Injury, Int. J. Care Injured (2004) 35, S-A68—S-A78

Computer aided high tibial open wedge osteotomy Peter Keppler1, Florian Gebhard1, Paul A. Grützner2, Gongli Wang3, Guoyan Zheng3, Tobias Hüfner4, Stefan Hankemeier4, Lutz-Peter Nolte3 1 2 3 4

Abteilung für Unfallchirurgie, Chirurgische Universitätsklinik Ulm, 89075 Ulm, Germany Berufsgenossenschaftliche Unfallklinik, 67071 Ludwigshafen, Germany M.E. Müller Forschungszentrum für Orthopädische Chirurgie, 3001 Bern, Switzerland Abteilung für Unfallchirurgie, Medizinische Hochschule Hannover, 30623 Hannover, Germany

KEYWORDS: High tibial osteotomy; intraoperative plan; navigation; fluoroscopy; accuracy evaluation; computer aided orthopedic surgery (CAOS).

Summary1 High tibial osteotomy is a widely accepted treatment of medial compartment osteoarthritis as well as other lower extremity deformities. However, it is a technically demanding procedure. The lack of exact intraoperative real time control of the mechanical axis often results in postoperative malalignments, which is one reason for poor long term results. These problems can be addressed with the use of a surgical navigation system. Following exposure, dynamic reference bases (DRBs) are attached to the femur, and the proximal and the distal part of the tibia. After intraoperative measurement of the deformities and correction planning, the osteotomy is performed under navigational guidance. The wedge size, joint line orientation, and tibial plateau slope are monitored during correction. The in vitro evaluation with a plastic bone model suggests that the error of deformity correction is less than 1.7° (95% confidence limits) in the frontal, and less than 2.3° (95% confidence limits) in the sagittal plane, respectively. On a cadaver study of 13 legs, the mechanical axis intersected the Fujisawa line in 80.7% (range 77.5—85.8%). The preliminary clinical experience confirms these results. A novel computer tomography free navigation system for high tibial osteotomy has been developed that holds the promise of improving the accuracy, reliability, and safety of this kind of approach.

Introduction High tibial osteotomy (HTO) is a widely accepted treatment of unicompartmental osteoarthritis of the knee as well as other lower extremity deformities, particularly in young and active patients for whom total knee replacement is not advised [2, 3, 6, 12, 18, 22, 24, 25, 27, 34, 38]. There are two basic techniques available: (1) the lateral closed wedge procedure, and (2) the medial opening wedge technique. Medial opening wedge HTO has some advantages over the closed wedge procedure, such 1

Abstracts in German, French, Italian, Spanish, Japanese, and Russian are printed at the end of this supplement.

0020–1383/$ — see front matter ß 2004 Published by Elsevier Ltd. doi:10.1016/j.injury.2004.05.013

as single transverse osteotomy, no alteration of the fibula, minimum invasiveness, lower soft tissues alteration, and no loss of osseous substance, etc. [12, 13, 17, 18, 34]. Retrospective studies show that the main reasons for failure include the high variability of postoperative outcome, loss of correction, and damage to tibial dorsal neurovascular structures such as popliteal arterial injury [24, 25, 32]. With conventional techniques, it is difficult to obtain the ideal correction in terms of both accuracy and precision. Compared to the clinical requirement, which is obtaining the mechanical axis of lower extremity within a narrow range of ±2°, high variability of postoperative alignment can be observed [27]. By

Computer aided high tibial open wedge osteotomy

S-A69

evaluating the reported clinical results, the overall standard deviation of the postoperative axial alignment is 2.5° if HTO is performed with a jig system, and 5.6° without the use of jig [9]. One reason for the postoperative malalignment is unreliable preoperative planning [7]. The conventional full leg x-ray radiograph is inaccurate because either lower extremity rotation or image parallax may introduce measurement error [15, 23, 41]. Moreover, the wedge size not only depends on the static measurement of deformity, but also on dynamic angle analysis, as well as osteotomy plane position, and patient anatomical size [7]. So far, surgeons have had to handle these problems with arbitrary personal experience, which often results in either under or over estimation of the wedge size. Besides the determination of wedge size, surgical technique is very important because once the correction is achieved and the implant is in place, little can be done to adjust for the ideal correction. Although new emergence techniques provide a way of removing a wedge precisely, the wedge opening remains a problem. An asymmetric opening might produce significant unwanted secondary deformity, which can be either malrotation in the transverse plane or alteration of the tibial plateau slope in the sagittal plane. So far, no reliable intraoperative method exists for verifying the wedge size, orientation, and axial alignment. Accurate osteotomy is also essential. An inappropriate cut may damage neurovascular structures at the tibial dorsal side, or penetration of the lateral cortical bone. The latter can result in fixation failure or loss of correction, which is also one of the common complications in HTO approaches. Recently, the new implants and surgical techniques have emerged, such as the Puddu plate introduced by Fowler et al in 2000, and more recently, the TomoFix® demonstrated by De Simoni et al [10, 31, 35]. Computer aided orthopedic surgery (CAOS) technologies may potentiality address these problems. As a main contribution of previous works, various preoperative planning and simulation systems have been reported whereby the surgeon is provided with a simulation of the operation and a prediction of postoperative outcome [5, 21, 39]. One such example is the osteotomy analysis simulation system (OASIS) developed by Chao et al, which is a preoperative planning software for tibial and femoral osteotomy based on a full leg x-ray image using a two dimensional (2-D) rigid body spring model [5]. Recently, a three dimensional (3-D) surface model derived from a full leg computer tomography (CT) scan has been introduced to simulate the 3-D manner of HTO [40]. Navigated ultrasound (2.5-D) as an effective tool has

also been reported for the registration of patient anatomy and simulation of osteotomy [20]. Besides the preoperative planning, an intraoperative guidance system and robot-assisted system have also been reported, which aim to increase the accuracy and reproducibility of removing a wedge for closed wedge osteotomy [9, 30]. However, these systems generally rely on the preoperative CT scan, which has disadvantage in terms of radiation exposure and potential risk of infection caused by implanted fiducial markers. Recently, the pioneering clinical results of using the CT-free OrthoPilot® system (Aesculap AG, Tuttlingen, Germany) for HTO have been reported. However, this system only provides a percutaneous digitization method for anatomical landmark registration, and more importantly, the surgical planning still relies on the conventional full leg x-ray image [33]. For these reasons, a novel CT-free computer aided intraoperative planning and navigation system is proposed. Instead of a CT scan, a fluoroscopy C-arm is integrated because it is routinely used in the operating theater with conventional techniques. The goal of our proposed system is to address all of the above-mentioned problems, thus making HTO an accurate, safe, and reproducible procedure. The main issues include: (a) accurate and reproducible intraoperative measurement of deformity, therefore eliminating the reliance on conventional x-ray radiographs; (b) intuitive intraoperative planning based on fluoroscopic images, thus easing the surgeon’s life, and (c) navigational guidance of osteotomy and opening wedge, therefore helping the surgeon to perform surgical procedures accurately.

Material and methods System components The system is based on the SurgiGATE® system (MediVision, Oberdorf, Switzerland). An optoelectronic infrared tracking localizer (OptoTrack 3020, Northern Digital Inc., Waterloo, Ontario, Canada), mounted on a movable stand, tracks the position of optical targets equipped with infrared light-emitting diodes (LEDs). These targets are attached to the surgical objects and the image intensifier of the C-arm, as well as other relevant surgical instruments. A Sun ULTRA 10 workstation (Sun Microsystems Inc., Mountain View, Canada) is chosen for the image processing and visualization task. According to the functionalities, there are three main parts of the navigation instruments. The first

S-A70

set comprises a calibrated fluoroscopy C-arm, a gravity direction measurement tool, an accuracy check for the verification of overall system accuracy, and a pointing device (pointer) for percutaneous digitization. This set of instruments is used for image acquisition and landmark registration. The next set consists of a surgical oscillating saw and an osteotomy chisel, which are used for cutting the bone. The final set comprises three DRBs, which are fixed at the femur, and the proximal and distal part of the tibia by 2.8 mm K-wires for tracking the motion of the corresponding anatomy when the leg is flexed, extended, and rotated. According to the surgical procedure, the system consists of the following steps: a) image acquisition and landmark registration, b) deformity measurement, c) planning of the osteotomy plane and wedge size, and d) navigational guidance of bone cutting and wedge opening.

P. Keppler et al

posterior condyles axis is defined as the connecting line of extreme point at lateral and medial femoral posterior condyles [1]. The tibial plateau size is registered with the most lateral, medial, anterior, and posterior point at the tibial plateau border. These landmarks are registered with bi-plane fluoroscopic reconstruction using knee AP and knee lateral images [14]. The knee center is calculated as the center of the tibial plateau [23]. At the ankle joint, the ankle center is defined as the middle point of talus, which can be registered with either fluoroscopic images or with percutaneous digitization by using a palpable pointer [23]. The ankle center is calculated as the middle point of the transmalleolar axis formed by the extreme point at the lateral and medial malleoli.

Deformity measurement

The anatomical landmarks are registered intraoperatively to represent surgical objects as well as to calculate the clinical parameters. A previously introduced hybrid concept is adapted to the system, which involves kinematic pivoting movement, pointbased percutaneous digitization by using a pointer, and fluoroscopic image-based bi-plane 3-D point reconstruction [14, 42]. At the hip joint, the hip center is defined as the spherical center of the femoral head, which can be registered with either a pivoting movement or with bi-plane 3-D point reconstruction using hip AP and hip lateral images (Fig. 2) [14, 23]. When the pivoting movement is used, a pointer is stuck at the anterior superior iliac spine of the pelvis to track the motion of the pelvis during the pivoting movement of the lower limb. At the knee joint, the femoral

The deformity of the affected limb is measured intraoperatively, which includes varus angle, flexion angle, tibial plateau slope, and medial proximal tibial angle. We do not rely on the conventional full leg radiograph because either malrotation of the lower extremity or the distortion of the image caused by parallax may introduce an error to the measurement [15, 23, 41]. Checking soft tissue balance is important [27]. The anatomical abnormalities of the knee joint not only consist of primary varus that refers to the tibiofemoral osseous geometry or damage of cartilage, but also secondary varus, resulting from the separation of the lateral tibiofemoral compartment due to the deficiency of soft tissues. It is important to differentiate between these two because only the primary deformity should be corrected with HTO [8, 27, 28]. The soft tissue balance is evaluated with the help of the joint line convergence angle (JLCA), which is

Fig. 1: System hardware comprise (a) optoelectric localizer and computer system, (b) fluoroscopy C-arm, (c) DRBs for patient anatomy tracking, (d) gravity direction measurement tool, (e) accuracy checker for the verification of system accuracy, (f) pointer for percutaneous digitization, (g) navigated saw, and (h) navigated chisel for bone cutting.

Fig. 2: Fluroscopic images used for landmarks registration, which comprise knee AP, knee lateral, and optionally hip AP, hip lateral, ankle AP, and ankle lateral image.

Image acquisition and landmark registration

Computer aided high tibial open wedge osteotomy

S-A71

formed by the intersection angle of the tibial and femoral joint line at the frontal plane [29]. By applying varus, valgus stress, and an upward push at the tibia, the surgeons can easily measure osseous deformities along with any ligament laxity (Fig. 3) [28].

weight-bearing axis; the software then automatically computes the wedge size, the overcorrection angle, and the increment of leg length accordingly. For example, for unicompartmental osteoarthritis of grade I and II, the target position of the weightbearing axis is 62.5% at the coordinate of the tibial plateau (0% corresponded to the media border and 100% to the lateral border), which approximates about 3° valgus overcorrection [27].

Intraoperative planning Planning of the osteotomy plane

Navigational guidance The planning of an osteotomy is performed with the help of fluoroscopic images, which includes the determination of the position and orientation of cutting plane at the frontal and sagittal planes. The cutting plane is superimposed on fluoroscopic images (Fig. 5), so that the surgeons can easily define the cutting plane according to clinical relevancies such as geometry of deformity, type of fixation, soft tissue coverage, bone quality, etc. The adjusted parameters consist of orientation at the frontal and sagittal planes, distance to the tibial plateau, length of the osteotomy plane, and the entry and end point of the osteotomy.

Planning of wedge size and orientation

Bone cutting The accuracy and safety of the osteotomy cannot be overemphasized. The lateral cortical bone hinge is vital for the stability and maintenance of the correction. Therefore, the penetration of the lateral cortex should be avoided. Inadequate surgical techniques such as poor lateral visualization, underestimation of the saw blade depth, or incorrect determination of osteotomy width may cause such problems. Tibial plateau fracture is another potential complication that can result from inappropriate osteotomy such as insufficient proximal fragment thickness, incomplete bone cut, or unintentional upper-lateral obliquity of the saw blade.

For the correction of pure frontal plane deformity (Fig. 4), which is the most common case of HTO, the surgeon simply specifies the target position of the

Fig. 3: Femoral and tibial joint lines and joint line convergence angle (JLCA). A biased JLCA value indicates that the measured deformity not only consists of primarily osseous deformity but also secondary deformity resulted from lateral soft tissues laxity. In this case, the soft tissues need to be balanced. The normal range of JLCA falls into 0.7º — 2.4º according to Paley [29].

Fig. 4: For the correction of frontal plane deformity, wedge size is determined according to the target position of weight-bearing axis at tibia plateau, where H is hip center, A is ankle center, T is the target position of weight-bearing axis, P is the position of hinge axis, G is corresponding target position of ankle center, and is the wedge size.

S-A72

Moreover, damage to the neurovascular structure might occur if the osteotomy is excessive at the tibial dorsal side. The navigational guidance (Fig. 5) is therefore provided to avoid these intraoperative pitfalls. With on-the-fly visualization, the position and direction of instruments can be controlled from multiple fluoroscopy images simultaneously. During the osteotomy, the surgeon is able to monitor the incision of the instruments continuously. With real-time feedback of the deviation of the instrument with the planned cutting plane, and the feedback of the distance of the instrument tip to the lateral cortical hinge, a safe and accurate osteotomy can be expected.

Wedge opening The opening of the wedge is one of the most crucial and difficult steps, because once the correction is achieved and the implant is in place, little can be done to adjust for the ideal anticipated correction. Excessive or insufficient opening at the osteotomy site, particularly the asymmetric opening, may introduce unwanted secondary deformity, which can be either malrotation in the transverse plane or alteration of the tibial plateau slope in the sagittal plane. With our proposed navigation system, however, surgeons are fully provided with feedback of the wedge size and wedge orientation, as well as axial alignment. When the wedge is being opened, the system allows surgeons to continually analyze and monitor the progress of the deformity correction.

Fig. 5: Osteotomy is performed under navigational guidance. The incision of instrument (eg, TomoFix® chisel, which is implemented as shin red line for the contour, and green polyhedron for the tip), as well as planned cutting plane (implemented as blue line for the direction at frontal plane, and red line for the direction at sagittal plane) can be visualized continuously from multiple fluoroscopic images.

P. Keppler et al

The 3-D wedge is decomposed and displayed as three components in frontal, sagittal, and transverse planes, which are then compared with the planned value using sliders (Fig. 6). Therefore, they are aware of any unintended asymmetric opening. The axial alignment and the joint line orientation, as well as the position of the weight-bearing axis at the tibial plateau coordinate are also navigated.

In-vitro plastic bone validation The proposed system is validated in the laboratory with a plastic bone model that aims to establish the accuracy and precision in terms of interobserver repeatability, intraobserver reproducibility, and end-to-end application accuracy. The mathematical simulation is used to calculate the mean error of application in the normal case and the maximum deviation in the worst case scenario.

Plastic bone model and test bench The plastic bone model is created with an entire normal leg (RR0119, Synbone, Switzerland). The proximal and distal part of the tibia are cut through and connected with a rubber hinge after removing a wedge along the AP and lateral/medial (LM) directions, respectively

Fig. 6: Wedge is opened under navigational guidance. The navigated parameters comprises components of wedge size in frontal, sagittal, and transverse plane, shown at the bottom of the window, the collinear parameters including varus/valgus angle, extension/flexion angle, and internal/external rotation angle of tibia in relation to the femur, the tibial plateau slope angle, medial proximal tibial angle (MPTA), the position of weight-bearing axis at the tibial plateau coordinate, shown at the top right of the window, and 3-D view of wedge and mechanical axes, shown at the top left of the window.

Computer aided high tibial open wedge osteotomy

S-A73

(Fig. 7). Thus a juxta-articular tibial deformity is simulated, which is a common case in HTO approach. All images used for landmark validation are acquired with this artificial model. The plastic bone is then mounted at the platform of the test bench with multi-freedom fixtures. The distal part of the tibia is connected to a measurement gauge through a spherical joint with which the position of the distal tibia can be measured. Uni- or multiplane deformities are created when the distal tibia is moved relative to the proximal part. When it is moved horizontally, a pure fontal plane varus deformity is created; when moved vertically, a sagittal plane deformity is created. If horizontal movement is combined with vertical movement, multiplane complex deformities, which exist at both frontal and sagittal planes, are created.

of the landmark position is defined as the mean position of these repeated measurements.

Interobserver repeatability In order to evaluate the robustness when landmarks are registered with different operators, sets of eleven repeated measurements were made by five independent operators with identical experimental conditions. The inter-observer repeatability is defined as the consistency of the individual landmark positions and the calculated functional parameters registered with each operator. The five operators present different skills. Operator #1 has an extensive clinical experience as well as knowledge of computer aided navigation systems. Operator #2 also has significant clinical experience but limited familiarity with navigation software. Operator #3 is an orthopedic specialist but has limited surgical experience and no experience with navigation software. Operators #4 and #5 have extensive experience of navigation software but no clinical experience. The reference positions of the landmarks are directly digitized on the plastic bone model by four independent operators with a palpable pointer. Each operator repeats each landmark six times. The value

Intraobserver reproducibility The bi-plane 3-D point reconstruction at the knee joint involves an AP and a lateral fluoroscopic image. The accuracy of this approach can be up to submillimeters based on phantom evaluation [14]. However, intraoperatively, it is cumbersome, and sometimes even impossible, to obtain an accurate AP and lateral fluoroscopic image. The misalignment of the C-arm orientation may make the projectional representation of the anatomical structures difficult to interpret, and thus potentially introduce error to the landmark registration. In order to assess such error, five images are acquired at AP and the lateral direction of the knee joint, which consist of one well-aligned image and four misalignment images. The misaligned images are obtained with the C-arm rotated 10° towards the proximal, distal, lateral, and medial directions in relation to the well-aligned position. The wellaligned AP image is defined as the patella projected at the center of femoral condyles; the well-aligned lateral image is defined as the lateral and medial femoral condyles overlapping each other. In a previous cadaver study [37], Wright et al demonstrated that the accuracy of achieving the true knee forward position using the patella was within 5°. Therefore, we believe that a 10° misalignment can simulate the worst clinical situation [41]. The landmarks are then registered with these images, which comprises 25 pair combinations formed by five AP and five lateral images. The intraobserver reproducibility is defined as the consistency of these registered landmark positions and the calculated functional parameters. Cadaver study In a preliminary study, the accuracy of navigated

Fig. 7: Test bench used for the evaluation of end-to-end application accuracy, which comprises (1) DRB (Dynamic Reference Base) of femur, (2) DRB of proximal tibia, (3) DRB of distal tibia, (4) multi-freedom fixtures, (5) measurement gauge that enable free movement of distal tibia inside the frame, (6) rulers used to measure the position of distal tibia, and (7) spherical joint that connects the distal tibia with the measurement gauge.

S-A74

HTO was analyzed. 13 legs of seven fresh cadavers were randomly assigned to navigated HTO (n=7) or conventional HTO (n=6). The main aim of the study was to align the mechanical axis to pass through 80% of the tibial plateaus (80% Fujisawa line) [11]. Secondary outcome parameters were the operation time and radiation dosage during the whole surgical procedure. An angle-fixed implant with interlocking screws, TomoFix®, was used for stabilization to minimize postoperative loss of correction. The intraoperative mechanical axis was examined either by the ‘cable technique’ for conventional HTO, or by the navigation module in the navigation group [19]. Postoperatively, CT scans were performed and the mechanical axis was measured with special 2-D computer software for preoperative planning (MediCAD®, Hectec, Altfraunhofen, Germany). Preliminary clinical trial The preliminary clinical trial was started on the basis of the encouraging in vitro and cadaver studies. From August to December 2003, five patients (two men and three women) with medial osteoarthritis or congenital varus deformity of the knee joint were operated on with the proposed navigation system by one of the authors (P. Keppler). The average age at the time of operation was 46 years (range, 23— 65 years). The left side was involved in two patients and the right side in three. They were managed with medial open valgus osteotomy with the use of the TomoFix® technique. Ultrasound as an effective 2.5-D tool for registering patient anatomy has been described [16]. A previous study demonstrated that such approach has an accuracy of 1° and standard deviation (95% confidence limits) of 2°. Although we did not actually need the preoperative planning, the ultrasound system was still used for the preoperative measurement of patient deformity and postoperative measurement of axial alignment in our primary clinical trial,

Fig. 8: The interobserver repeatability of functional parameters is evaluated by five independent operators with identical experimental condition.

P. Keppler et al

for the purpose of verification. The conventional radiograph evaluation was also made 4— 6 weeks after the intervention with full leg x-ray images (including the hip, knee, and ankle joint) with the patient standing and with the x-ray beam centered at the knees with the patellae oriented forward [8]. The measurements made from the ultrasound and radiograph were compared with the determinations of the navigation system.

Results In vitro plastic bone validation There was a complete consistency between the operators in registering all landmarks, and the calculation of clinical parameters, irrespective of their surgical experience and knowledge of a navigation system. The maximum interobserver difference of mean value was found to be 0.3° for varus, 0.6° for flexion, 1.0° for tibial plateau slope, and 0.8° for medial proximal tibial angle (MPTA), with the standard deviation (SD) of 95% confidence limits of 0.8°, 1.2°, 1.6°, and 1.2°, respectively (Fig. 8). The maximum discrepancy of the mean value with the reference value is 0.5° for the varus angle and 0.6° for the flexion angle. The position of the weight-bearing axis at the tibial plateau is meaningful from the biomechanical point of view. We found that the SD with 95% confidence limits is 3.1% for all operators, with a maximum interobserver difference of 2.8%. The maximum range of a single operator is 5.5%. With 25 pair images acquired at misaligned positions of up to 10o, the landmarks are registered and functional parameters are calculated. The mean deviation was found to be 0.4° for varus, 0.4° for flexion, 1.0° for tibial plateau slope, and 0.7° for MPTA, with the SD of 95% confidence limits of 0.6°, 0.8°, 1.2°, and 1.2°, respectively (Fig. 9). The

Fig. 9: The intraobserver reproducibility of functional parameters is evaluated with 25 pair fluoroscopic images. All measurements, ie, 11 repeated measurements for each pair, are performed with single operator.

Computer aided high tibial open wedge osteotomy

S-A75

maximum deviations are 0.9° for the varus angle, 1.1° for the flexion angle, 2.0° for tibial plateau slope, and 2.1° for MPTA. The mean discrepancy of the position of the weight bearing axis at the tibial plateau coordinate is 1.1%, with the SD of 95% and confidence limits of 2.5%.

be within this critical region with more than a 95% confidential limit.

Mathematical simulation Based on the interobserver and intraobserver landmark validation, a data cube is defined for each landmark, which is located at the mean position of each landmark with the size equal to three times the standard deviation in LM (lateral medial), AP (anterior posterior), and PD (proximal distal) direction. The cube represents the possible registration positions of each landmark. With the polar form of the Box-Müller transformation, which is an algorithm to generate randomized points according to normal distribution, 100 points are generated inside each cube [4]. After statistical analysis of all data combinations, the application accuracy with 95% confidence limits that reflects the normal case and maximum error in the worst case scenario are reported. We found that the mean error is 0.7° for varus and 0.9° for flexion, with the SD of 95% confidence limits of 1.0° for the varus angle and 1.4° for the flexion angle. In the worst case scenario, the maximum deviation of the varus and flexion angles is 2.6° and 3.6°, respectively. However, it corresponds to a possibility of less than 0.1%. Any discussion of accuracy would be incomplete without mentioning the clinical requirement. The retro-operative clinical study suggested that the goal of HTO is to obtain the axial alignment of the lower extremity at the frontal plane within a narrow range of ±2°, in order to get good long-term result [27]. Compared to the mathematical simulation, we can say that the clinical outcome is expected to

Preliminary cadaver study After conventional HTO, the mechanical axis intersected the Fujisawa line at 74.2% (range 60.4— 82.4%, SD 9.7%). In contrast, after the navigated HTO, the tibia plateau was passed through 80.7% (range 77.5— 85.8%, SD 3.1%). The completed operation time was 58 min (42— 84 min) for conventional HTO and 75 min (55— 87 min) for navigated HTO. Radiation dosage was reduced in the navigation group (average 41.8 cGy/cm2, range 28.3— 57.6 cGy/cm2) compared to conventional HTO (average 55.2 cGy/ cm2, range 36.0— 81.2 cGy/cm2). Particularly due to the improved accuracy of the corrective operation, navigation of the opening wedge HTO offers essential advantages.

Preliminary clinical trial All five patients had a varus deformity that ranged from 6.0–9.0° before the operation. The target position of the weight-bearing axis is determined by the surgeon according to clinical and arthroscopic relevancies. All patients were successfully supported with the above-described navigation system. The hip, knee, and ankle centers were determined by fluoroscopy C-arm. No complication was found in any patients. The average extra operation time is 30–60 minutes, which is mainly used for DRB fixation, fluoroscopic image acquisition and positioning of the OPTOTRAC camera. But there was a steep learning curve. Compared to the intended postoperative axial alignment, the mean error was found to be 1.0°, with a maximum error of 2° in all patients (Table 1).

PreOP No.

Diagnosis

Age

Sex

Side

1

Postraumatic gonarthrosis

39

m

2

Medial gonarthrosis

55

3

Congenital Genu varum

4 5

PostOP

Ultrasound

Navigation

Ultrasound

Navigation

Radiograph

left

172°

173°

182°

182°

181°

w

right

173°

173°

181°

183°

180°

23

w

right

174°

173°

179°

179°

180°

Medial gonarthrosis

65

m

left

171°

172°

184°

183°

181°

Posttraumatic gonarthrosis

46

m

right

173°

172°

182°

180°

178°

Table 1: Pre- and postoperative evaluation of the mechanical axis measured by 2.5-D ultrasound, navigation and long standing x-rays.

S-A76

Discussion Theoretically, a pivoting algorithm is sufficiently accurate to calculate the center of the hip [36]. However, clinically, many factors may limit the availability of this method, such as a limited range of motion, obesity of the patient, or a non-spheroid femoral head. Also, patients with severe dysplasia, ankylosis, or any other abnormalities of the hip joint are probably not applicable for such surgical procedures. In these cases, registration of the hip center with fluoroscopic images should be used, which is more accurate and reproducible compared with a pivoting algorithm according to our plastic bone evaluation. However, in this case, two more fluoroscopic images at the hip joint have to be acquired, which are unnecessary with the pivoting algorithm. Several previous studies 44— 45 have reported on the accuracy of ankle center registration based on percutaneous digitization [26, 37]. Our plastic bone evaluation confirms that digitization of the extreme point at the lateral and medial malleoli is a simple and reliable method. However, if a patient is too obese to palpate the malleolus, or if a deformity exists at the ankle joint, the bi-plane fluoroscopic reconstruction should be used. Moreland et al have already evaluated the different geometric methods to define the knee center in the frontal plane [23]. They reported that it is approximately the same when using the top of the femoral notch, the middle point of the femoral condyles, the center of the tibia spines, or the center of tibia plateau. However, no publication has addressed the reliability and reproducibility of the registration when bi-plane fluoroscopic reconstruction is used. We found that the borders of the lateral and medial tibial plateaus are the most reproducible landmarks compared with other landmarks such as tibial spine or femoral notch, because they can be unambiguously and objectively selected at fluoroscopic image. The SD of the repeated measurements (n = 21) with same fluoroscopic image was found to be 0.5 mm in LM and 1.5 mm in AP direction. However, if previously described misaligned images are used, the Euclidan length between the maximum and minimum position can be up to 5 mm in LM and 13.9 mm in the AP direction, which corresponds to an angular deviation of 1.5° and 3.9°, respectively. Such a large deviation results from the misinterpretation of the 2-D projection of the 3-D anatomic structures of the tibia plateaus. However, these errors tend to eliminate each other when the knee center is calculated, which is defined as the middle point of the tibial plateaus, because the tibial plateaus are almost symmetrical in the LM direction with a circular shape. We found

P. Keppler et al that the maximum discrepancy with the reference point is only 1.5 mm in the frontal and 2.9 mm in the sagittal planes, which corresponds to angular deviations of 0.5° and 0.9°, respectively. The wedge opening is a critical procedure that involves not only the control of wedge size, but also the correct placement of the hinge axis and wedge orientation. The hinge axis of a wedge is a line perpendicular to the plane of the angular deformity. From the technical point of view, to perform a correction strictly at the frontal plane (eg, varus), the opening of the osteotomy should be symmetrical in the anterior and posterior direction; and the hinge should be placed at the lateral cortical bone, perpendicular to the frontal plane. In contrast, for a correction at the sagittal plane (eg, recurvatum), the opening should be strictly symmetrical in the lateral and medial direction, and the hinge should be located at the posterior tibial cortical bone, perpendicular to the sagittal plane. The accurate placement of the lateral hinge axis is important because too much medial placement can make the opening procedure difficult, or even worse, result in tibial plateau fracture, while too much lateral placement will disrupt the lateral cortex bone, which may result in instability of the fixation during healing. With the conventional approach, it is difficult to accurately control the hinge position even with the jig system because it can only control the size of wedge angle. The orientation of the wedge is also very important because the asymmetrical opening of the wedge would introduce unwanted secondary deformity. For example, asymmetrical opening at the anterior posterior direction would change the tibial plateau slope, which may result in anterior or posterior instability due to excessive posterior or anterior slope. An important clinical advantage of the computer aided navigation system is the accurate control of the hinge axis and wedge orientation. With on-thefly visualization of projections of the instruments, the position of the hinge axis can be accurately controlled. With real-time feedback of wedge components in frontal, sagittal, and transverse planes, any misorientation of the wedge can be detected. In contrast, with conventional techniques, the surgical plan is derived from only the frontal plane x-ray film, and there is an assumption that the hinge axis is always perpendicular to the frontal plane. The intended kinematics of the plan will only be reproduced in the theater if this assumption is fulfilled. Therefore, if the hinge axis is somewhat rotated, which might happen because surgeons have no effective way of controlling the wedge orientation, the correction obtained in the frontal plane will not

Computer aided high tibial open wedge osteotomy

S-A77

be the one that was anticipated, and the unwanted alteration of tibial plateau slope in the sagittal plane or internal/external rotation of foot in the transverse plane would be induced. This situation would explain certain errors of postoperative outcome, even if the preoperative planning was accurate. The preliminary clinical results are very encouraging. Although the number of patients operated on is too small to show any statistically significant differences, it seems that navigation of the opening wedge HTO offers essential advantages. The system allows surgeons to measure the deformity during the operation, plan the surgical procedure, as well as perform the osteotomy under navigational guidance. This method is substantially different from other computer aided approaches for osteotomy planning or robotic preparation. Moreover, it does not require any additional image procedures or ancillary operative procedures, nor does it really require any information from preoperative plain radiographs. Currently, this software only focuses on the medial open valgus tibial osteotomy, which is one of the most common clinical cases. However, we are in the progress of extending the current module to the closing and Maquet (dome) type osteotomy at the tibial side, as well as the femoral and double osteotomy [2, 3, 22, 38]. The last of these osteotomies is much more difficult to correct with conventional techniques. In conclusion, while complications in HTO may occur, they can be eliminated or minimized with computer aided navigation techniques. The intraoperative planning and navigational guidance are able to reduce the potential complication risk and thus increase the safety, accuracy, and reproducibility, and improve the long term results of this kind of surgical osteotomy.

6. Coventry MB, Ilstrup DM, Wallrichs SL (1993) Proximal tibial osteotomy. A critical long-term study of eighty-seven cases. J Bone Joint Surg [Am]; 75: 196— 201.

References 1. Arima J, Whiteside LA, McCarthy DS, et al. (1995) Femoral rotational alignment, based on the anteroposterior axis, in total knee arthroplasty in a valgus knee. A technical note. J Bone Joint Surg [Am]; 77: 1331— 1334. 2. Babis GC, An KN, Chao EY, et al. (2002) Double level osteotomy of the knee: a method to retain joint-line obliquity. Clinical results. J Bone Joint Surg [Am]; 84-A: 1380— 1388. 3. Billings A, Scott DF, Camargo MP, et al. (2000) High tibial osteotomy with a calibrated osteotomy guide, rigid internal fixation, and early motion. Long-term follow-up. J Bone Joint Surg [Am]; 82: 70— 79. 4. Box G, Müller M (1958) A note on the generation of random normal deviates. Ann Math Stat; 29: 610— 611. 5. Chao EY, Sim FH (1995) Computer-aided preoperative planning in knee osteotomy. Iowa Orthop J; 15: 4— 18.

7. Dahl MT (2000) Preoperative planning in deformity correction and limb lengthening surgery. Instr Course Lect; 49: 503— 509. 8. Dugdale TW, Noyes FR, Styer D (1992) Preoperative planning for high tibial osteotomy. The effect of lateral tibiofemoral separation and tibiofemoral length. Clin Orthop; 248— 264. 9. Ellis RE, Tso CY, Rudan JF, et al. (1999) A surgical planning and guidance system for high tibial osteotomy. Comp Aid Surg; 4: 264— 274. 10. Fowler PJ, Tan JL, Brown G (2004) Medial opening wedge high tibial osteotomy: how I do it. Oper Tech Sport Med; 8: 32— 38. 11. Fujisawa Y, Masuhara K, Shiomi S (1979) The effect of high tibial osteotomy on osteoarthritis of the knee. An arthroscopic study of 54 knee joints. Orthop Clin North [Am]; 10: 585— 608. 12. Hernigou P, Ma W (2001) Open wedge tibial osteotomy with acrylic bone cement as bone substitute. Knee; 8: 103— 110. 13. Hernigou P, Medevielle D, Debeyre J, et al. (1987) Proximal tibial osteotomy for osteoarthritis with varus deformity. A ten to thirteen-year follow-up study. J Bone Joint Surg [Am]; 69: 332— 354. 14. Hofstetter R (1999) Fluoroscopy as an imaging means for computer assisted surgical navigation. Comp Aid Surg; 4: 65— 76. 15. Jiang CC, Insall JN (1989) Effect of rotation on the axial alignment of the femur. Pitfalls in the use of femoral intramedullary guides in total knee arthroplasty. Clin Orthop; 248: 50— 56. 16. Keppler P, Strecker W, Kinzl L, et al. (1999) Die sonographische Bestimmung der Beingeometrie. Orthopäde; 28: 1015— 1022. 17. Koshino T, Murase T, Saito T (2003) Medial opening-wedge high tibial osteotomy with use of porous hydroxyapatite to treat medial compartment osteoarthritis of the knee. J Bone Joint Surg [Am]; 85-A: 78— 85. 18. Koshino T, Murase T, Takagi T, et al. (2001) New bone formation around porous hydroxyapatite wedge implanted in opening wedge high tibial osteotomy in patients with osteoarthritis. Biomaterials; 22: 1579— 1582. 19. Krettek C, Miclau T, Grun O, et al. (1998) Intraoperative control of axes, rotation and length in femoral and tibial fractures. Technical note. Injury; 29 (suppl 1): C29— C39. 20. Krivonos O, Moskalenko D, Keppler P, et al. (2001) Computer simulation of osteotomy correction. Stud Health Technol Inform; 81: 247— 249. 21. Lin H, Birch JG, Samchukov ML, et al. (1995) Computer-assisted surgery planning for lower extremity deformity correction by the Ilizarov method. J Image Guid Surg; 1: 103— 108. 22. Mathews J, Cobb AG, Richardson S, et al. (1998) Distal femoral osteotomy for lateral compartment osteoarthritis of the knee. Orthopedics; 21: 437— 440. 23. Moreland JR, Bassett LW, Hanker GJ (1987) Radiographic analysis of the axial alignment of the lower extremity. J Bone Joint Surg [Am]; 69: 745— 749.

S-A78

24. Murphy SB (1994) Tibial osteotomy for genu varum. Indications, preoperative planning, and technique. Orthop Clin North [Am]; 25: 477— 482. 25. Naudie D, Bourne RB, Rorabeck CH, et al. (1999) The Install Award. Survivorship of the high tibial valgus osteotomy. A 10- to -22-year followup study. Clin Orthop; 367: 18— 27. 26. Nofrini L (2003) Estimation of Accuracy in Ankle Center Location for Tibial Mechanical Axis; Identification using computer assisted Surgery Systems. Comp Ass Orthop Surg; 1: 67. 27. Noyes FR, Barber-Westin SD, Hewett TE (2000) High tibial osteotomy and ligament reconstruction for varus angulated anterior cruciate ligament-deficient knees. J Sports Med [Am]; 28: 282— 296. 28. Paley D, Bhatnagar J, Herzenberg JE et al. (1994) New procedures for tightening knee collateral ligaments in conjunction with knee realignment osteotomy. Orthop Clin [Am]; 25: 533— 555. 29. Paley D, Pfeil J (2000) Prinzipien der kniegelenknahen Deformitätenkorrektur. Orthopäde; 29: 18— 38. 30. Phillips R, Hafez MA, Mohsen AM, et al. (2000) Computer and robotic assisted osteotomy around the knee. Stud Health Technol Inform; 70: 265— 271. 31. Puddu G, Franco V (2000) Femoral antivalgus opening wedge osteotomy. Oper Tech Sport Med; 8: 32— 38. 32. Rudan JF, Simurda MA (1991) Valgus high tibial osteotomy. A long-term follow-up study. Clin Orthop; 157— 160. 33. Saragaglia D, Pradel P, Chaussard C, et al. (2003) Orthopilot assisted osteotomy in osteoarthritic genu varus— results of the axial correction in case-control study of 56 cases. Comp Ass Orthop Surg; 1: 314. 34. Siguier M, Brumpt B, Siguier T, et al. (2001) Osteotomie tibiale de valgisation originale par ouverture interne sans perte de contact osseux. Technique et incidence sur la rapidite de consoliation: a propos des 33 premiers cas. Rev Chir Orthop Reparatrice Appar Mot; 87: 183— 188. 35. Staubli AE, De Simoni C, Babst R, et al. (2003) TomoFix: a new LCP-concept for open wedge osteotomy of the medial proximal tibia— early results in 92 cases. Injury; 34 (suppl 2): B55— B62. 36. Stindel E (2002) Detection of the Hip in Computer Assisted TKA: an Evaluation Study of the Accuracy and Reproducibility of the Surgetics Algorithm. Comp Ass Orthop Surg; 1: 65— 37. Stindel E (2002) The Center of the Ankle in CT-less Navigation System: What is really important to detect? Comp Ass Orthop Surg; 1: 63. 38. Sundaram NA, Hallett JP, Sullivan MF (1986) Dome osteotomy of the tibia for osteoarthritis of the knee. J Bone Joint Surg [Br]; 68: 782— 786. 39. Tsumura H, Himeno S, Kawai T (1984) The computer simulation on the correcting osteotomy of osteoarthritis of the knee. J Jap Orthop Assoc; 58: 565— 566. 40. Wirz P, Jakob RP (1999) Knee joint simulator: an anatomical reconstruction of the joint surfaces and of the ligamentous structures of the knee joint for teaching purposes. Knee Surg Sports Traumatol Arthrosc; 7: 59— 62. 41. Wright JG, Treble N, Feinstein AR (1991) Measurement of lower limb alignment using long radiographs. J Bone Joint Surg [Br]; 73: 721— 723.

P. Keppler et al

42. Zheng G, Marx A, Langlotz U (2002) Orthopilot assisted osteotomy in osteoarthritic genu varus— results of the axial correction in case-control study of 56 cases. Comp Ass Orthop Surg; 1: 129— 145.

Corresponding address: Dr. Peter Keppler Department of Traumatology, Hand and Reconstructive Surgery University of Ulm Steinhövelstraße 9 89075 Ulm, Germany phone: +49 731 500 33423 fax: +41 731 500 27349 email: [email protected]