‘Development and Validation of a Navigational Guidance
`System for Acetabular Implant Placement?‘
`
`D.A. Simon] '2, B. Jaramazm, M. Blackwellz, F. Morganz,
`A.M. DiGioia, M.D.,' 1 E. Kischellz, B. Colganl , T. Kanadez
`
`[Center for Orthopaedic Research, Shadyside Hospital Pittsburgh, PA 15232
`2Robotics Institute, Carnegie Mellon University, Pittsburgh, PA 15213
`
`Abstract. During the past year our group has been developing HipNav, a system
`which helps surgeons determine optimal, patient-specific acetabular implant
`placement and accurately achieve the desired implant placement during surgery.
`HipNav includes three components: a pre—operative planner, a range of motion
`simulator, and an intra-operative tracking and guidance system. The goals of the
`current HipNav system are to: l) reduce dislocations following total hip replace-
`ment surgery due to acetabular malposition; 2) determine and potentially increase
`the “safe” range of motion; 3) reduce wear debris resulting from impingement of
`the implant’s femoral neck with the acetabular rim; and 4) track in real—time the
`position of the pelvis and acetabulum during surgery.
`
`The original implementation of the HipNav system was a prooflof-concept pro-
`totype which was useful for demonstrating the efficacy of this technology in-vit-
`ro. As the HipNav system progressed towards a clinical implementation, our
`efforts focussed on several practical development and validation issues. This pa-
`per describes our experience transforming HipNav from a proof—of—concept pro-
`totype into a robust clinical system, with emphasis on technical development and
`Validation. Despite the highly applied nature of this endeavor, many fundamental
`research issues exist. The benefits of tightly coupling fundamental research to-
`gether with applied development in our work are discussed.
`
`Keywords: computer-assisted surgery, total hip replacement, navigational guid-
`ance, system validation.
`
`1
`
`Introduction
`
`Each year in the United States, approximately 200,000 primary and 40,000 revision to-
`tal hip replacement (THR) surgeries are performed. The most common early post-oper-
`ative complication following Tl-{R is dislocation of the femoral
`implant from the
`acetabulum, resulting in significant distress to the patient and surgeon, worse clinical
`outcome, and associated additional treatment costs. The approximate dislocation rate in
`the first year following primary THR is between 2 and 6 percent [10], and the most com-
`mon cause of early dislocation is malposition of the acetabular component I l l].
`
`The causes of dislocation following total hip replacement are multi-factorial and in-
`clude notonly malposition of the implants causing impingement, but also soft tissue and
`bone impingement, and soft tissue laxity l 1 ll. lmpingement between the neck of a fem-
`
`* This work was supported in part by a National Challenge grant from the NSF (IRI-9422734).
`Please address correspondence to the first author at das@ri.cmu.edu.
`
`EXHIBIT ,
`
` POIIN
`
`PENGAD800-631-6989
`
`
`
`err
`
`
`l\/lako
`
`Exhibit 1021 Pagel
`
`Mako Exhibit 1021 Page 1
`
`

`
`femoral implant
`
`x
`
`Fig. 1. Femoral and acetabular implants in impingement, and X—Ray of dislocation.
`
`oral implant and the rim of an acetabular implant is shown in Fig. l. impingement can
`lead to advanced wear of the acetabular implant rim resulting in polyethylene wear de-
`bris which may accelerate loosening of implant bone interfaces. The position of im-
`pingement is determined by implant design and geometry, and more importantly by the
`placement of the femoral and acetabular implants. In certain cases, impingement may
`result in dislocation, as seen in the X—ray of Fig. l.
`
`The HipNav system has been developed to permit accurate placement of the acetabular
`component during surgery [3]. HipNav includes three components: a pre—operative
`planner, a range of motion simulator, and an intra—operative tracking and guidance sys-
`tem. The pre—operative planner allows the surgeon to manually specify the position of
`the acetabular component within the pelvis based upon pre—operative CT images. The
`range of motion simulator estimates femoral range of motion based upon the implant
`placement parameters provided by the pre-operative planner. Feedback provided by the
`simulator can aid the surgeon in determining optimal, patient~specific acetabular im-
`plant placement. The intra—operative tracking and guidance system is used to accurately
`place the implant in the planned optimal position regardless of the position of the patient
`on the operating room table.
`
`By accurately placing the acetabular component in an optimally selected position, the
`HipNav system has the potential to reduce the risk of dislocations, reduce the generation
`of wear debris caused by impingement resulting from malpositioned components, and
`increase the “safe” range of motion. This paper focuses on the transition of the HipNav
`system from a laboratory prototype into a robust clinical system.
`
`2
`
`HipNav System Description
`
`Pre—operative planning in HipNav is based upon a CT scan of the patient’s pelvis. The
`pre—operative planner allows the surgeon to determine the appropriate implant size and
`placement. In the current version of the planner, the surgeon positions cross sections of
`a sphere (i.e., circles) upon orthogonal views of the pelvis to specify the implant’s cen-
`ter of rotation and size, as seen in Fig; 2. The surgeon specifies implant orientation us—
`ing 3-D surface renderings of the pelvis and the implant. We are currently evaluating
`
`Mako Exmbh 1021
`
`Page 2
`
`Mako Exhibit 1021 Page 2
`
`

`
`
`
`Fig. 2. Pre-operative planner - center of rotation and orientation components.
`
`several methods for presenting CT data to the surgeon, and for updating implant place-
`ment based upon surgeon input [I].
`
`Once the surgeon has selected the implant placement and size, the range of motion sim-
`ulator is used to determine the femoral positions (in terms of extension/flexion, abduc-
`tion/adduction, and internal/external rotation) at which impingement would occur for
`the specific implant design and placement. Based upon this range of motion informa—
`tion, the surgeon may choose to modify the selected placement in an attempt to achieve
`an “optimal” implant orientation (in terms of range of motion) for the specific patient.
`The range of motion simulator performs a kinematic analysis which determines an en-
`velope of the safe range of motion, as seen in Fig. 3. More detailed descriptions of the
`range of motion simulator appear in [7] and [8].
`
`The optimal patient—specific plan is used by the HipNav System in the operating room
`on the day of surgery. HipNav permits the surgeon to determine where the pelvis and
`acetabulum are in “operating room coordinates” at all times during surgery. Knowing
`the position of the pelvis during all phases of surgery. and especially during preparation
`and implantation of the acetabular implant, permits the surgeon to accurately and pre-
`cisely position the cup according to the pre-operative plan. Altemately, using HipNav
`
`15° Anteversion impingement
`
`Cup orientation: 45° Lateral Opening
`
`_
`
`Chem \‘
`
`\_"
`‘O
`
`to
`
`x
`20
`
`20
`
`,-” '
`' ‘K
`
`Implant
`Geometry
`
`2° ‘Y
`
`M050“
`E“V€10P€
`
`Fig. 3. Kinematic range of motion simulation.
`
`Mako Exmbh 1021
`
`Page 3
`
`Mako Exhibit 1021 Page 3
`
`

`
`the surgeon can align the implant to an accepted standard such as 45 degrees of abduc-
`tion and 15 degrees of anteversion.
`
`Several devices are used intra-operatively to allow the surgeon to accurately execute the
`pre—operative plan, as seen in Fig. 4. One device is an Optotrak optical tracking camera
`(Northern Digital lnc., Ontario) which is capable of tracking the position of special light
`emitting diodes (targets). These targets can be attached to bones, tools, and other pieces
`of operating room equipment to allow highly reliable tracking. Optotrak can achieve ac-
`curacies of roughly 0.1mm at speeds of 100 measurements per second or higher.
`
`In order to determine the locations of the pelvis and the acetabular implant during sur-
`gery, Optotrak targets are attached to several conventional surgical tools, as seen in
`Fig. 5. In the laboratory prototype of HipNav, the pelvis was tracked by attaching a tar-
`get to the pelvic portion of a Harris leg length caliper (Zimmer, Inc., Warsaw, IN), and
`inserting this device into the wing of the ilium. The acetabular implant was tracked by
`attaching a second target to the handle of an HGP II acetabular cup holder and position-
`er (Zimmer, lnc., Warsaw, IN). A third Optotrak target (which is only needed during
`system setup and calibration) is used to establish an operating room coordinate system
`(i.e., left, right, up and down with respect to the surgeon).
`
`Several key steps are necessary to use the HipN av intra—operative guidance system. One
`of the most important is the registration of pre—operative information (i.e., the CT scan
`and pre—operative plan) to the position of the patient on the operating room table. A lim-
`itation of some registration systems used in orthopaedics is the need for fiducial pins to
`be surgically implanted into bone before pre—operative images are acquired (e.g., see
`[16]). An alternative technique which has been applied by several groups including ours
`uses surface geometry to perform registration [2'][5J[9|[l2l[l3l[l5l. Using this ap-
`proach, the surfaces of a bone (such as the pelvis) can be used to accurately align the
`intra—operative position of the patient to the pre—operative plan without the use of pins
`or other invasive procedures. Using this technique, it is necessary to sense multiple
`points on the surface of the bone with a digitizing probe during surgery. These “intra-
`operative data points” are then matched to a geometric description of the bony surface
`of the patient derived from the CT images used to plan the surgery. A major focus of
`
`
`
`Fig. 4. intra—operative execution.
`
`Fig. 5. Surgical tools instrumented with
`tracking targets.
`
`Mako Exmbh 1021
`
`Page 4
`
`Mako Exhibit 1021 Page 4
`
`

`
`our registration research is the intelligent selection of these intra—operative data points
`in a manner which maximizes registration accuracy while minimizing the quantity of
`data [13].
`
`The registration process is illustrated in Fig. 6. The pelvic surface model was construct-
`ed from CT data using techniques similar to those described in ['4]. The discrete points
`were collected using an Optotrak digitizing probe which was physically touched to the
`indicated points. The goal of the process is to determine a “registration transformation”
`which best aligns the discrete points with the surface model. An initial estimate of this
`transformation is first determined using manually specified anatomical landmarks to
`perform corresponding point registration [6]. Once this initial estimate is determined,
`the surface-based registration algorithm described in [15] uses the pre- and intra—oper—
`ative data to refine the initial transformation estimate.
`
`Once the location of the pelvis is determined via registration, navigational feedback can
`be provided to the surgeon on a television monitor, as seen in Fig. 7. This feedback is
`used by the surgeon to accurately position the acetabular implant within the acetabular
`cavity. To align the cup within the acetabulum in the placement determined by the pre-
`operative plan, the cross—hairs representing the tip of the implant and the top of the han-
`dle must be aligned at the fixed cross hair in the center of the image. Once aligned, the
`implant is in the pre—operatively planned orientation.
`
`
`
`
`
`Fig. 7. Navigational feedback.
`A
`Fig. 8. Real—time tracking of the pelvis.
`
`Mako Exmbn 1021
`
`Page 5
`
`Mako Exhibit 1021 Page 5
`
`

`
`Registration also allows the position of the pelvis to be tracked during surgery using the
`Optotrak system, as demonstrated in Fig. 8. This eliminates the need for rigid fixation
`of the pelvis. In addition, this tracking ability allows us to record the position of the pel-
`vis during surgery, and especially at key times such as during the implantation of the
`acetabular component or during range of motion testing.
`
`3
`
`Development and Validation
`
`initial evaluation of the prototype l-lipNav system was performed in the laboratory un-
`der controlled conditions. As development progressed, we performed a series of evalu-
`ation trials which were progressively more realistic (i.e., similar to the clinical
`environment). At the time this paper was written, we had performed 4 cadaver trials in
`an operating room, with two additional cadaver trials scheduled before initiation of pre-
`clinical patient trials. The goal of the cadaver trials is to validate the various system
`components in terms of robustness, usability, safety and accuracy. The trials have been
`extremely helpful in the design of clinical procedures and protocols.
`
`We have classified development and validation issues into four categories: hardware,
`software, system and accuracy. A summary of the most interesting issues in each of
`these categories is presented below.
`
`3.1 Hardware Issues
`
`We attempted to use as many off-the-shelf hardware components in the l-lipNav design
`as possible. Many of these components are low-technology devices which are decep-
`tively simple. Despite their simplicity, poor selection or design can have serious conse-
`quences in terms of usability, reliability, accuracy and safety. For example, as seen in
`Fig. 8, it is necessary to rigidly fasten a tracking target to the pelvis. The initial device
`used for this purpose was an off-the—shelf component used for measuring leg lengths be-
`fore and after surgery. The device consists of three spikes attached to a rigid platform
`(the device on the left in Fig. 5) which are driven in to the iliac wing during surgery. ln
`HipNav, it is crucial that this device remains fixed relative to the pelvis once data col-
`lection for registration has begun. However, during the later stages of cadaver testing
`we noticed small motions (2-3 mm) of the target fixator relative to the pelvis, necessi-
`tating a re—dcsign of this component. The new design uses threaded screws, instead of
`smooth spikes, to ensure stability.
`
`Ergonomics plays an important role in component design. For example, our initial data
`collection probes were poorly balanced and excessively heavy, factors to which sur-
`geons are keenly sensitive. Data collection probe tips must be designed in a manner
`which allows data collection in a variety of anatomical locations (e.g., sciatic notch.
`percutaneous iliac wing, acetabular rim). Selection of a single probe tip which satisfies
`multiple accessibility constraints was important to eliminate the need for multiple
`probes or for probe re-calibration during surgery.
`
`Later stages of hardware validation focussed on practical problems such as:
`
`° Component sterilization requiring reactivity testing with the sterilization gas
`(ethylene oxide).
`
`Mako Exmbn 1021
`
`Page 6
`
`Mako Exhibit 1021 Page 6
`
`

`
`.0 Electrical isolation of tracking targets.
`
`° Evaluation of material biocompatability.
`
`- Electrical cable routing within the operating room to preserve sterility.
`
`Hardware development benefited greatly from the cadaver trials. In certain cases, hard-
`ware limitations would not have been identified without multiple trials (e.g., slippage
`of the pelvic target fixator, accessibility limitations of data collection probe tips).
`3.2 Software Issues
`
`Software development and evaluation are large fields of study encompassing many de-
`sign and evaluation philosophies. Two areas of software evaluation which are particu-
`larly relevant to computer—assisted surgical applications are functionality and usability
`testing. Functionality testing attempts to answer the question, “does the software cor-
`rectly perform the opeiation for which it was designed?” Usability testing attempts to
`determine whether the software can be efliciently operated by intended target users to
`complete a given task. Ensuring functional and usable software is paramount in com-
`puter—assisted surgical applications.
`
`The HipNav system consists of three primary software components: the pre—operative
`planner, the range of motion simulator, and the intra—opetative control software. Exten-
`sive functionality and usability testing has been performed on all three components, and
`highlights of these activities are summarized below.
`
`Large—scale, pre—operative planner usability tests are currently being performed by our
`group I l I. The goal of these tests is to compare several com petin g user interface designs
`to determine the best design for accomplishing the HipNav pre—operative planning task.
`Evaluation criteria are task accuracy, task completion time, and subjective factors such
`as user fatigue and confusion. In these experiments, test users are asked to reproduce
`particular acetabular implant orientations using the pre—operative planner, based upon a
`physical pelvic model and coupled implant which are presented to them. A complete
`description of this work appears in [1].
`
`For the range of motion simulator, we have concentrated our validation efforts on func-
`tional testing ]8]. In these tests, physical bone models are used to validate the accuracy
`of the computational kinematic simulator. The results of this validation process have
`been very encouraging, and suggest that our simulator is accurate to sub—degree toler-
`ances [8].
`
`The intra—operative control software provides a user interface to the HipNav system for
`use by surgeons. it is crucial that this software be reliable, easy to use, and easy to un-
`derstand. Simple factors such as selection of type fonts and font sizes, selection of back-
`ground and foreground colors, display layouts, and mechanisms for providing feedback
`to the surgeon regarding software state can have a profound impact on the acceptance
`of the HipNav system by clinicians. The software usability evaluations which we have
`performed were based on anecdotal feedback from two surgeons and a usability expert
`during and following the cadaver trials.
`
`Mako Exmbh 1021
`
`Page 7
`
`Mako Exhibit 1021 Page 7
`
`

`
`3.3 System Issues
`
`System issues are those which are not associated with a particular hardware or software
`component of the system, and which are not directly related to system accuracy.
`
`A crucial step in developing a clinical version of HipNav is the design of efficient CT
`scanning protocols. Initial cadaver trials were performed without regard to radiation ex-
`posure, monetary costs or scan time. The primary goal was to maximize the quantity
`and quality of information available for the planning and registration processes. For
`clinical use, it is desirable to minimize radiation exposure, monetary costs and scan
`time, while ensuring sufficient CT data. Two important parameters of the scanning pro-
`tocol include: inter—slice spacing, and extent of the imaged volume. The effects of vary-
`ing these parameters on the pre—operative planning and registration processes is
`currently being studied.
`
`The duration of the CT scanning process is primarily a function of the number of cross-
`sectional images required. The relation between scan time and number of images is not
`necessarily linear, and may depend on factors such as the rate of X—ray tube cooling. As
`the duration of scan time increases, so does the probability of patient motion during the
`scanning process resulting in significant artifacts or errors. We plan to investigate this
`problem by attaching a non-invasive fixturing device to patients receiving CT scans of
`the pelvis (not necessarily HipNav patients). Coupled to the brace are a series of rods,
`similar to those used in stereotactic neurosurgical head frames, which will allow us to
`assess and potentially correct for patient motion during the scanning process. Based
`upon the results of this study, we may include motion correction or patient immobiliza-
`tion as a routine component of the I-lipNav CT scanning protocol.
`
`The method which we currently use for generating accurate surface models of bones
`from CT data is labor intensive. The process requires a trained individual to semi-auto-
`matically extract bone contours from each of the CT slices, a time—consuming task. De-
`pending upon the number of Cl‘ slices, this procedure can take from 1
`to 4 hours. If
`systems such as HipNav are to be used routinely, the amount of manual labor required
`to construct accurate surface models must be reduced. We are pursuing parallel research
`with the goal of automatically generating accurate surface models of bones from CT im-
`ages.
`
`During surgery, the clock is always running. Therefore, minimizing procedural times
`and complications, and improving usage efficiency are important. Shifting setup and
`calibration procedures to the time period before the patient enters the operating room
`has obvious advantages. Additional time savings may be possible via intelligent design
`of hardware components which require assembly during surgery (e.g., the pelvic target
`tracker). Efficiency can also be improved by careful design of procedural transitions be-
`tween conventional portions of the total hip replacement surgery, and those performed
`using HipNav.
`
`Physical space is at a premium in the operating room. Therefore, it is important to min-
`imize the real-estate used by the hardware components of the system. The largest com—
`ponent of the I-lipNav system is the tracking camera. During cadaver trials, the camera
`
`Mako Exmbu 1021
`
`Page 8
`
`Mako Exhibit 1021 Page 8
`
`

`
`was positioned on a floor mounted stand, although we are considering a ceiling mount-
`ed camera for clinical use.
`
`Since the tracking camera requires line—of—sight between the sensor and the tracking tar-
`gets, it is crucial to know when a target is being obscured. This is especially important
`since tracking accuracy may degrade significantly as a function of the number of ob-
`scured LEDs. Therefore, the guidance software continually tests for obscured LEDs
`during the procedure and provides a warning when not enough LEDs can be seen.
`
`3.4 Accuracy Issues
`
`A major advantage of computer—assisted surgery is improvement in procedure execu-
`tion accuracy. In HipNav and related systems, there are many factors which contribute
`to system accuracy [ 14]. During HipNav validation, we have studied registration accu-
`racy, surface model generation accuracy, and tool calibration accuracy.
`
`In [ 14], we described a method for validating the accuracy of surface-based registration
`and the need for task—specific measures of registration accuracy. For the HipNav task,
`it can be shown that only orientation errors in registration are relevant to the implant
`placement task. This is because the system only provides feedback regarding implant
`orientation. Implant position is determined by a reaming process during which I-lipNav
`is not currently used.
`
`We have validated the accuracy of I-lipNav"s registration system using the techniques
`described in [14]. During the four cadaver trials which have been performed to date,
`registration orientation errors have varied between roughly 0.5 degrees and 1.5 degrees.
`If there were no other error sources contributing to implant misalignment, these mea—
`surements suggest that HipNav could position the acetabular implant within a 1.5 de-
`gree cone centered at the desired orientation. In practice, insertion error may be larger
`due to other sources of inaccuracy such as tool calibration errors, deviation of implant
`alignment during the insertion (impaction) process. and target sensing errors. Addition-
`al validation of HipNav registration is being done in the context of intelligent selection
`of intra—operative data points which maximize registration accuracy using minimum-
`sized data sets [13].
`
`4
`
`Conclusions
`
`Many of the issues addressed during the development of HipNav have strong technical
`research components which may have broad application. These issues include:
`
`- Automatic generation of bone surface models from CT—images, and validation
`of the accuracy of these models.
`
`- Optimization of registration accuracy as a function of the collected data [13],
`and subsequent validation of this accuracy [14].
`
`- Design of complex software interfaces to maximize usability for a target user
`group ll ].
`
`In addition to these technical issues, there are many clinical research problems related
`to Hi pNav, including demonstration of efficacy and cost—effeetiveness.
`
`Mako Exmbh 1021
`
`Page 9
`
`Mako Exhibit 1021 Page 9
`
`

`
`it has been our experience that performing technical research together with system de-
`velopment in a tightly coupled manner has several advantages. Perhaps the most impor-
`tant effect is to focus the research towards the solution of real problems. Experience
`gained during system development and validation tends to focus research efforts away
`from contrived problems and non—issues. For example, in the area of accuracy valida-
`tion it is important to define task-specific accuracy requirements to ensure that time is
`not spent achieving unnecessary accuracy.
`
`The HipNav system holds the promise of reducing dislocation rates in primary and re-
`vision total hip replacement by optimizing the placement of acetabular implants and
`minimizing impingement. It also provides a set of tools that will be useful for examin-
`ing assumptions made by conventional methods of total hip replacement surgery.
`
`References
`
`|
`
`I
`
`| M. Blackwell. et 21]., Design and evaluation of 3-d prc-operative planning software: Application to ace-
`tabular implant alignment. Rl tech report CMU-RI-TR-97—TBD, Carnegie Mellon. Pittsburgh, Feb i997.
`[2] E. Cuchet. et al.. Registration in neurosurgery and neuroradiotherapy applications. in Proc. 2nd Int‘!
`S_\v'fI1[J.i\/IRCAS. pp.31—38. Baltimore. Nov 1995.
`
`{3} A.M. DiGioia. et al., HipNav: pre—operative planning and intra—operative navigational guidance for ace-
`tabular implant placement in total hip replacement surgery. In PI'()(‘ CAOS S_vmp.. Bern. Nov. i995.
`[4] B. Geiger. Tltree-dinzen.tir)nal modeling of I11/man organs‘ and its application to diugn0.ti.v and .\'lll‘gi(‘a[
`planning. PhD thesis. Ecole des Mines de Paris. April 1993.
`13] C. J. Henri. et al.. Registration of 3-d surface data for intra—operative guidance and visualization in fra-
`meless stereotactic neurosurgery. in N. A_va(:/ze. ed., Prorr. 1st hit’! CVRA/led. pp. 47-56. Nice. France.
`April 1995. Springer-Verlag.
`
`[6] B. K. P. Horn. Closed-form solution of absolute orientation using unit quatemions. Journal afrhe Optical
`Sm‘ieI_\' 0fAmerii.'u A, 4(4):629—642. April 1987.
`|7l B. Jaramaz. et al.. Simulation of implant impingement and dislocation in total hip replacement. In Com-
`puter A.t'.s'i.s'ied Radiology‘ 96, Paris. June 1996.
`
`|8| B. Jaramaz. et al.. Range of motion after total hip arthroplasty: Experimental verification of the analytical
`simulator. in Prov I5/ Joint CVRA/led / MRCAS. Grenoble. March l997.
`
`|9l S. Lavallee. et al.. Computer-assisted spine surgery: A technique for accurate transpedicular screw fixa-
`tion using CT data and a 3-d optical localizer. Journal oflmage Guided Stirgery. l( l):65—73. April l995.
`[I0] D.E. McCollum. M.D. and WJ. Gray. M.D. Dislocation after total hip arthroplasty. Clinical Ortho-
`paetlic'.\'. 261 :
`I 59—l70. 1990.
`
`|l 1] B.F. Morrey. editor. Re<:'on.rti-uclive Surgerw of!/ze./0im‘.t'. pages l247—l260. Churchill Livingston, 1996.
`[12] L. P. Nolte. et al .. A novel approach to computer assisted spine surgery. in Pmc. I31 Int’! Symp. MRCAS.
`pp. 323—328. Pittsburgh. PA. September I994.
`
`l l3] D. A. Simon and T. Kanadc. Geometric constraint analysis and synthesis: Methods for improving shape-
`based registration accuracy. In Proc. Ix! Joint C VRMez1 / MRCAS. Grenoble. March l997.
`
`I 14] D. A. Simon. et al.. Accuracy validation in image—guided orthopaedic surgery. In PI‘()(‘. 2na'1nI’l Symp.
`MRCAS. Baltimore. Nov. i995.
`
`[15] D. A. Simon, M. Hebert. and T. Kanade. Techniques for fast and accurate intra-surgical registration.
`Journal (1_/Image Guided Surgery. l(I):l7—29. April i995.
`[16] R. H. Taylor. et al.. An image—directed robotic system for precise orthopaedic surgery. IEEE Trans". on
`R(>boIii:.\‘ and Aummalion. l0(3):Z6l—-275. lune l994.
`
`Mako
`
`Exhibit 1021
`
`Page 10
`
`Mako Exhibit 1021 Page 10