The Knee 21 (2014) 359–363
Contents lists available at ScienceDirect
The Knee
A method for assessing joint line shift post knee arthroplasty considering the preoperative joint space Shahram Amiri a, b,⁎, Bassam A. Masri a, Carolyn Anglin c, d, e, David R. Wilson a, b a
Department of Orthopaedics, University of British Columbia, Vancouver, BC, Canada Centre for Hip Health and Mobility (CHHM), Robert H.N. Ho Research Centre, Vancouver, BC, Canada Biomedical Engineering, University of Calgary, Calgary, AB, Canada d Department of Civil Engineering, University of Calgary, Calgary, AB, Canada e McCaig Institute for Bone and Joint Health, University of Calgary, Calgary, AB, Canada b c
a r t i c l e
i n f o
Article history: Received 25 September 2012 Received in revised form 23 January 2013 Accepted 13 March 2013 Keywords: Joint line Total knee arthroplasty Total knee replacement Joint space Knee Arthroplasty
a b s t r a c t Background: Accurate comparison of outcomes regarding various surgical options in knee arthroplasty can benefit from an improved method for joint line analysis that takes into account the preoperative joint space. Methods: This article describes a new preoperative-based registration method that measures changes in the joint line by overlaying the 3D models of the bones with implants using preoperative CT along with preoperative and postoperative biplanar radiography. The method was tested on six cadaveric specimens for measuring alteration to the medial and lateral joint lines in extension and flexion. Results: The joint line shift, when measured using the new method, was in the range of −0.2 to 1.3 mm on average (SD = 1.3 to 3.8 mm, for medial and lateral, in flexion and extension positions). This was significantly different (p ≤ 0.01) from the results of a previous postoperative-based registration method which did not account for the cartilage thickness in calculating alterations of the joint line (mean = 3.9 to 6.8 mm, SD = 1.2 to 4.3 mm). Conclusion: These results further highlight the importance of considering the preoperative joint space in analyzing the joint line, and demonstrate the utility of the newly introduced method for accurate assessment of changes in the joint line after arthroplasty. Clinical relevance: The introduced method provides accurate means for investigating joint line alterations in relation to different surgical techniques and the subsequent biomechanical effects after knee arthroplasty Crown Copyright © 2013 Published by Elsevier B.V. All rights reserved.
1. Introduction A change in the joint line after total knee arthroplasty (TKA) is one of the important factors influencing joint stability, range of motion, and patellofemoral mechanics. Even small changes to the normal level of the joint line can compromise the clinical outcome: deviations as small as 5 mm can cause joint instability [1] and changes as small as 2 mm can have a considerable effect on the range of motion post-arthroplasty [2]. For cruciate-retaining prosthesis designs these effects can be even more substantial: a 4 mm elevation of the joint line can produce significant additional strains in the posterior cruciate ligament [3,4]. The patellofemoral joint can also be adversely affected by changes in the joint line; a 3 mm distal shift to the joint line has been reported to cause patellofemoral pain and subluxation [5,6]. Because of these significant effects on the outcomes of knee arthroplasty, measurements of the joint line shift have been appreciated as an important factor in research studies that compare different surgical options and implant designs
⁎ Corresponding author. Tel.: +1 604 675 2575. E-mail address:
[email protected] (S. Amiri).
[7–9]. For these investigational studies an accurate and reproducible method is required to detect small alterations to the joint line for extension and flexion positions and for the medial and lateral sides of the joint individually. Current methods for measuring joint line alterations have limitations. Commonly used techniques are based on anteroposterior or lateral radiographs, and they measure landmarks that define the joint line with reference to common points such as the fibular head, the medial femoral epicondyle [10], or the tibial tuberosity [11]. These radiographic methods lack the ability to measure important differences between the medial and lateral joint lines [12,13], and they are prone to poor accuracies due to sensitivity to patient and X-ray beam positions [14]. To overcome these shortcomings, a more recent method overlays the 3D models of the preoperative bones and implants with reference to postoperative radiographs, using two-dimensional and three dimensional (2D–3D) matching [12]. This particular method, called ‘postoperative-based registration’ from this point forward, measures variations of the tibial and femoral joint lines in addition to the posterior condylar offsets (PCO), individually for the medial and lateral sides of the joint. Since in this method the joint line is measured based on subchondral bone levels that appear on the radiographs and not the
0968-0160/$ – see front matter. Crown Copyright © 2013 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.knee.2013.03.011
360
S. Amiri et al. / The Knee 21 (2014) 359–363
cartilage surfaces, it does not account for the preoperative joint space [12]. This limitation, as has been acknowledged by the authors of the paper, will be less if the cartilage on the articular surfaces is completely worn away in all regions. However, this is not usually the case, and typically there is a non-uniform pattern of cartilage degeneration. Although accounting for the cartilage thickness for joint line assessment seems trivial, there has not been a method available that can accurately measure joint line alterations after knee arthroplasty by directly comparing post-operative with pre-operative status. A recent study has shown considerable variability in the thickness of the cartilage of the posterior condyles, concluding that intraoperative measurements of the cartilage thickness must be added to the corresponding postoperative radiographic measurements [15]. This method, however, is not useful for the retrospective studies in which intraoperative caliper measurements of the cartilage thickness are not available. Even when the intraoperative assessment is available, since the bone cuts are made parallel to the coronal and transverse planes, measuring the cartilage thickness is only limited to flexion and extension points and not the important mid-flexion range for which alteration to the joint line has been shown to have the highest effects on instability [1]. Considering the shortcomings of the current techniques, there is need for an improved method that can assess the joint line shift after knee arthroplasty based on the preoperative joint space. The aim of this study was a) to introduce an accurate method to measure the joint line shift for any desired flexion angle of the joint by taking into account the cartilage thickness on both of the medial and lateral sides of the joint; and b) to compare this new method of assessing joint line shift to a previous post-operative based registration method that does not account for cartilage. 2. Methods The method requires a) preoperative joint imaging; b) postoperative joint imaging; c) registration of the 3D models of the bones and implants to the postoperative radiographs to determine implant positions relative to the corresponding bones; d) registration of the 3D models of the bones to the preoperative radiographs and subsequently finding the relative positions of the implant components according to the preoperative bone positions; and e) analyzing the changes in the joint line on the medial and lateral sides for both extension and flexion. This method, called ‘preoperative-based registration’ from this point on, takes into consideration the preoperative joint gap that represents the sum of tibial and femoral cartilage thicknesses. This technique superimposes the component onto the preoperative bone positions and measures the joint line shift as inter-penetration or separation of the implant components. 2.1. Preoperative imaging Computer models of intact cadaveric knee specimens were generated. Six previously frozen cadaveric knees from two male and four female donors (ages: 34 to 90), were obtained and prepared for testing. A computed tomography (CT) scan of each specimen was acquired using a Toshiba Aquilion scanner (TOSHIBA, Tokyo, Japan) with the specimens in an orientation simulating a supine patient position using the following settings: 120 kV, 160 mA, slice thickness = 2.0 mm, slice spacing = 2.0 mm, Bone Boost smoothing algorithm. Threedimensional models of the bones were constructed by segmenting the bone volumes using Analyze software, version 10 (AnalyzeDirect Inc., Overland Park, KS, US). The preoperative positions of the bones at knee extension and flexion were determined. Metal rods were fixed in the distal and proximal intramedullary canals of the tibia and femur using bone cement. Specimens were placed, using these rods, into a custom jig that held the specimens, simulating the patient's weightbearing poses at 0° or 90° of flexion (referred to as extension and flexion
positions). Orthogonal biplanar images of the specimens were acquired using a multi-planar imaging method (previously described [16]) that used a C-arm fluoroscope (Siemens Arcadis Orbic; Siemens AG, Munich, Germany) and motion tracking camera (Optotrak Certus, NDI, Waterloo, ON, Canada). The relative locations of the bones at both extension and flexion were reconstructed (Fig. 1a) by matching the 3D CT models of the bones to the biplanar fluoroscopic images using an edge and intensity matching algorithm [17] implemented in JointTrack biplanar 2D–3D registration software (University of Florida, Gainesville, US; http://sourceforge.net/projects/jointtrack/). 2.2. Postoperative imaging Total knee arthroplasty was performed on each specimen using posterior stabilized knee replacement prostheses (NexGen Legacy Posterior Stabilized; Zimmer, Warsaw, IN) and following the recommendations of the standard protocol. Each knee was then placed in extension, and biplanar images were acquired as described for the preoperative imaging. The tibia and femur were imaged separately to include as much bone as possible for each of the bones for the 2D–3D registration. 3D models of the femoral component and the metal tibial tray (obtained from high dose CT imaging of the components) were registered to the 2D biplanar postoperative images (Fig. 1b) using the JointTrack software. 2.3. Joint line analysis Joint line shift for the medial and lateral sides of the joint was assessed by transferring the implant position information from the postoperative to the preoperative images of the joint. The 3D models of the implant components were superimposed on the preoperative positions of the bones using Rapidform XOV (INUS Technology, Seoul, Korea) (Fig. 1c). This was done by combining the results from preoperative and postoperative 2D–3D registrations (as discussed above), and translating the relative coordinates of the 3D models of the implant components with respect to the bones from the postoperative into the preoperative cases. To determine the tibial joint line, the three-dimensional model of the polyethylene inlay was added to the metal tray according to the design of its locking mechanism. Two-dimensional cross-sectional slices of the combined bone and implant models were obtained using a sectioning plane that was passed perpendicular to the tibial tray and through the two most distal points on the medial and lateral condyles of the femoral component. Joint line shift was measured as the distance between the most distal point on the condyle of the femoral component and the most proximal point on the articular surface of the tibial polyethylene in the direction normal to the mediolateral edge of the tibial tray in the cross-sectional slice (Fig. 2a and c). It was assumed that for ideal restoration of the joint line, and for cases where no deformity or soft tissue contracture exists, there would be no joint line shift (after taking into consideration thickness of the worn cartilage) therefore for these cases zero distance is expected between the ‘predicted preoperative’ positions of the tibial and femoral TKA components. A positive shift in the joint line level according to this definition indicates interference between the components and suggests that TKA increased the joint gap, forcing extra tension in the corresponding collateral ligament. A negative shift indicates separation of the components, suggesting reduced ligament tension (or slackness compared to normal) after TKA. For the sake of comparison, the previously suggested postoperativebased registration method was applied to the same cross-sectional planes described above as an alternative way to measure the joint line shift as the sum of the tibial and femoral implant–bone distances. The implant–bone distance was defined as the distance between the articular surface of the implant and the corresponding subchondral surface of
S. Amiri et al. / The Knee 21 (2014) 359–363
361
Fig. 1. Imaging and registration steps of the methodology: (a) preoperative imaging was used to register the 3D models of the bones to the biplanar images of the patient (cadaveric specimen) in extension and flexion; (b) postoperative imaging was used to register each individual femoral or tibial component to the corresponding bones using biplanar 2D–3D registration software (JointTrack Biplanar); and (c) the implant components were overlaid over the preoperative locations of the bone models, based on the relative positions determined from step (b) for both extension and flexion. The resulting 3D models were used for joint line analysis.
the bone before cutting. The medial and lateral femoral implant–bone distances were measured as distances between the most distal points on the femoral component condyles to the preoperative femoral condyle in the direction parallel to the proximodistal edges of the femoral component in the cross-sectional slice (Fig. 2b and d). The corresponding tibial implant–bone distances were measured by taking the most proximal points of the bone and implant in the direction perpendicular to the mediolateral edge of the tibial tray in the 2D slice (Fig. 2b and d). Positive values were assigned to these implant–bone measures when the component was more distal compared to the bone for the femoral joint line, and when the implant was more proximal compared to the bone for the tibial joint line (i.e. when the component added thickness to the bone in each case). The sum of the individual tibial and femoral implant–bone distances was calculated as suggested by the previous postoperative-based registration method [12].
2.4. Statistical analysis A power analysis predicted that if ignoring cartilage produced errors in the joint line calculations with a mean value of 4 mm and with an underlying variance of 2 mm, five specimens would be required to detect that difference with 80% power and p = 0.05. A Student's paired t-test with a one-tailed distribution was used to determine whether the joint line shift measured using the new method was significantly different from the previous joint line shift measured for each of the medial and lateral compartments in extension and flexion (p b 0.05 was considered significant). 3. Results Joint line shift measurements using the new preoperative-based registration were significantly smaller (p ≤ 0.01) than those calculated by the previous postoperative-based
Fig. 2. Joint line analysis for extension (a and b) and flexion (c and d) on cross-sectional slices of the assembled bone and implant models (Fig. 1c). The postoperative joint line shift, using the new method, was measured as the distance between the implant surfaces (dMC, dLC in a and c). A positive shift in this definition indicates interference between the components which suggests that TKA increased the joint gap and produced extra tension in the corresponding collateral ligament. A negative shift indicates separation of the components which suggests reduced ligament tension (or slackness compared to normal) after TKA. The femoral and tibial joint line offsets, using the conventional method, were measured as the distance between the implant surface and the corresponding subchondral bone (dMF, dLF, dMT, and dLT in b and d).
362
S. Amiri et al. / The Knee 21 (2014) 359–363
Table 1 Joint line shift results: joint line shift (dMC and dLC in Fig. 2a and c), as measured using the new preoperative-based registration (the distance between the implant surfaces when assembled using the preoperative positions of the bones as the reference). Femoral joint line offsets (dMF and dLF in Fig. 2b and d), and tibial joint line offsets (dMT and dLT in Fig. 2b and d) are measured as the distance between the implant and boney surfaces of the condyles, using the postoperative-based registration described in the literature. The joint line shift from the previous postoperative-based registration is the sum of the femoral and tibial joint line offsets. Measurements (mm)
Extension Medial
Joint line shift: new preop-based registration Femoral implant–bone distance Tibial implant–bone distance Joint line shift: previous postop-based registration Difference between preop and postop-based registration p value
Flexion Lateral
Medial
Lateral
1.1 ± 1.3
1.1 ± 2.5
1.3 ± 2.0
−0.2 ± 3.8
−0.6 ± 1.1 4.5 ± 1.6 3.9 ± 1.2
0.5 ± 2.0 4.5 ± 2.6 5.0 ± 2.8
0.7 ± 1.9 4.4 ± 2.6 5.1 ± 4.1
1.7 ± 1.5 5.0 ± 3.0 6.8 ± 4.3
2.8 ± 1.3
4.0 ± 2.8
3.7 ± 3.8
7.0 ± 5.0
0.01
0.03
0.01
0.002
registration method (Table 1). Joint line shift calculated by the preoperative-based registration ranged from −0.2 to 1.3 mm (SD = 1.3 to 3.8 mm). The corresponding measures from postoperative-based registration, calculated as the sum of the individual femoral and tibial implant–bone distances, ranged from 3.9 to 6.8 mm (SD = 1.2 to 4.3 mm). The femoral implant–bone distance ranged from −0.6 to 1.7 mm (SD = 1.1 to 2.0 mm), and the tibial implant–bone distance ranged from 4.4 to 5.0 mm (SD = 1.6 to 3.0 mm). The average preoperative joint space was 2.8 mm on the medial side and 4.0 mm on the lateral side (Fig. 3). For the flexion position, these measures were about 2 mm larger: 4.7 mm and 6.0 mm for the medial and lateral sides, respectively (Fig. 3).
4. Discussion The new preoperative-based registration method measures changes to the joint line after TKA on both the medial and lateral sides of the joint for both extension and flexion by taking into account the preoperative joint space (combined tibial and femoral cartilage thicknesses). The results from the current method were up to 7 mm different on average compared to the previously suggested postoperative-based registration that calculated the joint line shift as the sum of the femoral and tibial implant–bone distances without considering cartilage thicknesses (Table 1). This appears to be an unacceptable margin of error for the previous technique, given that a 2 mm shift to the joint line can potentially make a difference between good and poor clinical outcomes [4]. The limitation of the previous postoperative-based method in not considering the articular cartilage has been acknowledged by its authors [12]. In case of severe cartilage losses on both the medial and lateral sides of the joint and for both flexion and extension, the errors resulting from their method can be minor. However, this is not typically the case for knee arthroplasty patients and there is normally remaining cartilage particularly on the lateral side and for flexion contact points, which should be taken into account.
Fig. 3. Preoperative joint space for the medial and lateral sides of the joint for both extension and 90° flexion. These measurements were from 3D analysis of the 3D models of the bones using six cadaveric specimens.
It is important to note that the current method only provides accurate means for studying the joint line shift, and does not necessarily suggest restoration of the preoperative joint line. It is clear that significant cartilage loss and deformity can exist in arthroplasty patients and the absolute priority of the knee arthroplasty is not to restore the preoperative joint line, but to produce a knee that is stable, balanced in flexion and extension, and properly aligned with the mechanical axis. These objectives can only be accomplished by considering multiple factors including deformities, structural deficiencies, contracture and soft tissue abnormalities. The methodology suggested in this study provides an appropriate tool for investigating the importance of the joint line alterations in relation to outcome variables, and allows comparing the efficacy of various surgical techniques for executing the planned alterations to the joint lines. This study highlights the importance of taking both of the tibial and femoral cartilage thicknesses into consideration for accurate and reliable joint line assessment. A recent study has demonstrated large variability of the cartilage thickness for the posterior femoral condyles (up to 4.0 and 5.0 mm for the medial and lateral posterior condyles, respectively) and concluded that changes to posterior condylar offsets cannot be solely judged based on radiographic measurements [15]. The current study shows an even larger range of variation for the combined tibial and femoral cartilage thicknesses: up to 7.5 mm for the medial side and up to 15.0 mm for the lateral side (Fig. 3), which is a 1.5 to 3 times larger range compared to the case when only the femoral cartilage thickness was considered [15]. This not only confirms the importance of variations of the femoral cartilage, but also suggests that taking into consideration the tibial cartilage is equally important when analyzing alterations to the joint line. An advantage of the new technique is that it allows accurate analysis of the joint line shift at any desired flexion angle, including mid-flexion where variation in the joint line has the most effect on joint instability [1]. The previously suggested intraoperative measures of the cartilage are not able to provide an equivalent measurement for mid-range flexion angles, because the osseous cuts in knee replacement are made in transverse and coronal planes. Navigation-based methods can provide exact measures of joint line offsets for each component based on intraoperative digitization [18]. The new technique, however, allows assessment of the joint line shift regardless of the surgical technique being used, and provides the means for comparing across conventional mechanical guides, navigation, or shape-matching techniques. Furthermore, in the new method the joint line calculations are based on more clinically relevant weightbearing positions, which can be different from the intraoperative navigation-based measurements in which the joint is unloaded. One limitation of the current method is the considerable effort that is required for imaging and analysis. This can be justified for research applications where accuracy is a critical factor for measuring changes to the joint line. Examples of these applications are when investigating the relationships between the joint line restoration and postoperative complications, and in comparing the efficacy of various techniques for restoring the joint lines. The combination of preoperative and postoperative imaging along with 2D–3D registration and analysis is more time-consuming and complex compared to the common 2D radiographs. However, this method takes into consideration the critical information of the local cartilage thickness on both the medial and lateral compartments of the tibial and femoral articular surfaces based on weightbearing positions of the patient. Two-dimensional lateral radiographs are limited since they require perfect lateral X-ray shots for both preoperative and postoperative conditions and may not be able to take into consideration the important differences between the medial and lateral compartments [12,13]. The accuracy of identifying the tibial tuberosity and the inferior surfaces of the bone and implant on a true lateral view is uncertain; the complex geometric shapes of the boney landmarks and their different appearances depending on the view angle suggest large inherent errors when a single radiographic view is
S. Amiri et al. / The Knee 21 (2014) 359–363
used for joint line analysis. The AP radiographs for measuring joint space width are also highly sensitive to the patient position and X-ray beam inclination especially for flexed joint positions [14]. The method suggested in this study has sub-millimeter accuracy for registering the 3D bone and implant positions, with an additional important advantage of not being sensitive to exact alignment of the patient and the X-ray beam position [16]. The introduced method in its current form due to inherent complexities is more appropriate as a research tool. However, current technical developments including more accessible and less expensive multi-planar radiography along with advanced image segmentation techniques and robust statistical shape models can significantly simplify the process toward convenient clinical use. The methodology of this study is based on the assumption that the joint contact points were located on the same cross-sectional planes before and after arthroplasty. This assumption was made in order to make a direct comparison possible between the preoperative and postoperative states. Many factors affect the location of the contact points after arthroplasty including the design of the components, the lines of action and magnitudes of the external loads and the effects of muscles and patellofemoral contacts. However, using similar cross-sectional planes can be justified considering that the coronal plane is the most clinically-relevant direction for joint line analysis and the kinematics of the joint in the transverse plane are less relevant when analyzing the joint space and the corresponding effects on tensions in the collateral ligaments. Another limitation was related to the lack of an external axial load applied to the cadaveric specimens during simulated weightbearing imaging. This is not expected to cause large errors since cartilage deformation due to weight is within a sub-millimeter range and the design of the test rig put the specimens under their own weight to prevent lift-off of the condyles. In conclusion, the method introduced in this study combines the preoperative and postoperative images of the patient's joint to provide accurate 3D measures of joint line shifts post-arthroplasty. The method incorporates important cartilage thickness information, and can be applied at any flexion angle of the knee, without sensitivity to patient and X-ray beam positions. The method yields much smaller measures of joint line shift than a previous approach, suggesting that the previous method may have substantially overestimated the joint line shift. The suggested method is recommended when investigating relationships between joint line alterations and numerous clinical and kinematic factors including the surgical techniques, subsequent biomechanical effects, and postoperative complications. Acknowledgments The authors acknowledge the generosity and support of Zimmer Inc. in donating the implant components and allowing the use of surgical
363
instruments for implanting the components. We also acknowledge the support of the Canadian Arthritis Network and Alberta Innovates — Technology Futures for funding.
References [1] Martin JW, Whiteside LA. The influence of joint line position on knee stability after condylar knee arthroplasty. Clin Orthop Relat Res 1990:146–56. [2] Ryu J, Saito S, Yamamoto K, Sano S. Factors influencing the postoperative range of motion in total knee arthroplasty. Bull Hosp Jt Dis 1993;53:35–40. [3] Emodi GJ, Callaghan JJ, Pedersen DR, Brown TD. Posterior cruciate ligament function following total knee arthroplasty: the effect of joint line elevation. Iowa Orthop J 1999;19:82–92. [4] Wyss TF, Schuster AJ, Munger P, Pfluger D, Wehrli U. Does total knee joint replacement with the soft tissue balancing surgical technique maintain the natural joint line? Arch Orthop Trauma Surg 2006;126:480–6. [5] Cope MR, O'Brien BS, Nanu AM. The influence of the posterior cruciate ligament in the maintenance of joint line in primary total knee arthroplasty: a radiologic study. J Arthroplasty 2002;17:206–8. [6] Insall J, Goldberg V, Salvati E. Recurrent dislocation and the high-riding patella. Clin Orthop Relat Res 1972;88:67–9. [7] Lee HJ, Lee JS, Jung HJ, Song KS, Yang JJ, Park CW. Comparison of joint line position changes after primary bilateral total knee arthroplasty performed using the navigation-assisted measured gap resection or gap balancing techniques. Knee Surg Sports Traumatol Arthrosc 2011;19:2027–32. [8] Snider MG, Macdonald SJ. The influence of the posterior cruciate ligament and component design on joint line position after primary total knee arthroplasty. J Arthroplasty 2009;24:1093–8. [9] Selvarajah E, Hooper G. Restoration of the joint line in total knee arthroplasty. J Arthroplasty 2009;24:1099–102. [10] Laskin RS. Joint line position restoration during revision total knee replacement. Clin Orthop Relat Res 2002:169–71. [11] Figgie III HE, Goldberg VM, Heiple KG, Moller III HS, Gordon NH. The influence of tibial-patellofemoral location on function of the knee in patients with the posterior stabilized condylar knee prosthesis. J Bone Joint Surg Am 1986;68: 1035–40. [12] Sato T, Koga Y, Sobue T, Omori G, Tanabe Y, Sakamoto M. Quantitative 3-dimensional analysis of preoperative and postoperative joint lines in total knee arthroplasty: a new concept for evaluation of component alignment. J Arthroplasty 2007;22:560–8. [13] Ishii Y, Noguchi H, Takeda M, Ishii H, Toyabe S. Changes in the medial and lateral posterior condylar offset in total knee arthroplasty. J Arthroplasty 2011;26:255–9. [14] Ravaud P, Auleley GR, Chastang C, Rousselin B, Paolozzi L, Amor B, et al. Knee joint space width measurement: an experimental study of the influence of radiographic procedure and joint positioning. Br J Rheumatol 1996;35:761–6. [15] Clarke HD. Changes in posterior condylar offset after total knee arthroplasty cannot be determined by radiographic measurements alone. J Arthroplasty 2012. [16] Amiri S, Wilson DR, Masri BA, Sharma G, Anglin C. A novel multi-planar radiography method for three dimensional pose reconstruction of the patellofemoral and tibiofemoral joints after arthroplasty. J Biomech 2011;44:1757–64. [17] Mahfouz MR, Hoff WA, Komistek RD, Dennis DA. A robust method for registration of three-dimensional knee implant models to two-dimensional fluoroscopy images. IEEE Trans Med Imaging 2003;22:1561–74. [18] Ensini A, Catani F, Biasca N, Belvedere C, Giannini S, Leardini A. Joint line is well restored when navigation surgery is performed for total knee arthroplasty. Knee Surg Sports Traumatol Arthrosc 2012;20:495–502.