Hipp, PhD, Nobuhiko Sugano, MD, Michael Millis, MD and Stephen
Murphy, MD
Department of Orthopaedic Surgery, Baylor College of Medicine, Houston, Texas, USA
Acetabular redirection osteotomy can be used to relieve pain, improve function, and
extend the life of dysplastic hip joints. To understand better the factors that may determine
the acetabular reorientation that minimizes pressure, joint contact pressures were calculated by computer assisted methods in 70 displastic and 12 normal hips. Calculated
pressures were consistent with pressures estimated and measured by other investigators.
Contact areas were 26% smaller, and contact pressures were 23% higher, in the dysplastic
hips compared with the normal hips. When the acetabula were reoriented to minimize
contact pressures for an activity such as the midstance phase of gait, then contact pressures
were elevated for dissimilar activities such as stair ascent. Contact pressures in the
dysplastic hips were reduced when the acetabula were rotated in the frontal plane to
increase lateral coverage or rotated in the sagittal plane to increase anterior coverage. In
most of the dysplastic hips, contact pressures were reduced twice as much when the
acetabulum was rotated in the frontal and the sagittal planes. Computer assisted methods
to quantify joint contact pressures can be used to assess potential candidates for reconstruction, plan acetabular redirection surgery, and possibly may improve the long term
success of acetabular redirection osteotomy.
Abstracts from CAOS/USA ’99
F. Langlotz, U. Langlotz, and L.-P. Nolte.
M.E. Müller Institute for Biomechanics, University of Bern, Bern, Switzerland
Objectives: In pelvic osteotomies to treat dysplastic hips, some of the intra-operative
chiseling has to be done with the osteotome blades out of view, and there is potential to
harm the hip joint and surrounding structures.
Total hip replacement is being regularly performed with a high benefit for the patient.
Incorrect spatial position is a key factor for the most common postoperative complication.
Assessment of inclination and ante-version angles for cup placement in vivo is insufficient, and a better means to assess the anteversion angle in particular is needed.
Background: We present two modules of the SurgiGATE navigation system developed at
our institute. The first application for the Bernese Periacetabular Osteotomy (PAO) can also be
applied to related osteotomies. It permits interactive visualization of osteotomes and subsequent reorientation of the acetabular fragment. The second module for cup placement incorporates a sophisticated planning software. Intraoperative feedback on the instrument position
is given throughout the implantation of the acetabular component.
Design/Methods: Both modules feature a workstation (Sun Microsystems), and an
optoelectronic navigator (Northern Digital). Modifications to the tool-sets consist of
marker probes with LEDs attached to the tracked instruments. For the computer-assisted
PAO, a CT-scan is loaded into the SurgiGATE system. Image data is segmented semiautomatically and 3 landmarks are selected for intraoperative registration: (1) spina iliaca
anterior inferior, (2) a point on the iliac rim, 2 cm posterior of the spina iliaca anterior
superior, and (3) the contra-lateral spina iliaca anterior superior. In a 3D reconstruction,
the part of the acetabulum that is supposed to be liberated is marked interactively.
Intraoperatively, a dynamic reference base is attached to the pelvis. This base is equipped
with LEDs and compensates for relative motion of the patient. It may be fixed to the ilium of
the operated side. However, a contralateral placement through a 1-cm skin incision is
advantageous, because the operating field is disturbed less. Registration is performed using our
restricted surface matching algorithm, which is fed with 3 anatomical landmarks plus a cloud
of 12 points on the bony surface. These 15 points are digitized using an LED-equipped pointer.
For the guidance of the osteotomies, the location of each osteotome is displayed in real-time
within a series of CT reconstructions. The image guided reorientation of the free acetabular
fragment is visualized three-dimensionally and by means of three angular values.
Total Hip Replacement. Preoperatively, a CT-scan of the patient is loaded and type and
size of the implant are chosen. Specific geometric implant data is imported from a
database. The surgeon translates and/or rotates the implant interactively to determine the
optimal position of the cup. Inclination and anteversion angles are computed and displayed during these adjustments in real-time. Once all parameters are determined, two
simulated X-rays are calculated from the CT data. Finally, 4-5 landmarks, namely a point
at the spina iliaca anterior superior, one at the fossa acetabuli, and 2–3 on the acetabular
rim are stored for intraoperative registration. Intraoperatively, the CAS system does not
dictate any specific approach. It is used as a flexible tool to guide according to the
preoperative plan. The DRB is attached to the pelvis with a standard bone screw. There
are two different attachment locations depending on the approach: for the lateral patient
position it is fixed cranially of the acetabulum, and for the supine position an attachment
is chosen at the spina iliaca anterior superior through a 1-cm incision. The next step is the
registration using the algorithm mentioned for PAO. During all relevant surgical actions
of the preparation and positioning of the implant, i.e., reaming, trial impacting, and
impacting, the system provides online feedback of the position of the surgical tool.
Results and Conclusions: Both modules have been introduced into clinical routine in
centers all over Europe. First results indicate that greater safety and higher accuracy can
be provided, but additional studies are necessary to finally prove the long-term clinical
N. Sugano, Y. Sato, T. Sasama, K. Nakahodo, S. Yoden, T. Nishii, T. Sakai, K. Haraguti,
K. Ohzono, K. Yonenobu, S. Tamura, T. Ochi
Department of Orthopaedic Surgery, Osaka University Medical School, Japan
Objectives and Background: We report additional benefits when acetabular navigation
is combined with femoral navigation during total hip arthroplasty (THA). Early results of
acetabular navigation with HipNav are encouraging, but more information about the
femoral side is also necessary. We therefore developed a combined acetabular and femoral
navigation system using infrared LED markers and optical cameras (Optotrak, Northern
Digital Inc.).
Design/Methods: Ten patients underwent THA for osteoarthritis secondary to hip
dysplasia. A cementless total hip bearing with modular neck and head system (ANCAFIT) was used. Preoperative transverse images from the level of the superior anterior iliac
spines to the level of the femoral canal isthmus were obtained using a helical CT scanner.
Images of the femoral condyles were used to measure femoral anteversion. Image data
were stored on an optical disk then transferred to a workstation. From the CT data, a
titanium alloy femoral stem was machined, and 3D and femoral bone surface models were
reconstructed. For measuring cup angles and femoral neck anteversion, 2 coordinate
systems were defined: the standard anatomic pelvic coordinate system, and the femoral
coordinate system. Cup size was determined from the A-P diameter of the acetabulum.
The cup was to be placed in 40° of abduction and 20° of anteversion relative to the pelvic
coordinate system and in the original acetabulum. In the OR, patients were placed in a
lateral decubitus position. The Optotrak position sensor camera was placed atop the wall
caudal to the patient. A posterolateral approach was used. For pelvic tracking, a plate with
6 LED markers was fixed to two pins inserted in the iliac crest with an extraskeletal
fixation system. For femoral tracking, an Optotrak digitizing probe with LED markers was
connected to a plate screwed to the greater trochanter. After shape-based surface registration with the bone models, the following were measured: Femoral neck osteotomy level
(planned and actual), position of the cup center, acetabular cup angle, femoral anteversion,
and limb length discrepancy. Relative movement of pelvis and femur was also recorded
when the final range of motion was tested intra-operatively.
Results: Shape-based registration was successful in all cases within 0.5 mm of error.
The femoral navigation system was used to identify the appropriate location for neck
resection, and to avoid malalignment of stem while broaching the femoral canal. In the
pelvic coordinate system, the patients’ cup angles (abduction/anteversion) were: 42/26,
46/28, 39/37, 37/28, 41/31, 42/32, 40/21, 39/30, 41/25, and 44/23 degrees (average
41/28°). In the original (table-based) CT coordinate system, the angles were: 40/35, 47/23,
44/18, 39/10, 42/17, 36/15, 37/15, 39/23, 42/25, and 46/24 degrees (average 41/21°). After
implantation of the cup and femoral stem, the limb length discrepancy, femoral neck
anteversion and offset were optimized according to the simulation. Limb length discrepancy improved from -9.8 mm preoperatively to –2.3 mm on the simulation and –1.4 mm
on postoperative x-ray. Average femoral anteversion was 31° (range 5–50°) preoperatively and was controlled to 25° (range 15–34°) postoperatively. After reduction of the hip
joint, range of motion was checked. During flexion-extension movement, the pelvis also
moved in the same direction with a range of 9-26°. In maximum flexion, the flexion of the
femur relative to the pelvis ranged from 79 to 96°.
Conclusions: Efficacy of computer navigation systems with an optical camera and LED
markers for acetabular cup placement was confirmed. Expanding computer navigation to
the femoral side of THA brought benefits in planning the femoral neck osteotomy level,
aligning the stem with the medullary axis, and modifying femoral anteversion, especially
in cases with hip dysplasia. Femoral navigation can provide information on limb length
without intra-operative x-ray. Navigation also revealed the relative movement of the
components during the intra-operative check of the safe range of motion. Data acquired
intraoperatively can be used for postoperative education regarding the safe range of
Abstracts from CAOS/USA ’99
E. Frederick Barrick, M.D
Inova Fairfax Regional Trauma Center, Falls Church, Virginia, USA
K Mallik, RD Zura, and DM Kahler
University of Virginia Health Sciences Center, Department of Orthopaedic Surgery,
Charlottesville, VA
Objectives: At present, registration for percutaneous placement of cannulated screws in
the pelvis requires use of an external frame with an attached fiducial array. We sought to
elucidate the effect of varying CT slice thickness and fiducial frame geometry when using
a point-based registration algorithm.
Background: Minimally invasive computer assisted procedures have the potential to
decrease the soft tissue trauma associated with surgical exposures. Increased accuracy in
screw placement is possible, while optical tracking of surgical instruments decreases the
need for intraoperative radiographic imaging. Accurate registration is essential to this
Design/Methods: Fiducial arrays (12.7-mm aluminum spheres attached to radiolucent
carbon fiber rods) were affixed to the left brim of a plastic pelvis. Three geometries were
tested: a 4-point planar array; the same array with a fifth fiducial on the contralateral iliac
wing; and a 4-point tetrahedral array. Seven titanium washers were applied to the pelvic
model to identify anatomic points and CT was performed. The digitized data was
transferred to a workstation (StealthStation) and used to build a 3D model. The pelvis was
registered, instrumented with a dynamic reference array, and continuously optically
tracked during the procedure. Registration error of ⬍1 mm was routinely obtained. A
surgical plan was defined by entering the washer centers as entry points and a central
fiducial as the target. The distance from each anatomic point to the reference point was
calculated, and the deviation from the surgical plan was displayed. This distance represented the registration error for each condition. Each trial was performed 3 times. Error
was evaluated for 3 different conditions. Using the 4-fiducial planar array, the registration
accuracy with 1.0-, 1.5-, and 3.0-mm CT cuts was determined. A fifth fiducial was then
added to the contralateral pelvis to determine the effect of greater fiducial spread. The
influence of frame geometry was examined by comparing the 4-fiducial planar array with
a 4-fiducial tetrahedral array.
Results: There was a trend toward increasing error in registration with increasing
distance from the fiducial array, although this was not significant.
Condition #1: Error for 4-fiducial planar array with variable CT thickness.
Mean Error (not significant)
CT cut thickness
1.0 mm
3.9 mm
1.5 mm
3.0 mm
3.0 mm
4.1 mm
Condition #2: Effect of adding a fifth fiducial with 3.0 mm CT slices.
SD and Range
Mean Error (p ⬍ .017)
Four fiducial array
4.1 mm
1.3 1.8–6.7 mm
Five fiducial array
3.3 mm
1.2 1.5–5.2 mm
Condition #3: Comparison of planar and tetrahedral four fiducial arrays.
SD and Range
Mean Error (p ⬍ .0005)
Planar array
4.1 mm
1.3 1.8–6.7 mm
Tetrahedral array
2.9 mm
0.6 1.7–3.8 mm
Conclusions: Large spherical fiducials offer no significant improvement in registration
accuracy when CT cuts ⬍ 3.0 mm are used. Clinically, thicker cuts are desirable to
decrease radiation exposure. A fifth contralateral fiducial increases registration accuracy,
but trapezoidal frame geometry is superior to both the planar frame and the 5-fiducial
array. As expected, there was a trend toward greater error with increasing distance of
target points from the array. With the planar array, 24% of the trials produced errors ⬎ 5
mm. A fifth fiducial decreased this to 9%. With the trapezoidal array, accuracy of targeting
was within 5 mm for all trials. The increase in accuracy with the modified protocols was
statistically significant. This technology potentially offers great advantages over conventional techniques, but is not yet sufficiently reliable to completely do away with intraoperative radiographic imaging (especially when a margin of ⬎5 mm cannot be tolerated).
Validation of registration accuracy with use of limited fluoroscopy is essential.
Objectives: This clinical study was undertaken to see if surgical navigation with a
CT-guided computer-assisted system would improve the accuracy, reduce the risk, and
expand the indications of iliosacral fixation.
Design/Methods: The exact location of the pelvis is triangulated by an optical tracking
system then matched with a preoperative CT scan transferred to a computer workstation.
The trajectory of the intended screw is planned using planar and 3D reconstructions
produced by the workstation. A drill or drill guide is tracked to permit a guide pin to
follow the planned path, then the screw is inserted. Fluoroscopy was used for backup.
Indications included acute unstable 61-B and 61-C posterior pelvic ring disruption with
sacroiliac dislocation or sacral ala fracture, and nonunion of sacral ala fracture. Surgical
navigation was used in 16 cases. Fiducials were spheres on an external fixator in 3 cases,
titanium guide pins in 6, and cutaneous radiopaque markers in 7. After placement of
fiducials, a CT of the pelvis was obtained using a helical scanner with 1-mm cuts. In the
OR the pelvis was registered with the CT on the workstation. Open reduction was
performed if necessary. The computer displayed the position of the guide-pin trajectory in
relation to the virtual pelvis and thus to the real pelvis almost instantaneously. In later
cases, a drill guide on a passive locking articulated arm was used to reduce excess motion,
with a stop to assure appropriate depth. Lag screw insertion was the final step.
Results: Registration was successful in 14 of 16 cases of attempted surgical navigation.
Two could not be registered because the patient with attached external fixator was too
large for the CT scanner, or fiducial pins did not hold in severely osteoporotic bone. Screw
insertion was attained in all registered patients except one who had plates and screws from
concomitant fixation of an iliac fracture blocking the path of the guide pin. Placement of
13 iliosacral screws was excellent in 11 patients. Two cases had misplaced screws: one
through the first sacral foramen and another exiting anterior to the sacral body. Of the 6
patients with titanium pin fiducials, 2 were placed accurately, 2 were misplaced, 1 had
hardware in the way, and 1 had osteoporosis. Of the 3 with external fixation spheres, 1
patient was too large, the other 2 were successful. All 7 with cutaneous markers registered
successfully, all screws were accurately placed; 4 were prone and 3 supine. Two pediatric
patients had 4.5-mm cannulated screws inserted using cutaneous markers for registration
in the supine position. Three patients in whom visualization by fluoroscopy was impossible had 4 screws placed accurately, while two patients whose anomalous sacrum gave
minimal clearance had well-placed screws also.
Fluoroscopic iliosacral screw fixation is currently the least invasive and best tolerated
means for fixing disruption of the posterior pelvis. It is essential to aim accurately to keep
the screw inside the bone to avoid major nerves and vessels. Using fluoroscopy has the
problem of poor visualization due to bowel gas, contrast media, and obesity. Also, it may
be difficult to execute a safe trajectory if there are anatomic variations of the sacrum.
In addition to executing navigation, the program identifies anatomic variations and
assists in surgical planning. Viewed together, triaxial images and 3D reconstruction can
identify patterns of sacral fractures that cannot be recognized on plain x-ray or axial CT
images. The major disadvantage with this system is the registration method using invasive
fiducial insertion prior to the CT scan. An external fixation frame with spheres is only
practical when needed for resuscitation. The use of guide pins requires an extra procedure,
and is thus impractical, costly, and time consuming. It was also less accurate than the other
two means. Cutaneous radiographic markers worked elegantly. They are convenient,
inexpensive, avoid two operations and extra incisions, and are thus far effective. They can
be placed posteriorly closer to the sacrum, and thus provide better accuracy if the patient
can be prone for the operation.
Conclusions: The system works well with an external fixator in the supine position and
with the cutaneous markers supine or prone. It is accurate when fluoroscopic visualization
is not possible and when CT precision is needed to maneuver through anatomic variations.
Abstracts from CAOS/USA ’99
ViewPoint™ SYSTEM
George A. Brown, M.D., Michael C Willis, M.D., Keikhosrow Firoozbakhsh, Ph.D.,
Adam Barmada, M.D., Charles L. Tessman, R.N., Andrew Montgomery, Tom DeCoster
MD, Mark Crawford MD, Brenton Milner MD
Department of Orthopaedics, University of New Mexico School of Medicine, Albuquerque, New Mexico, USA.
Objectives: To report on the intraoperative use of CT image-guided navigation in the
open treatment of complex acetabular fractures and in the percutaneous treatment of
acetabular and sacroiliac injuries.
Background: Baumgaertner in his 1999 review article discusses the problem of acetabular joint penetration by screws or pins. However, he does not mention computer
imaging as an option for reducing this risk. The integration of CT imaging modalities with
modern frameless stereotactic techniques has not been incorporated in operative fixation
of displaced acetabular fractures treated with ORIF. Minimizing surgical exposure in
complex acetabular fracture surgery would potentially reduce muscle weakness, improve
recovery time, and reduce heterotopic bone formation and all are potential benefits of CT
image-guidance systems. Placement of anterior column screws should be more accurate
and the screws should be more easily inserted. Percutaneous fixation of acetabular
fractures, sacroiliac separations, and sacral alar injuries should be possible improving
patient safety.
Design/Methods: Eleven complex acetabular fractures were treated using image-guided
software intraoperatively. Two patients were treated with percutaneous fixation, one for
acetabular fracture and the second for sacroiliac fracture dislocation. The goals set to be
achieved in this work are: (a) evaluation of the clinical usefulness of this system in pelvis
surgery, (b) find registration techniques that work and are clinically useful, (c) report on
accuracy of registration techniques employed.
Results: Accuracy in the range of 1 mm was found when registration is precise, and is
in the range of 3.5 mm when registration is only approximate.
Conclusions: Each of the implants placed under image guidance was found to be
accurate and without penetration of the acetabular joint space, the neural foramina, or the
spinal canal of the S1 vertebral body on CT scans obtained postoperatively. The set-up
time for the system was minimal and added benefits include reduced intraoperative
fluoroscopy time and obviation of additional surgical approaches in some cases. Compared to a series of similar fractures treated traditionally, the total operative time was
significantly less when CT-imaging was employed (16 & 20%). Intraoperative CT-guided
imagery was found to be an accurate and suitable method for use in the operative treatment
of complex acetabular fractures with substantial displacement. CT image guided surgery
(IGS) helped to prevent articular hardware penetration in all nine posterior wall fractures
and offers significant improvement in guidance for iliosacral screw placement in injuries
to the pelvic ring. Iliosacral screws can be placed in alignment with the sacral ala
(posterior to anterior) significantly increasing the safe zone within the sacrum. This cannot
be accomplished without 3D-computer guidance.
F.J. Ortiz Zaragoza, J.M. Fernández Meroño, R. Torres Sánchez
Dpto. Automática y Electrónica Industrial, Universidad Politécnica de Cartagena, Cartagena (Murcia), Spain
Objectives: Distal locking and drilling of intramedullary nails is problematic due to the
radiation exposure required. A method based on computer vision and geometric modeling
has been developed to achieve nail fixation with minimal radiation exposure. A new
technique to establish the correspondence between the projected images of the bone and
a 3D virtual nail has been developed. To achieve this goal, a real-virtual matching has
been adopted.
Background: The nail is an object of constant cross-section, though irregular, with the
main inertia axis as an invariable property. Therefore, if the main inertia axis of the
projection is calculated in any view, it would coincide with the nail axis projection. This
conclusion confirms that the 3D position of the longitudinal axis can be obtained through
two projections. A unique trajectory can be defined as a straight line passing through two
points. Therefore, the drill trajectory would be clearly defined by knowing the 3D position
of two points of the nail, which can be the centers of the two holes in the nail.
The 3D reconstruction from two projections to obtain the orientation of a perpendicular
vector through the distal holes is almost impossible, due to insufficient information. As
there is a great probability that the intersection of the holes could be hidden, or could not
be detected to determine the vector orientation, it is necessary to pursue an alternate
method that will permit the reconstruction of the trajectory that defines the drilling of the
Design/Methods: Although it is impossible to completely reconstruct the nail by means
of classic reconstruction techniques, it is possible to do so with its longitudinal axis. The
direction of the guidance vector for the drilling tool is set up by two angles: Rotation and
Elevation, with reference in the center of the segment from one center to the other. Since
the orientation of the principal axis is known, and thus one of the trajectory angles, the
proposed solution consists of the comparison between the intersection shape of the holes
of a virtual model of the nail, generated by computer, and a real projection of the nail,
which offers, visually, the intersection of the holes named before.
Knowing the position of the holes relative to the nail’s main axis, it will be situated in
such a way that, turning the virtual nail on its main axis (which coincides in position with
the real nail), the two intersections come to coincide with each other. This is achieved
through an optimization method based on least squares technique. Once the matching
algorithm has found the best coupling of the images, the rotation of the virtual nail stops,
and that will be the spatial position of the real nail, coinciding with the virtual one.
The necessary source coordinates are obtained from the fluoroscope calibration. The
entire process has been integrated into a software which also contains the linearization
process and the fluoroscope calibration. The final goal of the system is to obtain the angles
which define the spatial orientation of the drilling tool. These turning and elevation angles
are referred to the line joining the centers of the holes; the center of this segment is
referred to a mobile coordinate system fixed to the nail. In turn, this system is related to
a reference system fixed to the surgical table.
Results: The system has been tested in laboratory using images from a CCD camera and
positioning the nail with the desired orientation angles. The software obtained the correct
angles in a 0.5° scale, and the correct reference center in 0.5 mm scale. It was also tested
starting from fluoroscope images and the same good results were obtained.
Conclusions: We have developed a registration technique for distal locking that allows
correspondence between the computer world and the real world, without the use of
fiducials. The registration technique is shape based, and we have developed an original
method to perform the correspondence by comparing projections with virtual figures
modeled on the computer. The proposed architecture is low cost, PC-based, and with an
easy-to-use interface to the surgeon. This technique has demonstrated good results for the
correspondence and, we believe to great advantage, makes the use of fiducials unneccesary: No fiducials 3 no previous surgery.
Abstracts from CAOS/USA ’99
F. Langlotz, M. Sati, R. Hofstetter, M. Slomczykowski, L.-P. Nolte.
M.E. Müller Institute for Biomechanics, University of Bern, Bern, Switzerland
Allen P. Schlein, MD
Orthopaedic Surgery Associates, Bridgeport, CT, USA
Objectives and Background: As an in-situ X-ray imaging means, the mobile C-arm
provides information about shapes and types of fractures, relative positions of fragments,
implants and surgical instruments. A continuous or pulsed mode provides motion image
feedback of surgical actions. Use of the C-arm has brought about intramedullar nailing to
fix, e.g., femoral shaft fractures in a now-standard low-invasive surgical procedure.
However this technique still has significant problems such as high radiation exposure to
patient and OR staff, especially if the continuous mode is used. In this mode, the imaged
motion sequences contain static elements (e.g., the bony anatomy) and dynamic elements
(e.g., the instrument performing the surgical action).
In fluoroscopy-based navigation, the static elements are recorded with a single C-arm
image, which is stored on a computer workstation. Motion of the dynamic parts is tracked
with an optoelectronic camera, and their current positions are drawn on the C-arm image.
This approach provides a virtual continuous fluoroscope mode without the device actually
in place.
Objectives: To describe a method of reducing femoral fractures and facilitating distal
targeting that is efficient, inexpensive, and limits the radiation exposure of the surgical
Design/Methods: The aim of this work was to not only visualize a surgical instrument
(drill) but also a femoral nail and bone fragments through super-imposed line graphics to
inform the surgeon about their spatial relations in real-time. Additional information about
torsional relations of fragments is provided through intraoperative measurements. That
makes this system a versatile tool for the fixation of femoral fractures using intramedullar
Referencing. Marker shields with 4 LEDs are attached to each bone fragment, the
femoral nail, the drill, and the C-arm image intensifier. The LEDs define local coordinate
systems (COS) with a fixed relation to every point on the assigned rigid body. As the
surgical object consists of several moveable rigid bodies, we introduced the concept of
multiple patient referencing by attaching patient references to the nail (N-COS) and the
proximal and distal fragments (PF-COS and DF-COS). A C-arm image reflects the
momentary set-up of the objects and must be registered to each patient reference separately.
Bi-planar 3-D Reconstruction. Two registered C-arm images from different view angles
are used for 2D/3D reconstruction. An anatomic landmark is identified manually in both
images before calculating the related 3D position. We first determine the 3D straight-line
of the X-ray beam associated to a 2D point in the first digital C-arm image. The second
digitization then defines where on this line the landmark point is located. In this way, the
bone fragments to be navigated are marked by digitization of their main axes.
Navigation of Fragments. With their known axes identified, fragments are navigated
similar to tools and implants allowing visual control of the reduction of a fracture. Since
the image is static and the fragments have been moved after its acquisition, the part of the
image with the distal fragment does no longer reflect its position. The reconstructed and
navigated distal axis does, however, reflect the real movement and can be used for
fragment alignment.
Rotational Adjustment. The superimposed line graphics do not allow control of the
remaining torsional correction. The anteversion (AV) angle, specifying rotation around
the femoral shaft axis is, however, of high significance in the reduction procedure.
Calculation of AV angle is possible based on multiple registered images and the landmark
reconstruction method that allows precise geometric measurements. Multiple patient
referencing also enables real-time measurements for precise intraoperative torsional
adjustments. Six anatomic landmarks are obtained from bi-planar 3D reconstruction. They
can be uniquely identified using 3 pairs of registered fluoroscopic images. With all
landmarks present in the DF-COS the AV-angle is calculated through a trigonometric
function that implements the AV-angle definition.
Results and Conclusions: This system for fluoroscopy assisted femoral fracture reduction has been successfully used in one surgery so far. It was possible to reduce the amount
of radiation exposure considerably. Further applications of the described concept are
currently under development.
Background: Since World War II, intramedullary rodding of long bones has progressively gained acceptance. Over the last two decades, advancement in the technique has
been principally related to changes in materials and to the locking of the rods, the latter
being intended to prevent the shortening and rotational deformities often associated with
treatment of unstable fractures.
Distal locking requires increased skill and experience on the part of the surgeon, and
this portion of the rodding technique has been known to frustrate even the acknowledged
expert. An increase in the time required for reducing a femoral fracture and performing the
distal locking also leads to prolonged radiological exposure for both the patient and
surgical staff.
The development of the Euclid system and the associated surgical technique was meant
to facilitate the procedure of intramedullary rodding of the long bones. The device is
inexpensive and easy to operate as it is a semi-automatic device not requiring expensive
and complicated computer software.
Design/Methods: Euclid is a floor-mounted device with an outrigger arm that is
controlled in elevation and medial-lateral motion by linear actuators. The device resembles the base of an image intensifier, both in appearance and function. The purpose of the
base is to manipulate both the reduction head and the targeting head in the field of an
image intensifier.
Initially, the patient is positioned on a fracture table and traction utilised to reduce the
femoral fracture. The sterile reduction bow is snapped into place on the extended arm of
the machine, and the machine draped with a split sheet. The reduction bow is then placed
under the distal fracture fragment. A standard flexible guide wire is placed down the shaft
of the femur. As the wire passes through the fracture site, the fracture is manipulated via
the control pedals, allowing for its passage down the femur. This obviates the need for the
surgical assistant to place his hands in the radiological field while manipulating the
fracture fragments.
After reaming of the canal and placement of the fixation rod, the reduction yoke is lowered
and removed. The targeting head is then attached to the outrigger arm. The image intensifier
head is positioned over the distal holes in the rod, and the leg rotated until they appear as
perfect circles on the video screen. The targeting head is now aligned in space with the joystick
control, so that the contained stainless steel bushing appears on the video screen as a perfect
circle. This indicates that the holes in the rod and the bushing lie in a parallel plane in space
as described by Euclidean Geometry. The foot pedals are used to superimpose the target
bushing and the holes in the femoral rod. An appropriate sized drill is placed through the
targeting device and the holes placed across the femoral shaft. By exchanging the inner
cannulated bushing, the depth gauge and subsequently the screwdriver can be placed through
the targeting head. This allows the targeting head to act as a tool steady rest, facilitating
transverse screw placement and minimising the associated radiation exposure.
Results: Construction of a simple device, which can aid in the reduction of a femoral
fracture and facilitate targeting of the distal locking screws, has been shown to be feasible
at minimal cost.
Conclusions: This paper describes the development of a device utilised to facilitate the
technique of distal locked femoral nailing and minimising both operating time and
radiological exposure. It can be utilised to advantage with either a simple image intensifier
or with sophisticated tools and digitalized imaging equipment.
Abstracts from CAOS/USA ’99
R. Bächler, J. Kowal, Y. Bourquin, M. Sati, L.-P. Nolte
Maurice E. Müller Institute for Biomechanics, Bern, Switzerland
Objectives: The main goal of biological plate osteosynthesis is to optimally reposition
the fracture with a subsequent stabilization of the fragments. Damage to nearby soft tissue
should be minimized. This calls for minimally invasive surgical techniques that are
extremely demanding. Computer aided surgery (CAS) systems have proven their capability to augment the safety and accuracy of difficult interventions. This study presents a
system for improving the navigation of fracture plate positioning and fixation during
biological osteosynthesis.
Background: In minimally invasive plate osteosynthesis interventions, the plate is
inserted percutaneously with a manipulation handle from either the distal or proximal side.
Usually, the plates are bent prior to insertion in order to optimally fit to the bone surface.
For biological osteosynthesis, the form is determined using contour X-rays. Insertion of
the plate is controlled using either different fluoroscopic views or with constant fluoroscopic control. After positioning of the plate, the fracture fragments have to be screwed
to it. Each screw hole is made accessible through a small incision. The location of the hole
is determined using fluoroscopy or by searching for it with the drill tip. However, intense
use of fluoroscopy results in significant radiation exposure for both the patient and the
surgical staff.
Design/Methods: The fracture plate navigation system is based on the CAS system
developed at the Müller Institute. A fractured plastic bone, various osteosynthesis plates,
a surgical drill, and a screwdriver were instrumented with markers for the navigation
system. To allow easy navigation, 2D X-ray images of fractured bones were integrated
into a 3D graphics environment (OpenGL/OpenInventor). X-ray images were acquired
with the fluoroscopy-based navigation system, and the X-ray projection parameters were
imported as well. Using a special projection model derived from these parameters, the 3D
graphical representations of the surgical tools and implants can be displayed on top of the
X-ray images. Using a specially-developed hole-pointer, the position of each hole of the
deformed plate can be digitized. The system then updates the shape of the bent plate based
on its generic 3D model. During the positioning and fixation steps, the modified shape is
displayed. The holes need not necessarily be digitized in a fixed order, as the system is
able to detect the corresponding hole for a given pointer position.
Positioning of the fracture plate is assisted by showing different X-ray views of the
fractured bone. The surgeon then guides the plate according to the representation on the
computer screen. The accurate representation of the plate shape allows the surgeon to
choose the correct location of the screw holes. After the plate has been correctly
positioned, its fixation is assisted by the system. A special guidance mode allows the drill
to be correctly aligned for each screw hole. The drilling-depth can then be controlled with
the help of the system. Using the instrumented screwdriver, each screw can be correctly
inserted. The system automatically detects which screws have been inserted, and provides
the surgeon with visual feedback on which screws have already been inserted, which
screw is currently being inserted, and which screws remain to be inserted. Thus, it is
possible to prevent insertion of a screw into a hole that already contains one, and detect
screws that have not been inserted.
Results: The system provides the surgeon with realistic and interactive feedback on
normally hidden surgical actions. The drilling of the screw holes and the screw insertion
can be guided for arbitrarily deformed fracture plates. Using a fracture plate database, the
system can be easily extended to support plates from different manufacturers and for
different surgical procedures. The accuracy has not yet been measured using a proven
method, but first trials on plastic bones turned out to be very promising.
Conclusions: The existing system showed its potential to improve minimally invasive
fracture fixation using advanced image interactive technology. Moreover, replacing constant mode fluoroscopy with a fluoroscopy-based CAS system reduces radiation exposure
to both the patient and the surgical staff significantly.
Y. Raja Rampersaud MD, David A. Simon PhD, Kevin T. Foley MD.
Image-Guided Surgery Research Center, Memphis, TN, USA
Objectives: To derive theoretical translational and rotational accuracy requirements for
image-guided (IG) spinal screw placement using existing morphometric data.
Background: Previous studies have examined the clinical accuracy of particular IG
systems, underlying causes of inaccuracy in these systems, and methods for quantifying
this inaccuracy. However, application-specific accuracy requirements have not been
adequately addressed.
Design/Methods: We developed a geometric model relating spinal anatomy to accuracy
requirements for IG surgery. Modelling the pedicle as a cylinder, we determined the
accuracy required to avoid pedicle perforation at its narrowest point using clinically
relevant screw diameters.
Results: An inverse relationship was found between allowable translational and rotational errors for safe screw placement. As anticipated, accuracy requirements were
greatest at levels where the requisite screw diameter approximated the dimensions of the
pedicle. These requirements were highest for T5, followed in descending order by T4, T7,
T6, T3, T12, L1, T8, T11, C4, L2, C3, T10, C5, T2, T9, C6, L3, C2, T1, C7, L4 and L5.
Accuracy requirements ranged from 0.0 mm/0.0° at T5 to 3.8 mm/12.7° at L5.
Conclusions: This study demonstrates that extremely high accuracies are required to
safely place pedicle screws at certain levels of the spine. In some cases, these accuracies
are beyond the capabilities of existing IG systems. We hypothesize that other factors, such
as the mechanical constraints imposed by the pedicle wall and the surgeon’s tactile
feedback, are critical for achieving the improved clinical accuracy of image guidance
systems demonstrated in the literature.
Michael L. Swank1, MD, Crystl D. Willison2, MD, Kimberly Wilkens2, RN
Freiburg Spine Institute and 2Riverhills Neurosurgery, Cincinnati, OH, USA
Background: Pedicle screws as an adjunct to lumbar spine fusion have been controversial, at least in part because of the difficulty in accurately placing screws in the pedicle.
Several intraoperative navigational systems have been developed to help improve the
accuracy of operating in and around delicate neural and vascular tissues. The purpose of
this study was to evaluate prospectively the use of an intraoperative navigation system
(Radionics OTS™)for the placement of pedicle screws, as compared to a traditional
fluoroscopic-controlled method.
Design/Methods: An IRB-approved prospective study for the placement of lumbar
pedicle screws was performed. Screws were placed by a standard fluoroscopy-controlled
technique on one side and by a computer-assisted technique on the other side. Intraoperative anterior, posterior and lateral fluoroscopy was used after all screws were placed to
confirm proper screw placement. Most patients had laminectomies after the surgery to
directly evaluate intramedial corticle and inferior screw placement. Postoperative CT
scans were done and screw position was rated by independent radiologist blinded as to
which side the computer-assisted orthopaedic surgery (CAOS) technique was used.
Patients were also followed clinically.
Results: 26 patients were enrolled in the study. Seven patients who had a CT scan
preoperatively did not have the OTS used in surgery for a variety of reasons, including
computer system failure (1), instrumentation failure of an early spinous process clamp (2),
CT scan done improperly (1), scheduling conflict with a craniotomy (1), intraoperative
medical complications requiring aborted procedure (2). Of the 19 patients who completed
the study, 55 screws were placed by a CAOS technique and 66 screws were placed by
fluoroscopic technique. All screws were properly started within the pedicle. Four of 66
screws in the fluoroscopic technique group had pedicle wall violations; three lateral and
one medial. Two of the 55 screws in the OTS group resulted in a pedicle wall violation;
one medial and one lateral. Five of the 66 screws in the fluoroscopic group, and two of the
55 in the OTS group either violated the end plate or anterior vertebral body. One of the
36 screws in each group was identified as possibly placing a nerve at risk, and no screws
were felt to put a vessel at risk.
Conclusions: The overall screw malposition rate in this clinical series compares to other
reports in the literature. The OTS performed as well as or better than the fluoroscopic
technique in all instances, with half as many screws being misplaced in the OTS group as
the fluoroscopically-controlled group. Based on this data, we can state that the OTS not
only allows proper placement of the screws but also may improve the accuracy of
placement of screws in the clinical setting compared to a traditional fluoroscopic
Abstracts from CAOS/USA ’99
DM Kahler and K Mallik
Dept. of Orthopaedic Surgery, University of Virginia Health System, Charlottesville, VA
Objective: Computer-assisted surgical navigation enables accurate placement of guide
wires and cannulated screws into the pelvis using percutaneous technique, without
intraoperative radiographic guidance. Early experience with this technique is compared to
conventional fluoroscopic technique to determine its relative merit with regard to reliability, operative time, and radiation exposure to both the patient and surgeon.
Background: Fluoroscopically guided cannulated iliosacral screw placement into the S1
body has gained popularity for treatment of traumatic posterior pelvic ring disruptions.
Unfortunately, this technique may be hindered by inadequate visualization of bone landmarks
due to patient obesity or retained intra-abdominal contrast media. Computer assisted technique
can potentially decrease operative time, eliminate failures due to poor fluoroscopic visualization, and decrease the need for intraoperative radiography, while providing equivalent or better
accuracy than fluoroscopic technique.
Design/Methods: Between February 1997 and May 1999, 20 posterior pelvic ring disruptions in 13 patients were stabilized with 1 or 2 7.3-mm cannulated screws using computerassisted technique. Injuries consisted of 14 sacroiliac joint disruptions, 6 sacral fractures, and
1 crescent fracture/dislocation. Most patients underwent preliminary pelvic external fixation as
part of their initial trauma resuscitation. Preoperative CT was obtained with the fixator frame
and fiducial array in place to allow segmentation for construction of a virtual model. CT data
was loaded onto an integrated system consisting of a work-station, digitizer, and CCD camera
boom (Stealth Station). Optimal entry and target points for screw placement were defined as
a surgical plan and saved on the workstation. Following intraoperative point-based registration, an optically tracked drill guide was used to place the guide wire and cannulated screws
into the safe zone in the S1 body. Limited intraoperative fluoroscopy was used to verify
registration and confirm accurate screw placement. Operative time per screw and total
fluoroscopy time was prospectively monitored. In the 3 years prior to this study, traditional
fluoroscopic iliosacral screw fixation was attempted in 16 posterior pelvic ring disruptions, but
was abandoned in 2 cases due to inadequate radiographic visualization. Total operative time
per screw and total fluoroscopy time for the other 14 cases was obtained from records. This
series was then compared to the computer assisted series in terms of accuracy, operative time,
and fluoroscopy time.
Results: Accurate screw placement was possible with both techniques. There was no
radiographic or clinical evidence of screw malposition in either series. Using computerassisted technique, 25 screws were successfully placed into the S1 body in 20 cases. Mean
total operative time was 24.6 min/screw (range 15–56 min). Mean fluoroscopy time for
validation of screw placement was 12.9 sec/screw (range 10 –28 sec). Using traditional
fluoroscopic technique, 18 screws were successfully placed into the S1 body in 14 cases.
Mean operative time was 33.8 min/screw (range 23– 61 min). Mean fluoro-scopy time was
74.1 sec/screw (range 63–104 sec). Computer-assisted technique required on average 58
seconds less fluoroscopy time and 9 minutes less operative time per screw than conventional technique (p ⬍ .05).
Computer assisted
No. of screws
33.8 minutes
24.6 minutes
74.1 seconds
12.9 seconds
Conclusions: Use of computer-assisted technique in treating posterior pelvic ring disruptions significantly decreases radiation exposure without increasing operative time.
Exposure with a portable C-arm amounts to 4 rads/min for pelvic imaging, while a routine
chest radiograph requires ⬃10-15 millirads. Computer-assisted technique results in a
radiation cost saving per patient of ⬃250 routine chest radiographs. Though poorly
defined, the risk of malignancy related to radiation exposure in orthopaedic procedures is
real. Despite the need for specialized equipment, this technique appears to offer significant
advantages over conventional fluoroscopic technique.
R. Bächler, R. Hofstetter, M. Slomczykowski, D. Schlenzka, M. Sati, L.-P. Nolte
Maurice E. Müller Institute for Biomechanics, Bern, Switzerland
Objectives: We present a novel computer-based technique for spinal surgery. It combines intraoperative fluoroscopy-based imaging with modern surgical freehand navigation.
The system requires no preoperative CT examination, and intraoperative radiation exposure is significantly reduced. The system has been evaluated in a laboratory setting as well
as in the operating theater.
Background: Transpedicular implant systems are used for rigid segmental fixation for
a wide variety of indications. Previous studies have shown the incidence of incorrectly
placed transpedicular screws to range from 10% to 41%. Additionally, there is considerable radiation exposure for both the patient and the surgical staff. In the past few years,
CT-based CAS systems have been introduced to further enhance accuracy and safety of
the procedure. Despite the many advantages of this technique, e.g., major reduction in the
misplacement rate, criticism has focused on the limited medical benefit for lumbar cases
with a relatively normal morphology, costs for the additional tomographic examination,
and the additional time spent intraoperatively.
Design/Methods: For the placement of transpedicular screws, 1 lateral and 2 obliqueregistered fluoroscopic images are acquired. The complete navigation is then based on
these images and no further radiation is required. To prepare the pedicle, the tip of the
instrumented pedicle awl is positioned in the middle of the pedicle oval and its axis is
aligned with the pedicle axis using the corresponding oblique view. The cranio-caudal
angulation of the awl is controlled on the lateral image of the vertebra. After perforating
the cortex, the screw canal is prepared using the instrumented pedicle probe. Finally, the
pedicle screw may be delivered with a standard or instrumented screwdriver.
In a laboratory setup, 40 pedicle screws were inserted into 4 plastic lumbar spine
models (L1-L5), 20 simulating an open and 20 simulating a per-cutaneous surgical
approach. Then, 3 human lumbar spines (L1-L5) were used for in vitro testing. The
dorsolateral spinal musculature was left intact, thus simulating a percutaneous approach. A total of 30 screw holes were prepared, and a 4.0-mm aluminum cylinder was
inserted into each hole. Each pedicle was carefully sectioned perpendicular to the
screw axis, and each section was X-rayed. A circle with a diameter of 6.0 mm was
projected concentrically onto the shape of the cylinder for classification.
The system was evaluated clinically at the ORTON Orthopaedic Hospital in
Helsinki. For safety, a software module was developed to allow both the fluoroscopy
and CT-based systems to run simultaneously. The latter was used as a reference. Three
patients diagnosed with lumbar spinal stenosis and idiopathic thoracic scoliosis were
chosen for this study. Open insertion of 11 pedicle screws was planned at levels
ranging from T11 to L2.
Results: Gross visual inspection of the plastic spines showed all pedicle screws to be
located within the pedicles without perforation. 122 histological sections from 30 pedicles
were analyzed for the in vitro study. An ideal position of the aluminium cylinder was
found in 101 pedicle sections. In 17 sections the pedicle cortex was touched by the screw
template. Cortex engagement of no more than half the pedicle cortex’ thickness was
observed in 4 sections. The histological examination showed no cortex perforation.
For the OR evaluation, the entry points found with the fluoroscopy based system were
accepted in 9 out of 11 cases. In 2 cases, the entry point was shifted by 2 mm following
the CT based system. The screw trajectories were accepted in 10 out of 11 cases.
Postoperative CT scans, acquired with special focus on the pedicle region, showed the
final positions of all pedicle screws to be in full agreement with the data from the CT based
system. No cortex perforations were observed.
Conclusions: The results of the in vitro study compare well to results from similar
studies using the CT based system. All screws could be inserted with no cortex perforation. The clinical trial supported the proof of concept, where 11 pedicle screws could be
placed with only minor problems due to image quality in one case. However, the small
number of patients does not allow a final assessment and comparison to CT based systems
at this point in time.
Abstracts from CAOS/USA ’99
Neil Glossop1, Richard Hu2, Dominic Young2, Gary Dix2, and Yaser Behairy2
Traxtal Technologies, Bellaire, Texas, USA
Division of Orthopaedic and Neurosurgery, Foothills Hospital, University of Calgary,
Calgary, Alberta, Canada.
Objectives: To investigate non-fluoroscopic techniques for performing percutaneous
registration and placement of screws in the spine and pelvis.
Background: In spine surgery, the use of CAS systems is often limited to open surgery
in which extensive exposure of the vertebrae is required to register. In pelvic trauma, it is
possible to percutaneously access anatomical landmarks to register, but these must be
distributed over the pelvis. We investigated new registration techniques to decrease the
invasiveness of registration and the morbidity of CAS-based screw fixation.
Design/Methods: Two methods were used to register the spine. The first is a percutaneous matching technique. Stab incisions were made posterolaterally, and locations along
the transverse processes and medial pars were selected by probing through the skin to
perform a paired point matching. Surface fitting was then used to assist and improve this
initial registration.
A second registration method that used a radiolucent, trackable fiducial carrier was
developed for use in both the spine and pelvis. The carrier was attached by a screw to the
vertebra or pelvis fragment prior to the scan. The carrier contained fiducials that were
imaged on the preoperative CT but also functioned as a dynamic reference base (DRB)
which could be tracked by an optical position sensor, so the positions of the fiducials was
known at all times. By selecting these as registration points, it was possible to register the
carrier (and the attached bone) semi-automatically. Since the “patient-space” positions of
the fiducials were accurately known a priori from the carrier’s construction, it was never
necessary to touch the points with a probe. Our software processed the data to obtain an
accurate registration prior to surgery, enabling the surgeon to proceed without performing
the matching step.
Results: Our early results have demonstrated percutaneous registrations using bony
landmarks is an effective technique in the spine, with 4/6 pedicle screws being placed
successfully in the cadavers. One miss was attributed to DRB motion after registration,
and the second accidentally targeted the wrong vertebral segment. Both misses could have
been avoided with the better technique that is already built into the CAS system. The
iliosacral and anterior column screws as well as 4 additional pedicle screws were
successfully placed using the fiducial marker carrier. Registrations using this method
tended to be simple and of high quality, with low RMS values reported from the system
and excellent correlation with the surface.
Conclusions: Both solutions have proved to be able to cope with the requirements of
registration. The trackable fiducial carrier was felt to be more accurate and convenient for
the surgeon. The additional surgery required for implantation of the carrier in the spine
may be justified if it can be shown that there is improved patient recovery and a more
accurate registration.
Use of the methods for spinal fixation is practical in their current form and especially
when an external pelvic fixator is used, as in trauma cases. It will be some time before
percutaneous spinal stabilization becomes realistic in all situations however. Despite the
progress that has been made, we urge caution as these methods are still evolving. We are
currently working to improve the instrumentation used.
Richard D. Bucholz, M.D., F.A.C.S.
Bakewell Section of Image Guided Surgery, Division of Neurological Surgery, St. Louis
University School of Medicine, St. Louis, Missouri.
The first application of image guidance occurred in neurosurgery in the form of stereotactic surgery. The interest of neurosurgeons in image guidance is due to the nature of the
procedures they perform, which are characterized by high risk due to the delicacy and
importance of the targeted organ. Image guidance now represents the standard of care for
cranial procedures.
Navigational systems for neurosurgery have become commonplace and are in their
second generation of development. A good example is the StealthStation by Sofamor
Danek, a division of Medtronic, Inc., of Minneapolis, with over 350 systems installed
world wide, and over 6,000 procedures performed. This system, developed in conjunction
with St. Louis University, consists of a high-speed graphics computer, an optical camera
system consisting of two cameras sensitive to infrared light, and instruments modified to
either emit or reflect infrared light. The instruments used in this system are used to track
the surgeon’s movement as well as devices that attach to the patient’s body (termed
reference arcs or arrays). By using both forms of instrumentation simultaneously the
system correlates the position of the patient with the position of the surgical instrument.
The result of this correlation is displayed on a monitor in the operating room.
In spite of their commercial success, these systems have been approved for marketing
in the US only in the past four years, and can therefore be considered to be still in their
developmental infancy. As all image guided systems use the same steps to couple imaging
to surgery, the components of every image-guided system are therefore familiar. Therefore reviewing the progress occurring in neurological systems can be useful in determining the future of systems optimized for orthopedic surgery. The components of image
guidance comprise the following:
Preoperative imaging
Image registration
Intraoperative tracking
System control
Intraoperative imaging
Outcome analysis
Each component will be described and discussed in this talk in the order they are
encountered in a standard image guided intervention. Possible improvements in the
components, with focus on applications within orthopedic surgery, will be touched upon,
and the utility of systems for orthopedic applications will be emphasized. The utility of
using two-dimensional images for guidance instead of 3D images as commonly employed
for neurosurgical interventions will be considered given the financial realities of orthopedic interventions.
Russell Taylor, PhD.
NSF Engineering Research Center for Computer-Integrated Surgical Systems and
Technology, Johns Hopkins University, Baltimore, MD
The impact of Computer-Integrated Surgery (CIS) on medicine in the next 20 years will
be as great as that of Computer-Integrated Manufacturing on industrial production over
the past 20 years. A novel partnership between human surgeons and machines, made
possible by advances in computing and engineering technology, will overcome many of
the limitations of traditional surgery. By extending human surgeons’ ability to plan and
carry out surgical interventions more accurately and less invasively, CIS systems will
address a vital national need to greatly reduce costs, improve clinical outcomes, and
improve the efficiency of health care delivery. As CIS systems evolve, we expect to see
the emergence of two dominant and complementary paradigms: Surgical CAD/CAM
systems will integrate accurate patient-specific models, surgical plan optimization, and a
variety of execution environments permitting the plans to be carried out accurately, safely,
and with minimal invasiveness. Surgical Assistant systems will work cooperatively with
human surgeons in carrying out precise and minimally invasive surgical procedures.
This talk will focus on the possible evolution of Surgical CAD/CAM and Assistant
systems in orthopaedic surgery, with a particular emphasis on the relation between current
and future surgical applications and key enabling engineering research problems.
Abstracts from CAOS/USA ’99
Jon C. Bowersox, MD, PhD
University of California, San Francisco, California, USA
The feasibility of remote, robotic-assisted surgery was first demonstrated in animal studies
less than 4 years ago. Using a prototype system developed at SRI International, surgeons
sitting at a console viewed a stereoscopic display of the remote operative field, and
grasped standard surgical instrument handles. The remote worksite had instrument tips
positioned such that they appeared to extend from surgeons’ hands in a natural orientation.
Unlike robotic positioning devices, in which laparoscopic cameras can be steered by
mechanical arms, telemanipulators translate hand motions to a distant field, thus truly
enabling remote surgery. The precision and accuracy of this system was demonstrated in
swine by performing repairs in femoral arteries, aortic interposition grafting, cholecystectomies, and repair of simulated injuries of the gastrointestinal tract. Unlike laparoscopic
surgery, no special training was needed. Maintaining natural eye-hand orientation (oculovestibular axis), using standard instruments, and providing force feedback all contributed to the natural, immersive performance using the SRI system. Surgeons described the
system as intuitive, not requiring specialized skills or training, as is common in laparoscopic surgery. The naturalness of the experience has been termed “telepresence” by the
engineers and scientists who developed the technology.
Among the many possible uses of telepresence surgery, the potential to restore intuitive
feel to minimally invasive surgery is the most exciting, at least in the near future.
Advanced laparoscopic procedures have been limited by the difficulties working with
counter-intuitive instruments, with a loss of stereoscopic visual cues, dampening of the
haptic senses of touch and feel, and restricted degrees-of-freedom of movement. The
potential value of working through trocar access ports with the same dexterity as
conventional, open surgery is tremendous. The feasibility of using a telepresence surgery
workstation coupled with instrument tips inserted through 10-mm trocars was demonstrated by surgeons performing end-to-end vascular anastomoses of 6-mm PTFE grafts,
with significant performance improvement over that achieved using standard laparoscopic
techniques. No special training or practice was required to achieve technically adequate
Potential Applications of Telepresence Surgery
When the surgeon is separated from the operative field by
Physical Barriers: abdominal wall, laparoscopic surgery
Size: dexterity limitations, microsurgery
Hazardous environments: biological or nuclear contamination, BL4 containment,
Distance: remote battlefield surgery
Training: simulation interface
The telepresence surgery operator’s workstation has been coupled with micro-manipulators and an operating microscope to perform 1-mm diameter arterial anastomoses in rats,
and to fiberoptic endoscopes for remote diagnostic evaluation. A most exciting area of
development is displaying a simulated surgical scene to surgeons at the operator’s
workstation. By manipulating the simulated environment, surgeons can train in a practice
environment with the same interface that will be used to complete an actual surgical
procedure. The potential for practicing extremity wound debridement has been demonstrated, and incorporation of ATLS-training is anticipated in the near future.
The first clinical results using a surgical telemanipulator system for minimally invasive
surgery were reported from France, where six patients underwent cardiac surgery using
6-degree-of-freedom telemanipulator arms introduced through 10-mm trocars. Although
all six patients had their procedures completed using conventional techniques, none had
a sternotomy, and no mechanical or technical limitations were noted.
Potential applications of telemanipulators other than cardiac surgery include minimally
invasive spine surgery, neurosurgery, gynecology, and vascular surgery. Challenges that
remain to be addressed before surgical telemanipulators become widely used include cost
(anticipated at nearly a million dollars for a single system), and efficiency.
R.J. Bale2, C. Fink1, R. Rosenberger, W. Hackl1, C. Hoser1, M. Rieger2, W. Jaschke2, K.P.
Deptartment of Traumatology and 2Department of Radiology, University Hospital Innsbruck, Austria
Objectives and Background: Open treatment of osteochondral lesions of the talus
usually requires extensive arthrotomy and dissection. Our purpose was to develop a new
minimally invasive method for accurate targeting of bone lesions of the extremities using
3D navigation. We have therefore designed a foam immobilization technique for the
extremities. The accuracy of pin placement was evaluated in ten cadavers, and our first
clinical experiences with the new method are presented.
Design/Methods: We used two different frameless stereotactic navigation systems in
combination with a new foam immobilization technique and a targeting device similar to
the EasyTaxis (Philips). The EasyGuide (Philips) was used for the first 5 and the Stealth
Station (Sofamor Danek) for the remaining 5 cadaver experiments and the first three
patients. The lower extremity is immobilized with foam sparing an 8 ⫻ 8 cm area for the
entrance of the pin(s). The immobilized extremity is rigidly mounted to the base plate.
Radio-opaque fiducials are glued to the foam and a CT scan is performed. The dataset is
transferred to the navigation system in the simulation room via local network. Entrance
and target point are determined on the 3D navigation system. After using the reference
points on the foam for registration, the targeting device is adjusted with the aid of the
“navigation views” of the frameless stereotactic system. The length of the pin is measured
on the monitor. Then, the pin is advanced through the preset aiming device as far as the
clamp allows. In the cadavers, the position of the pin tip was evaluated by anatomical
preparation. In the three patients, the pin position was confirmed with intraoperative
fluoroscopy and postoperative CT scan, with which the drill hole could be identified.
Results: In 10 cadavers and three patients, the pin was placed within the range of 1–2.5
mm at the first attempt. There was no significant difference in accuracy between the two
different navigation systems.
Conclusions: Our first cadaver experiences and our first clinical applications indicate
that frameless stereotactic navigation systems allow accurate, preplanned targeting of
bone lesions for various purposes. The new technique was helpful in localizing osteochondral lesions in the talus, thus reducing surgery time and invasiveness. The method
presented above can also be used for biopsy of bone tumors. Modifications of the
immobilization device are required in order to use our method in clinical routine.
Abstracts from CAOS/USA ’99
D. Glozman1, M. Shoham1, A. Fischer2
Robotics Laboratory and 2CMSR Laboratory of Computer Graphics and CAD, Dept. of
Mechanical Engineering, Technion–Israel Institute of Technology, Haifa, Israel
Objectives: Registration is a critical stage in robotic–assisted surgery, in which a
geometric relationship such as position and orientation of the patient’s bone relative to the
robot’s tools is established intra-operatively. Current registration techniques often need
implantation of artificial fiducial markers or digitizing devices such as optic or magnetic
sensors or laser scanners, which complicate the registration procedure.
The registration process proposed uses a surface matching technique, and thus requires
no marker implantation. Three ideas simplify the registration process: 1) the robot itself
is used as a digitizer, eliminating the need for an extra localizer; 2) bone modeling is based
on the multi-resolution technique for adaptive registration; 3) an algorithm to determine
the minimal number and location of sampled points needed for registration was developed,
thus easing the intra-operatively sampling process. The proposed method was applied to
the Total Knee Arthroplasty (TKA) procedure, and special care was taken in adapting the
method to the surgical application in hand.
Background: In TKA, the distal, femoral, and proximal tibial compartments are resected and replaced with 2 prosthetic components. About 200,000 cases are performed
annually in the US. Progress in implant design and surgical technique led to success rates
close to 85%. During the intervention, the articular surfaces of the femur and tibia are
replaced by 2 prosthetic components. Alignment errors of such components exceeding 1°
in orientation and 1 mm in position can severely affect the kinematic and kinetic
functionality of the operated limb and may eventually lead to implant failure. Therefore,
careful pre- and intra-operative planning is required. Use of robotic assistance during the
execution phase improves the absolute accuracy in positioning and guiding surgical tools.
Robotic execution of the planned bone resections can ensure further improvement of the
procedure because of the higher intrinsic geometric accuracy of a robot as compared to
that of a human operator. Based on prior experiments in orthopedic surgery, it is expected
that a robotic assistant will overcome implant misalignment which is the major cause for
aseptic loosening and failure in TKA.
Results/Methods: The robot used in the surgical procedures can also be used as a
digitizer, thus considerably reducing the potential inaccuracies in reference frame registration. Registration is then performed directly between the bone and the robot, which in
turn guides the surgical tools. To speed up the computation, we introduce the hierarchical
multi-resolution approach and use a level of detail data model.
To ensure the required accuracy of 1° rotation and 1 mm translation, we consider
optimal number of sampling points and their location. The search for the best
sampling points is viewed in terms of grasping theory. It is possible to view the
problem as that of grasping the bone with a multi-fingered hand. The contacts between
the fingertips and the grasped object are modeled as frictionless point contacts. Each
contact is modeled as a virtual linear spring directed normal to the surface passing
through the point of contact. In order to determine the optimal set of sampling points
for each set of points, a worst-case transformation is calculated. That is, we look for
the bone motion which is minimally detected by the sensor, and choose the set of
contact points that are maximally dislocated by the minimal motion. This configuration of the grasp gives the best stiffness properties; hence, the points of contact are the
best candidates for sampling during registration.
Conclusions: This technique for 3D data registration is specifically oriented towards
robot-assisted TKA, and is based on surface matching with ICP algorithm and on a
multi-resolution model. Sufficient registration accuracy could be achieved even without
the fully-detailed model. The robot can be used as a digitizer, since sampling with a robot
as a probe has many advantages: there is no need for an additional complex localizer or
an additional transformation of coordinates, and the calibration procedure is diminished.
Grasping theory was found useful in determining the minimal number and location of the
sampling points for a given registration accuracy.
Без категории
Размер файла
194 Кб
Пожаловаться на содержимое документа