`
` Navigation in Neurosurgery
`
`Ehsan Azimi (1), Jayfus Doswell (2), Peter Kazanzides (1)
`
`(1) Dept. of Computer Science, Johns Hopkins University
`
`(2) Juxtopia, LLC
`
`ABSTRACT
`
`image-guided
`in
`is crucial
`identification
`tumor
`Precise
`neurosurgical procedures. With existing navigation systems,
`the surgeon must turn away from the patient to view the
`imaging data on a separate monitor. In this study, an
`innovative system is introduced that illustrates the tumor
`boundaries precisely augmented on the spot where the tumor
`is located with regard to the patient. Additionally, it allows the
`surgeon to track the distal end of the tools contextually, where
`direct visualization is not possible. In this approach, the
`tracking system is compact and worn by the surgeon,
`eliminating the need for additional devices that are bulky and
`typically limited by line of sight constraints.
`
`KEYWORDS: Augmented reality, HMD, neurosurgery,
`surgical navigation
`
`1
`
`INTRODUCTION
`
` Surgical resection is one of the most common treatments for
`brain tumors. The treatment goal is to remove as much of the
`tumor as possible, while sparing the healthy tissue. Image
`guidance (e.g., with preoperative CT or MRI) is frequently
`used because it can more clearly differentiate diseased tissue
`from healthy tissue. Most image guidance devices contain
`special markers that can be easily detected by the tracker.
`Registering the tracker coordinate system to the preoperative
`image coordinate system gives the surgeon “x-ray vision.”
`
`It can, however, be challenging to effectively use a navigation
`system because the presented information is not physically co-
`located with the operative field, requiring the surgeon to look at
`a computer monitor rather than at the patient. This is especially
`awkward when the surgeon wishes to move an instrument
`within the patient while observing the display.
` Such
`ergonomic issues may increase operating times, fatigue, and the
`risk of errors. Furthermore, most navigation systems employ
`optical tracking, due to its high accuracy, but this requires line-
`of-sight between the cameras and the markers in the operative
`field, which can be difficult to maintain during the surgery.
`
`(1) 3400 N. Charles St., Baltimore MD, {pkaz, eazimi1}@jhu.edu
`(2) 1101 E 33rd St, B304, Baltimore MD, jayfus@juxtopia.com
`
`IEEE Virtual Reality 2012
`4-8 March, Orange County, CA, USA
`978-1-4673-1246-2/12/$31.00 ©2012 IEEE
`
` After observations of surgeries, particularly neurosurgeries,
`and discussions with surgeons, we identified a need to overlay
`a tumor margin (boundary) on the surgeon’s view of the
`anatomy. It was also desired to correctly track and align the
`distal end of the surgical instruments with the preoperative
`medical images. The aim of this paper is to investigate the
`feasibility of implementing a head-mounted tracking system
`with an augmented reality environment to provide the surgeon
`with visualization of both the tumor margin and the surgical
`instrument in order to create a more accurate and natural
`overlay of the affected tissue versus healthy tissue. It allows the
`surgeon to see the precise boundaries of the tumor for
`neurosurgical procedures, while at the same time providing
`contextual overlay of the surgical tools intraoperatively which
`are displayed on optical see-through goggles worn by the
`surgeon. This makes it feasible for augmented reality, as the
`overlay provides the most pertinent information without unduly
`cluttering the visual field. It provides the benefits of navigation,
`visualization, and all of the capabilities of the existing
`modalities and is expected to be comfortable and intuitive for
`the surgeon.
`
`The majority of related research has focused on augmented
`reality visualization with HMDs, usually adopting video see-
`through designs. Many of these systems have integrated one or
`more on-board camera subsystems to help determine head pose
`[1,2,3] and some have added inertial sensing to improve this
`estimate via sensor fusion [4,5,6]. None of these systems,
`however, attempt to provide a complete tracking system and
`continue to rely on external trackers. Other researchers [7]
`have implemented a video see-through augmented reality
`system that also includes a magnification capability, using both
`internal and external tracking systems. The internal tracking
`system is a single camera to provide orientation. The external
`tracking system has 4 cameras, and tracks a pointer, reference
`frame, and the display position.
`
`2
`
`SYSTEM DESCRIPTION AND METHODS
`
`2.1
`
`System Description
`This section provides an overview of the prototype setup.
`As illustrated in Figure 1, the user wears the optical see-
`through goggles (Juxtopia LLC, Baltimore, MD) and a helmet
`that supports a compact optical tracking system (Micron
`tracker, Claron Technology, Toronto, CA). The first step is a
`registration procedure (Fig 1), in which the surgeon uses a
`tracked probe to touch markers that were affixed to the patient
`
`123
`
`Medivis Exhibit 1011
`
`1
`
`
`
`Calibration
`2.3
` Since the position, orientation and view angle of the user's
`eyes is different from those of the tracker, an additional
`calibration step is necessary to correctly reorient the image on
`the goggles' display. This step differs from user to user, and
`therefore a modified version of SPAAM technique [8] is
`implemented so that the user can complete this subjective task.
`
`3 CONCLUSIONS AND FUTURE WORK
`
`To the best of our knowledge this is the first time that a
`head mounted tracking, registration and display system are
`integrated for surgical navigation. This would reduce the line of
`sight problem (because the tracker’s line-of-sight is the same as
`the surgeon’s), the difficulty in association of preoperative
`images, and the bulkiness that exists in the current systems.
`
`This was a limited study intended to demonstrate the
`feasibility of the approach. The tracking system, which works
`by processing the markers, currently provides 20Hz update
`rates and cannot keep up with sudden head movements. To
`overcome this problem, we plan to add inertial sensing (e.g.,
`accelerometers, gyroscopes) to the tracking system both as a
`backup and as concurrent sensors, using a Kalman filter for
`sensor fusion to improve the robustness with respect to marker
`occlusions and compensate for the delay of the optical tracking
`system.
`
`In the future, the system should include magnification to
`add the capability of the existing pure optical systems (surgical
`loupes). We are also considering an eye tracking system for
`more accurate overlay positioning.
`
`ACKNOWLEDGMENTS
`
`We thank Dr. George Jallo for his clinical guidance and
`Kamini Balaji and David A. Bosset for their technical
`contributions. This work was supported by NSF IIP-0646587.
`
`References
`
`[1] H. Fuchs, M.. Livingston, R. Raskar, D. Colucci, K. Keller, A. State, J.
`Crawford, P. Rademacher, S. Drake, A. Meyer, “Augmented reality
`visualization for laparoscopic surgery,” MICCAI, pp. 934-943, 1998.
`[2] W. Hoff, T. Vincent, Analysis of head pose accuracy in augmented
`reality, IEEE Trans. Visualization and Comp. Graphics, Vol. 6, 2000.
`[3] F. Sauer, F. Wenzel, S. Vogt, Y. Tao, Y. Genc, A. Bani-Hashemi,
`“Augmented Workspace: designing an AR testbed,” Proc. IEEE Intl.
`Symp. on Augmented Reality (ISAR), 2000.
`
`[4] R. Azuma, G. Bishop, “Improving Static and Dynamic Registration in an
`Optical See-through HMD”, SIGGRAPH, 1994.
`
`[5] S. You, U. Neumann, R. Azuma, “Hybrid Inertial and Vision Tracking
`for Augmented Reality Registration,” Proc. IEEE Conf. on Virtual
`Reality, pp 260-267, Houston, TX, March 1999.
`[6] L. Chai, WA Hoff, T Vincent, “Three-Dimensional Motion and
`Structure Estimation Using Inertial Sensors and Computer Vision for
`Augmented Reality,” Presence, 11(5):474-492, Oct 2002.
`[7] A Martin-Gonzalez, S Heining, N Navab, “Head-mounted virtual loupe
`with sight-based activation for surgical applications”. ISMAR, pp 207-
`208, 2009.
`
`[8] M. Tuceryan, N. Navab, “Single-Point Active Alignment Method
`(SPAAM) for Optical See-Through HMD Calibration for Augmented
`Reality,” Presence: Teleoperators and Virtual Environments, Vol. 11,
`No. 3, Pages 259-276, June 2002.
`
`Figure 1. User wearing the optical see-through goggles
`and head mounted tracker. Figure shows the registration
`procedure, where the surgeon touches markers using a
`tracked probe.
`
`prior to the preoperative imaging. A paired point rigid
`registration technique computes the transformation that aligns
`the preoperative data (i.e., tumor outline) to the real world.
`
`After registration, the surgeon can see a registered preoperative
`model overlayed on the real anatomy using the optical see-
`through goggles (Figure 2, which illustrates the concept with a
`skull model, rather than the tumor margin that would be used in
`an actual surgery).
`
`SoftwareDesign
`2.2
`The software is written in C++, using a component-based
`architecture. It has three main interconnected components: 1)
`tracking module, 2) registration module, and 3) 3D graphical
`rendering. The 3D graphical rendering component uses the
`Visualization Toolkit (VTK). The transformation matrix
`obtained from
`the registration step
`is applied
`to
`the
`preoperative model in order to represent a correctly-posed
`scene in the tracker reference frame (Figure 2). The model is
`rendered after defining a VTK actor and camera. Based on the
`tracker information, the relative position and orientation
`between the Micron and the tracked object (i.e. the skull) is
`known. We then have to reorient the VTK camera by adjusting
`its parameters to put it in the same relative position to the actor
`as the tracker is with respect to the skull. This task is achieved
`by calculating the focal point, position, view angle, and view-
`up direction of the camera. The resulting image is illustrated in
`Figure 2.
`
`Figure 2. Graphical display. Right image shows real
`video output from one Micron camera, while the
`overlayed model of the skull, after registration, is
`depicted on the left.
`
`124
`
`Medivis Exhibit 1011
`
`2
`
`