The Journal of Arthroplasty Vol. 24 No.

7 2009

Three-dimensional Analysis of Computed TomographyBased Navigation System for Total Knee Arthroplasty
The Accuracy of Computed TomographyBased Navigation System
Hideki Mizu-uchi, MD, PhD,* Shuichi Matsuda, MD, PhD,* Hiromasa Miura, MD, PhD,* Hidehiko Higaki, PhD,y Ken Okazaki, MD, PhD,* and Yukihide Iwamoto, MD, PhD*

Abstract: We evaluated the postoperative alignment of 37 primary total knee arthroplasties performed using a computed tomographybased navigation system (Vector Vision Knee 1.5; Brain Lab, Germany) with a new 3-dimensional analysis. The mean coronal femoral angle was 89.0 1.4 (85.5-92.8), and the coronal tibial component was 89.2 1.0 (87.4-91.6). The hip-knee-ankle angle was observed to be 178.2 1.5 (173.9-181.8). The external rotational alignment of the femoral component relative to the surgical epicondylar axis was 0.5 1.7 (3.2 to 3.4). The results demonstrated that a computed tomographybased navigation system provided a reasonably satisfactory component alignment. The discrepancy between the 2-dimensional and 3-dimensional evaluations was 1.0 0.9 (0.1-3.4). Threedimensional analysis is necessary to evaluate the accuracy of the navigation system. Keywords: total knee arthroplasty, navigation system, alignment, 3-dimensional analysis, CT. 2009 Elsevier Inc. All rights reserved.

Optimal postoperative alignment of the lower extremity is important for achieving long-term survival after total knee arthroplasty (TKA). Any misalignment of the components could lead to various types of implant failures, such as aseptic loosening and instability, polyethylene wear, and patellar dislocation. Many surgeons use the alignment guide system because the center of the hip and
From the *Department of Orthopaedic Surgery, Graduate School of Medical Sciences, Kyushu University, Fukuoka City, Japan; and yFaculty of Engineering, Kyushu Sangyo University, Fukuoka City, Japan. Submitted February 15, 2008; accepted July 9, 2008. No benefits of funds were received in support of the study. Reprint requests: Shuichi Matsuda, MD, PhD, Department of Orthopaedic Surgery, Graduate School of Medical Sciences, Kyushu University, 3-1-1, Maidashi, Higashi-ku, Fukuoka City 812-8582, Japan. 2009 Elsevier Inc. All rights reserved. 0883-5403/08/2407-0016$36.00/0 doi:10.1016/j.arth.2008.07.007

ankle joints cannot be observed directly. However, it has been reported that these surgical techniques have an upper limit with respect to the accuracy of the postoperative alignment [1-9]. Navigation systems have been developed for TKAs to decrease the number of outliers required for achieving proper alignment. Many clinical and experimental studies of these navigation systems have demonstrated that the accuracy of implantation has improved following their use [10-22]. For both computed tomography (CT)-based navigation systems and image-free navigation systems, more than 90% of the operated knees achieved a postoperative mechanical axis alignment of the leg within 3 of neutral alignment [10-16,18-20,22]. Studies of navigation systems largely agree upon the necessity of obtaining accurate coronal postoperative alignments. However, in almost all these studies, analyses have been conducted using

1103

1104 The Journal of Arthroplasty Vol. 24 No. 7 October 2009 radiographs and/or CT scans without 3-dimensional (3D) evaluation. These analyses comprised 2-dimensional (2D) evaluations that are affected by the positioning of the limb as well as the scanning direction. Therefore, it was difficult to accurately measure the angle of the position of the implants. Recently, 3D analyses of image-free navigation systems have been reported by Chauhan et al [13] and Matziolis et al [16]. The present study established a new method for 3D reconstruction from postoperative CT images to accurately measure the alignment of the component relative to any designed plane. Thus far, few data have been published on such a 3D analysis of CT-based navigation systems. It is worth evaluating the postoperative alignment of the CT-based navigation system by a 3D method. Our hypothesis was that component alignment would be evaluated accurately by a new 3D analysis compared to conventional radiographic evaluation, and a CT-based navigation system would provide a satisfactory component alignment.
Table 1. Preoperative Demographic Data
Mean age (y) Sex male/female Diagnosis OA/RA HKAA Mean follow-up time (y) 76.1 6.0 (61-87) 5 knees (4 patients)/ 32 knees (29 patients) 34 knees (30 patients)/ 3 knees (3 patients) 170.8 9.1 (157.0-192.0) 2.2 0.8 (0.9-4.9)

OA indicates osteoarthritis; RA, rheumatoid arthritis.

Preoperative planning procedure and surgical techniques For CT scans, a 100-mm section of the femoral head, a 200-mm section whose midpoint was the knee joint, and a 100-mm section of the distal tibia were scanned with a slice thickness of 2 mm. From these data, we defined the center of the femoral head, the center of the knee joint, and the center of the ankle joint after adjusting for the bone threshold and the window level and width. The bone threshold value was between 100 and 150 HU according to the patient's CT data; hence, extra artifacts were deleted, and the bone surface was identified as clearly as possible. For the femoral component, the coronal alignment was set perpendicular to the mechanical axis of the femur. The planned sagittal alignment was set parallel to the anatomical axis of the distal femur to avoid notching of the femur due to its anterior bowing (average 3.0 1.2 [1.0-6.0] in flexion to the mechanical axis of the femur). The rotational alignment was adjusted to the surgical epicondylar axis (SEA) [23], which was a line connecting the sulcus of the medial epicondyle and the most prominent points of the lateral epicondyle of the femur. After aligning the femoral component to the axis, we adjusted the size of the femoral component to be as close as possible to the posterior condyles. For the tibial component, the planned coronal alignment was set perpendicular to the mechanical axis of the tibia, which was defined by a straight line between the center of the cutting line of the proximal tibia and the center of the ankle joint [8]. The planned sagittal alignment was set parallel to the lateral anatomic tibial slope. We determined the position of the tibial component to be 10 mm distal to the highest point of the tibial plateau, and the size was chosen such that the component would not overhang the medial border of the tibia. For surgical techniques, standard medial parapatellar exposure was used. The reference clamp was fixed to the distal femur or the proximal tibia with a pin (2 pins, instead of one, were used from

Materials and Methods

Patients This study was approved by the institutional review board. One hundred thirty-six primary TKA were performed using the NexGen Legacy Posterior-Stabilized prosthesis (Zimmer, Warsaw, Ind) by surgeons (S.M, H.M and K.O) between November 2002 and May 2006. Of these, 53 primary TKAs (39% of total TKAs) were performed in 47 patients using a CT-based navigation system (Vector Vision Knee 1.5; Brain LAB Inc, Heimstetten, Germany) after patients were informed of the risk of radiation exposure and the duration required by the study. After the surgery, 1 patient (1 knee) died and 1 patient (1 knee) had an ipsilateral fracture of the femoral shaft due to falling down. Four patients (4 knees) were not evaluated taking into consideration the possibility of over radiation because of the use of CT scans for examining deep vein thrombosis (3 patients) and headache (1 patient). Informed consent was not obtained in the case of 10 patients (10 knees) because of the risk of radiation exposure from the postoperative CT scan. Consequently, only 37 primary TKAs (69.8% of TKAs using the navigation system) performed for 33 patients have been included in this study. The demographic data for patients are presented in Table 1.

Three-dimensional Analysis of CT-Based Navigation System for TKA  Mizu-uchi et al

1105

April 2005). Registration using surface matching of bones was achieved with a pointer to match the actual femur or tibia to the corresponding 3D CT images on the screen. A minimum of 8 points to a maximum of 20 points were registered until an accuracy of 1.9 mm or better was achieved. By using the cutting block adapter, the femoral and tibial cutting blocks were positioned to match the preoperatively planned cutting plane, which was shown on the navigation system. After resection, all the planes were checked by the verification tool of the navigation system. The patella was resurfaced in all the patients. All femoral, tibial, and patellar components were fixed with cement. We did not change preoperative planning, navigation, and surgical procedure for this period. Evaluation of the postoperative alignment The postoperative knee statuses were assessed using The Knee Society scoring system. Postoperative CT was performed in the same manner as the preoperative CT. A 100-mm section of the hip joint (around the femoral head) and a 100mm section of the ankle joint were scanned with a 2-mm slice thickness. A 200-mm section of the knee joint was scanned with a 1-mm slice thickness. We analyzed the CT scan data using computer software (Real INTAGE Ver.3.0; KGT Inc, Tokyo). The hip joint, the ankle joint, and the implants in the knee joint were extracted with adequate thresholds. We determined the coordinate of the center of the femoral head and

the ankle joint, the distal anatomical axis of the femur, the center of the femoral and tibial components, the SEA, the geometric centers of the femoral component pegs, and the posterior edge of the tibial component by using the 3D method. The outer shape of the extracted images was modeled by using a software developed inhouse and then loaded onto a CAD software (CATIA Ver 5.12; Dassault Systemes, France) (Fig. 1). We defined the mechanical axes and planes by substituting the coordinates and analyzed the postoperative alignment of the femoral and tibial components from these images. 1. Coronal alignment: The coronal femoral component angle (CFA) was defined as the medial angle between the mechanical axis of the femur and the horizontal axis of the 2 prosthetic condyles (Fig. 2A). The coronal plane of the femur was defined as corresponding to the mechanical axis of the femur and the SEA. The coronal tibial component angle (CTA) was defined as the medial angle between the mechanical axis of the tibia and the horizontal axis of the tibial tray (Fig. 2B). The coronal plane of the tibia was defined as corresponding to the mechanical axis of the tibia and the line connecting the posterior edge of the tibial component. The coronal alignment of the femoral and tibial components had been planned such that the mechanical axis passed through the center of each com-

Fig. 1. Three-dimensional model loaded onto a CAD software.

1106 The Journal of Arthroplasty Vol. 24 No. 7 October 2009

Fig. 2. Measurement of the femoral and tibial component alignment. (A) The CFA, (B) the CTA, (C) the SFA, (D) the STA, (E) the RFA.

ponent perpendicularly. Therefore, the optimal CFA and CTA were 90 each. The medial angle between the mechanical axis of the femur and that of the tibia was measured (hipknee-ankle angle [HKAA]). The CFA, CTA, and HKAA were also measured by the conventional radiograph method (full-length weight-bearing anteroposterior radiographs) and compared with the angle obtained using the 3D method. A Mann Whitney U test was used to determine statistically significant differences in absolute value of the difference between the 2 methods using these parameters (significant: P b .05). 2. Sagittal alignment: The sagittal femoral component angle (SFA) was defined as the posterior angle between the mechanical axis of the femur and a line drawn perpendicular to the distal part of the femoral component (Fig. 2C). The sagittal plane of the femur was defined by a normal to a line connecting the geometric centers of the femoral component pegs. The optimal SFA was 90the difference between the mechanical axis and the anatomical axis in the sagittal plane. The average optimal SFA

was 87.0 1.2 (84.0-89.0). The sagittal tibial component angle (STA) was defined as the posterior angle between the mechanical axis of the tibia and a line on the tibial base plate (Fig. 2D). The sagittal plane of the tibia was defined by the normal to the line connecting the posterior edge of the tibial component. The optimal STA was the angle between the mechanical axis and the lateral anatomic tibial slope in the sagittal plane. The average optimal STA was 83.1 0.5 (82.0-85.0). 3. Rotational alignment: The rotational femoral component angle (RFA) was defined as the angle between the SEA and the posterior condylar line of the femoral component (+: External rotation in relation to the SEA; Fig. 2E). The rotational plane of the femur was defined by the normal to the mechanical axis of the femur. The optimal RFA was 0.

Results
The clinical data of the preoperative and the postoperative data were presented in Table 2.

Table 2. Comparison Between the Preoperative and the Postoperative Clinical Data
Preoperative Knee Society score Maximum extension angle Maximum flexion angle * Significantly different from preoperation. 53.6 12.2 (20 to 73) 9.7 8.6 (30 to 0) 116.9 16.6 (85 to 145) Postoperative 95.5 5.0 (81 to 100) * 2.2 4.3 (15 to 0) * 115.0 15.4 (90 to 140)

Three-dimensional Analysis of CT-Based Navigation System for TKA  Mizu-uchi et al Table 3. Comparison Between the 3D Method and the Conventional Radiograph Method
Optimal HKAA (180) CFA (90) CTA (90) 3D Method 178.2 1.5 (173.9-181.8) 89.0 1.4 (85.5-92.8) 89.2 1.0 (87.4-91.6) Radiograph Method 178.5 1.8 (173.5-181.2) 89.0 1.5 (85.0-91.4) 89.5 1.3 (87.0-93.0)

1107

Absolute Value of Difference 1.0 0.9 (0.1-3.4) * 0.8 0.6 (0.0-2.8) * 0.7 0.6 (0.1-2.2) *

* Significantly different between 2 methods.

Statistically significant differences were detected in Knee Society score and the maximum extension angle (P b .0001 and P = .0001, Wilcoxon test). No patients had flexion contracture more than 20 or a flexion angle of less than 90. There were no severe complications such as infection, delays in wound healing, and patellar problems. For the coronal alignment, Table 3 showed the average angle and the mean absolute value of the difference between 3D evaluation and the conventional radiographic evaluation. Statistically significant differences were detected in absolute value of the difference (P b .0001). The CFA for 36 knees (97.3%) and the CTA for all the 37 knees (100%) were obtained within 3 from the optimal angle (Figs. 3 and 4). For the sagittal alignment, the average SFA was 86.5 1.8 (82.7-90.6). The mean differences from the optimal SFA was 0.5 1.7 (3.7 to 2.8); further, the SFA for 33 knees (89.2%) was obtained within 3 from the optimal SFA (Fig. 5). The average STA was 84.3 1.5 (81.0-87.0). The mean difference from the optimal STA was 1.3 1.4 (1.9 to 3.5), and the STA for 33 knees (89.2%) was obtained within 3 from the optimal STA (Fig. 6). For rotational alignment, the average RFA was 0.5 1.7 (3.2 to 3.4); moreover, the RFA for 34 knees (91.9%) was obtained within 3 from the optimal RFA (Fig. 7).

Discussion
Total knee arthroplasty has become one of the most successful orthopedic procedures. However, its success is dependent on many factors, including preoperative patient condition, implant materials, component design, and surgical technique. With respect to surgical technique, it is important to correctly align the femoral and tibial components and to balance the soft tissues adequately. However, it is quite possible that the femoral and tibial components are malpositioned by the conventional method [1-9]. It has been reported that an ideal positioning of the components can be achieved in 70% to 80% of the patients by using intramedullary or extramedullary alignment guides [5,7]. The entry points for intramedullary alignment guides may change the alignment [4,6]; further, the extramedullary alignment guide is easily affected by the condition of the ankle joint [2,5,9]. Navigation systems for TKA have been developed to achieve a greater accuracy than that obtained by the conventional method. Many clinical and experimental studies of these navigation systems have shown that the accuracy of implanted components has been improved [10-22]. However, most of these studies used 2D evaluations such as radiographs and/or CT scans (without 3D reconstruction), which may be affected by many factors such as positioning

Fig. 3. A distribution of varus/valgus alignment of the femoral component in relation to the mechanical axis of the femur.

Fig. 4. A distribution of varus/valgus alignment of the tibial component in relation to the mechanical axis of the tibia.

1108 The Journal of Arthroplasty Vol. 24 No. 7 October 2009

Fig. 5. A distribution of flexion of the femoral component in relation to the mechanical axis of the femur.

Fig. 7. A distribution of axial rotational alignment of the femoral component in relation to the SEA. Positive values imply external rotation.

of the legs, flexion angle of knees, and the scanning direction of radiation. It is important to measure the postoperative alignment 3-dimensionally for an exact evaluation. We established a new alignment analysis system by using the postoperative CT data. This study is the first study to evaluate postoperative alignments using a 3D model reconstructed from CT data. The postoperative alignments were measured on the basis of a defined plane that can be selected separately for the femoral and tibial components. The present study showed a discrepancy between 2D and 3D evaluations. The mean difference in absolute value was within 1; however, the maximum value was up to 3.4, which may be a critical error in evaluating postoperative alignments. This discrepancy may have been caused by a standing position, a flexion-extension of the knee joint, the direction

Fig. 6. A distribution of the posterior slope of the tibial component in relation to the mechanical axis of the tibia.

of radiation, and the rotation of the components. It may be difficult to obtain genuine coronal planes for both the femoral and tibial components in full-length weight-bearing anteroposterior radiographs when there is a rotational mismatch between these components. In this study, the coronal plane of the femur was defined by the mechanical axis of the femur and the SEA. The coronal plane of the tibia was defined by the mechanical axis of the tibia and the line connecting the posterior edge of the tibial component. These planes can reduce the adverse affect of this discrepancy. The sagittal planes were defined as a normal to the line connecting the geometric centers of the femoral component pegs and the line connecting the posterior edge of the tibial component. These planes can reduce the adverse affect of the varus-valgus position and rotational mismatch that occur frequently on lateral radiographs. The rotational plane of the femur was defined by the normal to the mechanical axis of the femur. Previous CT studies [13,21] evaluated the rotational alignment only from CT slices that are strongly dependent on the scanning direction. The results of this study demonstrated that a CT-based navigation system provided a reasonably satisfactory component alignment as planned preoperatively. In the case of the coronal alignment, the CFA and CTA were achieved within 3 from the optimal CFA and CTA in more than 95% of the procedures. We can conclude that for coronal alignment, the CT-based navigation system ensures the accuracy of positioning the components as reported by previous studies [18,19]. In the case of the sagittal alignment, the SFA and STA were achieved within 3 from the

Three-dimensional Analysis of CT-Based Navigation System for TKA  Mizu-uchi et al

1109

optimal SFA and STA in approximately 90% of the procedures. In the case of femoral rotational alignment, controversy still exists regarding the efficacy of a computer navigation system. Some studies [13,21] have reported an improved accuracy of rotational alignment with image-free navigation, but a large variability has also been reported [24,25]. On the other hand, few studies have so far shown the advantages of the CT-based navigation system. Our results demonstrated that for rotational alignment, the RFA was achieved within 3 of the optimal RFA in 91.9% of the procedures; moreover, the range (3.2 to 3.4) for this difference was narrow. Theoretically, in the detection of the epicondylar axis, a surface-matching method with CT data has advantages over manual detection that is used with image-free navigation systems. This study confirms that improved postoperative rotational alignment of the femoral component can be achieved by using CT-based navigation systems. In summary, this study is the first to evaluate postoperative alignments using a 3D model reconstructed directly from CT data. It is possible to measure the postoperative alignment for TKA more accurately on the basis of the defining plane. Further, the discrepancy with conventional radiograph analysis was up to 3.4. Our results thus demonstrated that good postoperative alignment was achieved as planned preoperatively with CTbased navigation systems.

6.

7.

8.

9.

10.

11.

12.

13.

14.

Acknowledgments
The author thank Kurata K, PhD, and Fukunaga T, Faculty of Engineering, Kyushu Sangyo University, Fukuoka city, Japan, for the help in analyzing the data.

15.

References
1. Bathis H, Perlick L, Tingart M, et al. Intraoperative cutting errors in total knee arthroplasty. Arch Orthop Trauma Surg 2005;125:16. 2. Dennis DA, Channer M, Susman MH, et al. Intramedullary versus extramedullary tibial alignment systems in total knee arthroplasty. J Arthroplasty 1993; 8:43. 3. Kinzel V, Ledger M, Shakespeare D. Can the epicondylar axis be defined accurately in total knee arthroplasty? Knee 2005;12:293. 4. Laskin RS. Instrumentation pitfalls: you just can't go on autopilot! J Arthroplasty 2003;18(3 Suppl 1):18. 5. Mahaluxmivala J, Bankes MJ, Nicolai P, et al. The effect of surgeon experience on component position-

16.

17.

18.

19.

ing in 673 press fit condylar posterior cruciatesacrificing total knee arthroplasties. J Arthroplasty 2001;16:635. Mihalko WM, Boyle J, Clark LD, et al. The variability of intramedullary alignment of the femoral component during total knee arthroplasty. J Arthroplasty 2005;20:25. Petersen TL, Engh GA. Radiographic assessment of knee alignment after total knee arthroplasty. J Arthroplasty 1988;3:67. Matsuda S, Mizu-uchi H, Miura H, et al. Tibial shaft axis does not always serve as a correct coronal landmark in total knee arthroplasty for varus knees. J Arthroplasty 2003;18:56. Mizu-uchi H, Matsuda S, Miura H, et al. The effect of ankle rotation on cutting error of the tibia in total knee arthroplasty. J Bone Joint Surg Am 2006; 88:2632. Amiot LP, Poulin F. Computed tomography-based navigation for hip, knee, and spine surgery. Clin Orthop Relat Res 2004;421:77. Bathis H, Perlick L, Tingart M, et al. Alignment in total knee arthroplasty. A comparison of computer-assisted surgery with the conventional technique. J Bone Joint Surg Br 2004;86:682. Bolognesi M, Hofmann A. Computer navigation versus standard instrumentation for TKA: a singlesurgeon experience. Clin Orthop Relat Res 2005; 440:162. Chauhan SK, Scott RG, Breidahl W, et al. Computerassisted knee arthroplasty versus a conventional jigbased technique. A randomised, prospective trial. J Bone Joint Surg Br 2004;86:372. Jenny JY, Clemens U, Kohler S, et al. Consistency of implantation of a total knee arthroplasty with a non-image-based navigation system: a case-control study of 235 cases compared with 235 conventionally implanted prostheses. J Arthroplasty 2005; 20:832. Kim SJ, MacDonald M, Hernandez J, et al. Computer assisted navigation in total knee arthroplasty: improved coronal alignment. J Arthroplasty 2005;20 (7 Suppl 3):123. Matziolis G, Krocker D, Weiss U, et al. A prospective, randomized study of computer-assisted and conventional total knee arthroplasty. Three-dimensional evaluation of implant alignment and rotation. J Bone Joint Surg Am 2007;89:236. Nabeyama R, Matsuda S, Miura H, et al. The accuracy of image-guided knee replacement based on computed tomography. J Bone Joint Surg Br 2004;86:366. Nizard RS, Porcher R, Ravaud P, et al. Use of the Cusum technique for evaluation of a CT-based navigation system for total knee replacement. Clin Orthop Relat Res 2004;425:180. Perlick L, Bathis H, Tingart M, et al. Navigation in total-knee arthroplasty: CT-based implantation compared with the conventional technique. Acta Orthop Scand 2004;75:464.

1110 The Journal of Arthroplasty Vol. 24 No. 7 October 2009

20. Sparmann M, Wolke B, Czupalla H, et al. Positioning of total knee arthroplasty with and without navigation support. A prospective, randomised study. J Bone Joint Surg Br 2003;85:830. 21. Stockl B, Nogler M, Rosiek R, et al. Navigation improves accuracy of rotational alignment in total knee arthroplasty. Clin Orthop Relat Res 2004; 426:180. 22. Victor J, Hoste D. Image-based computer-assisted total knee arthroplasty leads to lower variability in coronal alignment. Clin Orthop Relat Res 2004; 428:131. 23. Berger RA, Rubash HE, Seel MJ, et al. Determining the rotational alignment of the femoral component in total knee arthroplasty using the epicondylar axis. Clin Orthop Relat Res 1993;286:40. 24. Jenny JY, Boeri C. Low reproducibility of the intraoperative measurement of the transepicondylar axis during total knee replacement. Acta Orthop Scand 2004;75:74. 25. Siston RA, Patel JJ, Goodman SB, et al. The variability of femoral rotational alignment in total knee arthroplasty. J Bone Joint Surg Am 2005; 87:2276.