SURGICAL TOOL ALIGNMENT BY LASER GUIDANCE
USING FLUOROSCOPIC-BASED NAVIGATION TECHNIQUE
A System Implementation and Validation Study
Jack T. Liang
1
, Shinya Onogi
2
and Yoshikazu Nakajima
1,2
1
Graduate School of Engineering, University of Tokyo, Yayoi 1-1-1, Tokyo, Japan
2
Intelligent Modling Laboratory, University of Tokyo, Tokyo, Japan
Keywords: Laser Guidance, Computer Aided Surgery, Fluoroscopic Surgery.
Abstract: This paper provides a novel method for intuitive and CT-less surgical navigation based on fluoroscopic-
based navigation and laser guidance technology for minimal invasive orthopaedic surgery. This method
does not require intra-operative registration of three-dimensional surface model derived from pre-operative
CT/MRI volumes and is able to project surgical path planed intra-operatively onto the patient's skin directly.
In this paper, implementation of this method and basic in vitro guidance accuracy validation were
performed. Tool insertion path planning was performed on three 2D images from a pinhole imaging source
taken at different incident angles. A 3D insertion pathway was generated and projected using two laser
beams. Our Fluorolaser system has a planning accuracy of 1.07±0.60 mm, 0.73±0.38 degrees and an overall
guidance accuracy of 1.11±0.62 mm, 0.80±0.68 degrees. These results demonstrate that the proposed
method has great potentials to ensure accurate and intuitive surgical procedures.
1 INTRODUCTION
Fluoroscopy is commonly used for real-time
viewing of patient anatomy during percutaneous or
less invasive orthopaedic surgical procedures. It
provides real-time 2D views using a fluoroscope.
However, its usage comes at the expense of
increased exposure to radiation; difficulties in
obtaining and processing multi-planar visualization;
and the mental burden for surgeons to interpret 3D
positions from 2D images.
To address the visualization problem, X-ray
computed tomography (CT)-based and magnetic
resonance (MR)-based are developed to simplify
surgical procedures. However, CT-based navigation
has large amounts of radiation exposure while MR-
based navigation does not provide high bone
intensity resolution and can only be used in non-
metal containing environments. To address radiation
exposure, X-ray fluoroscopic navigation, or virtual
fluoroscopy, has been developed to guide surgical
tool placement (Foley, 2000; Hofstetter, 1999;
Leloup, 2008). However, these systems superimpose
surgical tool location over pre-acquired x-ray images
and displays navigational plans via a secondary
display device. This unintuitive guidance method
forces surgeons to mentally calculate real-world tool
locations from a display panel during surgery.
Laser-based navigational guidance systems have
been developed for spinal pin insertion and pelvic
implant fixation surgeries (Sasama, 2002). It
generates surgical plans and uses two laser beams to
project point-of-entry and insertion orientation
directly onto the patient’s skin (Nakajima, 2004).
Thus, a direct guidance is achieved and surgeons no
longer need to mentally interpret positioning
information from a screen. However, CT or MR-
image is required for surgical planning.
This paper presents an improved system that
employs two laser sources to guide surgical tool
insertion using surgical planning from two or more
fluoroscopic X-ray images taken intro-operatively.
In particular, we elicited the virtual fluoroscopic
planning technique. As far as the authors know, this
is the first study to integrate the virtual fluoroscopic
planning technique with the dual-laser guidance
technique for practical manual execution of
surgeries. This novel combination yields an original
surgical navigation and guidance system that
displays tool insertion path directly in the surgical
field without use of any pre-surgical techniques. We
showed the feasibility of this method with a basic
system validation using a pin-hole camera.
319
T. Liang J., Onogi S. and Nakajima Y..
SURGICAL TOOL ALIGNMENT BY LASER GUIDANCE USING FLUOROSCOPIC-BASED NAVIGATION TECHNIQUE - A System Implementation and
Validation Study.
DOI: 10.5220/0003763903190323
In Proceedings of the International Conference on Biomedical Electronics and Devices (BIODEVICES-2012), pages 319-323
ISBN: 978-989-8425-91-1
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
2 MATERIALS
The proposed fluoroscopic-based dual-laser guided
(Fluorolaser) navigation system consists of three
major components: a Laser Guidance System
(Nakajima, 2004) with built-in tracking capabilities
using the hybrid Polaris Spectra (Northern Digital,
Waterloo, ON, Canada); active infrared markers
(AdapTrax 15 Marker, Northern Digital, Waterloo,
ON, Canada); and a web camera (P1C30 series,
Elecom, Osaka, Japan) used in place of an X-ray
fluoroscope. We implemented the fluorolaser
software using C++ (Visual Studio 2005, Microsoft
co.) with OpenCV library. It connects to the laser
guidance system over TCP/IP and to the Polaris
tracking device via RS-232c. The two functional
parts of this new system, guiding and planning, will
be discussed in this section.
2.1 Insertion Guidance Using Lasers
The intersection of two emitted laser-beams
simultaneously guides surgical tool’s entry position
and its inserting orientation as shown in Fig. 1. The
intersectional line l
0
made by laser planes S
1
and S
2
represents the insertion path in 3D space. The
insertion point p
0
is the intersection of l
0
and the
patient’s body surface S
0
. Insertion point guidance is
robust to deformations of the body surface,
represented by changes from S
0
to S
0
in Fig. 1.
Surface S
0
represents any deviations from S
0
in both
height and shape due to body surface relocation,
deformation or obstruction by soft tissues. For
proper surgical tool alignment, the tip of the surgical
tool is placed at p
0
and the length aligned to l
0
.
Orientation alignment can be verified using the
projections of the lasers on the side of surgical tools.
Proper alignment is indicated by two parallel lasers.
Figure 1: Left: Schematic diagram of the laser guidance
method. Right: Robust entry point displayed for surface
relocation, deformation or occulsion by soft tissue using
dual laser-beams.
2.2 Navigational Planning
Infrared markers are placed on the imaging device
and on the surgical object to create the localizer-
fluoro coordinate system (CS
F
) and localizer-subject
coordinate system (CS
S
), respectively. The surgical
planning process requires multiple sets of inputs.
Each set consists of three pieces of information
obtained from the surgical environment: an image of
the surgical object and the insertion site obtained by
the imaging device and the three-dimensional
coordinates of the two markers when the image is
taken. For each set of inputs, a two-dimensional
surgical tool insertion path is manually drawn onto
the acquired two-dimensional image. Using camera
calibration parameters, this 2D plan is projected
along the ray-lines of the imaging device to obtain a
plane in 3D space. The plane is formulated in CS
F
and transformed for use in CS
S
.
Following Fig. 2, two 2D plans from two sets of
inputs yields two planes in 3D space. The
intersection of two planes yields a line in 3D space
where the directional vector of the line is the cross
products of the normal of the two planes. This
intersection line represents the insertion path in 3D
space. If three sets of inputs are used, three 2D plans
are made to give three planes in 3D space. This
gives rise to three intersection lines in 3D space.
Regression of these lines is performed to obtain a
line of best fit, which represents the optimal surgical
insertion path. This optimal insertion path is stored
in CS
S
and thus is robust towards any slight
movements of the surgical object.
The optimal path is then transformed into the
localizer coordinate systems (CS
L
). Laser guidance
is realized by using two laser sources to project the
mathematical representation of this insertion path.
3 METHODS
We validated the Fluorolaser system using a pinhole
web-camera, which is mathematically similar to a
fluoroscope (Navab, 1999). We prepared a simple
validation setup using a flat surface and surgical
insertion tool, the K-wire (150 mm length, 3 mm
diameter). Guidance accuracy was validated by
tracking the position of the K-wire, which
represented the desired surgical tool insertion path
and the surgical tool itself. An active infrared marker
and a guidance sleeve were attached for position
tracking and alignment, respectively. Proper tool
orientation alignment was indicated by parallel laser
beams on the guidance sleeve, as stated previously.
3.1 Experimental Design
We created a virtual coordinate system on a flat
BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices
320
Figure 2: Fluorolaser transformation
matrices: localizer to laser 1 and laser
2, localizer to subject, and localizer to
fluoroscope. Fluoroscope is positioned
at two arbitrary locations. Two 2D
plans are used to calculate insertion
path in 3D space.
Figure 3: Experimental setup includes
the laser guidance system, a flat
surface, a web camera (fluoroscope
imitation device) and a K-wire.
Location markers are used to locate
imaging device, K-wire and surface.
Flat surface represents the surgical
object.
Figure 4: Black figure represents the
initial K-wire position; white figure
represents the final K-wire position;
and green line represents the planned
surgical insertion path in 3D space.
surface (Fig. 3) and selected the z-axis as the
direction normal to the surface. We used a four
factorial Box-Behnken experimental design method
to create 28 validation trials by altering the intended
insertion path mathematically. Its factors and levels
are: x-axis (-50, 0, 50 mm), y-axis (-100, 0, 100 mm),
polar angle (-30, 0, 30 degrees), and azimuthal angle
(-30, 0, 30 degrees). Experiments were conducted
for both two-image and three-image configurations
for planning.
3.2 Experimental Protocol
First, the intended insertion path is shown by the
dual-lasers in 3D space. This ground truth path is
represented mathematically and altered according to
the experimental design. The K-wire was placed
according to the lasers. Using this initial k-wire
position, 2D images were taken for planning. The K-
wire was removed while planning was performed.
Once the 3D insertion path was generated, the laser
was turned on. The K-wire was repositioned using
guidance by laser. Once lasers became parallel, final
position of the K-wire was recorded.
3.3 Analysis
Fluorolaser accuracy validation looked at the
displacement of relative tip locations between
planed 3D insertion path, initial K-wire position and
final K-wire position. We calculated the difference
in tip positions in a plane that is perpendicular and
Figure 5: Positional and orientational error values shown
in blue and red, respectively. Results include accruies for
2-image planning, 3-image planning, and 3-image
guidance.
intersects the initial K-wire tip (Fig. 4). Planning
accuracy refers to the error between actual insertion
path and user-determined insertion path. Guidance
accuracy refers to the error of the overall system.
Transformation matrix between K-wire tip and
marker was pre-obtained.
4 RESULTS
Results of 28 validation experiments are shown in
Fig. 5. System planning accuracy using two 2D
images for planning was 2.08±1.47 mm, 0.68±0.37
deg (degrees). Planning accuracy after linear
regression of three 2D images was 1.07±0.60 mm,
0.73±0.38 deg. Positional accuracy is significantly
improved using three images (P-value: 2.00x10
-6
).
SURGICAL TOOL ALIGNMENT BY LASER GUIDANCE USING FLUOROSCOPIC-BASED NAVIGATION
TECHNIQUE - A System Implementation and Validation Study
321
Using three-images, max planning error was
found to be 2.22 mm, 1.85 deg. The RMS and
RMSE were 1.22 mm, 0.82 deg and 0.59 mm, 0.37
deg, respectively. The overall guidance accuracy
using three 2D images was 1.11±0.62 mm,
0.80±0.68 deg. RMS and RMSE were 1.27 mm,
1.05 deg and 0.62 mm, 0.67 deg, respectively.
Box-Behnken surface responses for planning and
guidance errors were desirable for the 28 different
four-factor three-level combinations. There was no
correlation between the system accuracies and the
positional changes of imaging device.
5 DISCUSSION
Conventional navigation systems use a secondary
viewing device to indirectly guide surgery. These
operations are often mentally challenging and
cumbersome to perform. The Fluorolaser system
guides surgeries directly using two lasers to guide
surgical tool alignment. Planning is performed using
two or more images taken intra-operatively as
oppose to complete CT or MRI registration data.
The entire surgical navigation process using the
Fluorolaser system consists of two types of errors:
planning error and guidance error. The latter has
been validated previously (Lim, 2010). The former
includes camera calibration error, and human errors
for planning and re-positioning. In this study, we
have evaluated planning and guidance errors of the
system. Planning accuracy is limited by the camera
calibration technique as well as user-based insertion
path planning. Furthermore, this validation study
was performed with the assumption of point-
accuracy, which cannot be obtained from the K-wire
tip or the crosshair created by the lasers. Generally,
navigational systems designed for precutaneous
surgeries using two-dimensional images are
considered precise and accurate when position and
orientation errors are within 2 mm and 2 deg,
respectively. However, acute accuracies may be
required for special types of surgeries such as
surgeries in the brain and neck regions.
The presented evaluation of the Fluorolaser
system may be limited. In our experiments, we could
not use fluoroscope images and instead used normal
camera images. Although different in the imaging
techniques used, the system is invariable for both
camera and x-ray images since both devices are
theoretically assumed to be pinhole devices (Navab,
1999; 2000). Validations with x-ray images are
necessary for actual surgeries.
Although this system is an improvement to
conventional systems, clinical assessment is
necessary to fully validate its robustness in actual
surgeries. We intend to further validate this system
using fluoroscopes and saw-bones or biological
specimens to better mimic the surgical environment.
We are also considering the implementation of
insertion depth guidance. The lack of the depth
guidance may result in the need to acquire additional
X-ray images or to use positioning trackers. We plan
to introduce a physical stopper attachment for the
surgical tool; however, problems may arise when
during skin deformation. Adequate implementation
of depth guidance using laser-beams may overcome
these problems as well as expand the applicability of
the system in different types of surgeries.
6 CONCLUSIONS
We have proposed a novel integration of the
fluoroscope-based navigation and the laser guidance
systems. This Fluorolaser system provides intuitive
surgical tool insertion positioning in precutaneous
surgeries without pre-operative CT/MRI volumes.
Specifically, in vitro insertion path planning was
performed using 2D images from a pinhole imaging
source and insertion guidance was directed by lasers.
The planning accuracy was 1.07±0.60 mm,
0.73±0.38 deg and overall guidance accuracy was
1.11±0.62 mm, 0.80±0.68 deg. This demonstrates
the potential of the Fluorolaser system to be used in
accurate and intuitive surgical procedures without
registration of pre-operative CT/MRI volumes.
REFERENCES
Foley, K T., et al., 2000. Virtual Fluoroscopy. Oper. Tech.
in Ortho, 10(1), pp. 77.
Hofstetter, R., et al., 1999. Fluoroscopy as an imaging
means for computer-assisted surgical navigation.
Comp. Aid Surg, 4, pp. 65.
Leloup, T., et al., 2008. A Novel Technique for Distal
Locking of Intramedullary Nail Based on Two Non-
constrained Fluoroscopic Image and Navigation. IEEE
Trans on Med Img, 27(9), pp. 1202.
Sasama, T., et al., 2002. A Novel Laser Guidance System
for Alignment of Linear Surgical Tools. LNCS 2489
MICCAI’02, pp. 125.
Nakajima, Y., et al., 2004. Available range analysis of
laser guidance system and its application to monolithic
integration with optical tracker. International
Congress Series, 1268, pp. 449.
Lim, S., et al., 2010. Assessment for the feasibility of
external-fixation pin guidance using laser navigation.
Jap. Soc. Of Comp. Aid Surg. 12.
BIODEVICES 2012 - International Conference on Biomedical Electronics and Devices
322
Navab, N., et al., 1999. Camera-Augmented Mobile C-arm
Application. LNCS 1679 MICCAI’99, pp.688.
Navab, N., et al., 2000. Visual Servoing for Automatic and
Uncalibrated Needle Placement for Percutaneous
Procedures. IEEE CVPR’00, 2, pp. 1063.
SURGICAL TOOL ALIGNMENT BY LASER GUIDANCE USING FLUOROSCOPIC-BASED NAVIGATION
TECHNIQUE - A System Implementation and Validation Study
323