THE INTERNATIONAL JOURNAL OF MEDICAL ROBOTICS AND COMPUTER ASSISTED SURGERY ORIGINAL Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. Published online 21 September 2014 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/rcs.1614
ARTICLE
Advancing computer-assisted orthopaedic surgery using a hexapod device for closed diaphyseal fracture reduction
Hailong Du1 Lei Hu2 Changsheng Li2 Tianmiao Wang2 Lu Zhao2 Yang Li2 Zhi Mao1 Daohong Liu1 Lining Zhang1 Chunqing He1 Licheng Zhang1 Hongping Hou1 Lihai Zhang1*† Peifu Tang1*† 1
Department of Orthopaedics, Chinese PLA General Hospital, Beijing, People’s Republic of China
2
Robotics Institute, Beihang University, Beijing, People’s Republic of China *Correspondence to: P. Tang, L. Zhang, Department of Orthopaedics, Chinese PLA General Hospital, 28 Fuxing Road, Haidian District, Beijing 100853, People’s Republic of China. E-mail: [email protected]; [email protected]
†
These authors contributed equally to this study.
Accepted: 13 August 2014
Copyright © 2014 John Wiley & Sons, Ltd.
Abstract Background Surgical complications such as healing problems, in fractures treated using the Arbeitsgemeinschaft für Osteosynthesefragen (AO) technique, present functional and economic challenges to patients and treatment dilemmas for surgeons. Computer-assisted orthopaedic surgery using minimally invasive techniques focused on biological osteosynthesis is a novel direction for fracture treatment. Method We modified the hexapod computer-assisted fracture reduction system by introducing a new reduction strategy, building a new system configuration and upgrading the corresponding software. We then validated the entire system, using a fracture model of bovine femur. Results Precision tests were performed seven times on a bovine femur with a transverse fracture. Residual deviation was 1.23 ± 0.60 mm in axial deflection, 1.04 ± 0.47 mm in translation, 2.34 ± 1.79° in angulation and 2.83 ± 0.96° in rotation. Conclusion Our new reduction system described here is detachable, flexible and more precise in coordinate transformations. The detachable, modular design will allow for more analogous applications in the future. Copyright © 2014 John Wiley & Sons, Ltd. Keywords CAOS
hexapod; computer assisted orthopaedic surgery; diaphyseal fracture;
Introduction In orthopaedics, the Arbeitsgemeinschaft für Osteosynthesefragen (AO) technique maintains the original skeletal geometry and provides anatomical reduction and solid fixation for fracture treatment. However, healing problems in the soft tissue and blood supply can impair drainage and cause hematoma. Biological osteosynthesis (BO) aims to preserve the biology and create solid fracture unions, but requires radiative over-exposure (1) and an experienced surgeon and still shows poor accuracy and security, with rotational residuals always remaining (2,3). Owing to advances in computer
Advancing CAOS with a hexapod device
and engineering technology, computer-assisted orthopaedic surgery (CAOS) is attracting increasing clinical attention, is widely used in the orthopaedic field and shows a promising future. CAOS is performed by digitalizing the patient’s anatomy, visualizing the fractures and local physical characteristics, digitally combining the images and integrating the surgical instruments into the digitized image background. This allows surgeons to perform any procedure in an improved, practical and real-time visual environment in a minimally invasive way, with increased accuracy and decreased radiation exposure. CAOS was first described as a hypothesis by BouazzaMaroufin 1995 (4). In the late 1990s, CAOS was first introduced into practical application (5) and was then rapidly developed. Koo et al. (6) and Seide et al. (7) developed the algorithm and prototype, respectively, to perform closed fracture reduction. The hexapod robot kinematics that Seide et al. used was first described in engineering terms as the Stewart platform by Stewart in 1965 (8), and then in medicine for limb lengthening and deformity correction by Taylor in 1994 (9) as the Taylor spatial frame. The hexapod device has several advantages, such as ease of application, computer accuracy for the correction of six degrees of freedom (6-DOF) deformities, and the ability to perform residual serial postoperative adjustments and manipulate fractures into excellent alignment in an
349
outpatient, less invasive, manner through struts adjustments only (10–12). Using the hexapod kinematics and computer technology, we designed a novel 3D hexapod CAOS system for diaphyseal fracture reduction of long bones that can provide precise parameters and convenient procedures for fracture reduction and deformity correction compared to conventional techniques (13). A novel reduction strategy, a registration and reduction technique using mirror imaging, for one-legged fracture (14,15) was also created. To address several problems we encountered in the last version (13), we advanced the system and performed serial specific tests to determine its precision, practicality and feasibility. We had endeavoured to make precise configuration changes during reduction procedures, so we used electric telescopic struts with DC motors as the actuators, whose length did not vary freely. However, when released after the procedures, the neutral position could not be regained easily. Also, the DC motors mounted on the actuators occupied too much space, limiting the surgeons’ operation and risking jamming between adjacent motors; this would be a major inconvenience, a waste of time and a burden on the surgical crew. The other problem was that the system operated with the patient’s legs placed through two rings, which could make the patients uncomfortable, jam during the procedure or occupy space during CT scans (Figure 1).
Figure 1. (A, B) Twelve specific location balls were fixed on each ring for registration. (C, D) The cadaver precision tests showed acceptable results Copyright © 2014 John Wiley & Sons, Ltd.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
350
Materials and methods Reduction strategy We developed a novel reduction strategy, a registration and reduction technique using Townsend and Richards (14) mirror imaging for one-legged fractures on the hexapod system. The physical kinematics features of the human skeletal system create human body mirror symmetry. Two mirror-symmetrical identical bones share the same physical parameters (15). The non-fractured mirror image leg should ideally match the prefractured injured leg. Thus, the mirror image can provide precise parameters for anatomical fracture reduction. Furthermore, we introduced 3D image processing techniques and programmed the corresponding software (Figure 2). For registration, mirror transformations of the contralateral long-bone models were used as reference standards for reduction. Two major fragments of long-bone fractures were made to coincide with those of the contralateral bones, and the reduction trajectory was planned according to the relative displacement transformation matrix after coincidence. The reduction procedure was then performed by the reduction unit. During image processing, we introduced a threedimensional (3D) surface stereolithography (STL) model of the fractures as our reconstruction method, based on computed tomography (CT) data from Mimics software v. 10.01 (Materialise NV, Leuven, Belgium), using the surface rendering algorithm of marching cubes (16,17) to accurately measure and analyse the geometric parameters of the bones. The 3D image was reconstructed from DICOM slices of high-resolution CT scans made of point cloud data, whose surface consists of characteristics of the femur anatomy. We considered the 3D surface STL models of the contralateral long bone after mirror
transformation to be essentially identical to the fractured long bones after reduction. Two shape-specific sections were selected for registration with the contralateral 3D image. The mapping matrix ensured that the two point clouds were the minimum distance apart, enabling the reduction robot to move the fractured point cloud to the normal femur point cloud. ICP algorithms are commonly used to compute the transform matrix (18), and we adopted the following data processes, based on an ICP algorithm, to calculate the mapping matrix for the two point clouds. The mathematical model for registration, based on the ICP algorithm, was as follows: f ¼ min
Xn i¼1
kpi ðR qi þ SÞk2 ; i ¼ 1; 2; ⋯; n
(1)
where pi stands for the points in the specific region of the two fracture fragments, qi stands for the points in the specific region of the two contralateral fragments, and R and S stand for the rotation and translation matrix after the registration of fractures. The parameters for automatic registration were obtained by finding optimal solution of this equation. If deviations arose because of the automatic registration, manual trimming was needed. For selection, priority should be given to the region with the least impurity on the 3D surface of the STL model of pi and qi.
Construction of a novel hexapod reduction configuration We designed the original hexapod configuration based on a Taylor spatial frame, which can perform any six-degree of freedom (6-DOF) movements without loss of stability or the need to change parts during any reduction
Figure 2. Operator interface of custom software program; a displaying and operating window is on the left and four categories of menu are integrated and listed on the right Copyright © 2014 John Wiley & Sons, Ltd.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
351
Advancing CAOS with a hexapod device
procedure for long-bone fractures. The two major fracture fragments were tightly attached to the proximal and distal rings and then the fracture was reduced, along with configuration changes of the hexapod device. Twelve balls at specific locations were fixed on each side of the rings for registration (Figure 1A, B). To address the problems described above, we made several improvements through a redesign after deep discussions. Figure 3A shows the new fracture reduction device in its neutral configuration, which maintains all of the original hexapod advantages.
Hybrid-driven system A hybrid-driven system, based on a motor drive and a hydraulic transmission, was used in place of the electric driving mode, and consisted of a motor, a screw and two groups of hydraulic cylinders. Six electric telescopic actuators used previously in the reduction mechanism were replaced by a group of six hydraulic cylinders as an internal hydraulic cylinder set (Figure 3A, yellow) that provided more non-working freedom and higher stability and precision during all procedures. An identical external
Figure 3. (A, B) Key hexapod construction and ordinate transformation. (C–F) Six degrees of freedom movements: (C) translation movements along the X, Y axes; (D) translation movement along the Z axis; (E) rotation movement around the X, Y axes; (F) rotation movements around the Z axis. (G, H) Upper and lower components in the series and parallel configurations: (G) parts in blues and green demonstrate lower or static components; (H) parts in red and orange demonstrate upper or dynamic components Copyright © 2014 John Wiley & Sons, Ltd.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
352
hydraulic cylinder set in the control box was connected with the internal set by hydraulic pipes and sealed. The adjustment of the hydraulic cylinder on the mechanism was controlled by driving the external hydraulic cylinder via a stepper motor drive and hydraulic transmission. Non-compressed liquid was infused into the cylinder and pipe. The rotary motion of the servo-motors was converted into reciprocating motion of the hydraulic cylinder, which transferred the power to the internal hydraulic cylinder set. The biggest advantages of this electro-hydraulic hybriddriven system are that, when released, the length of the hydraulic cylinders can be adjusted freely and randomly and, when validated, the length can be solidly fixed and precisely adjusted. This mechanical part makes the system more reliable and precise, with higher freedom of movement, while simultaneously requiring less space and facilitating the entire reduction mechanism and surgical procedure. In order to improve transfer efficiency for long-standing depreciation, the single-thread triangular screw in the last version was replaced by ball screws (lead, P = 5 mm; transmission efficiency, η = 90%). Neglecting the influence of transmission efficiency of the reduction gearbox, the output maximum force and velocity of each hydraulic cylinder can be found using the following formula: F¼
2 πMiη n ; V¼ P P 60i
(2)
Series and parallel configurations In addition to the hydraulic drive units, a new series and parallel configuration was introduced. In the previous version, the mechanism configuration was designed directly from the hexapod robot kinematics and was characterized by a parallel robot manipulator or Stewart platform, which worked with the patients’ broken legs in the rings. This configuration could jam during the reduction procedure and make the patients uncomfortable with parts of the mechanism between their legs. Thus, we designed a new series and parallel configuration, which extends a specific positioning unit outside the hexapod device. The reduction mechanism includes two major units, the positioning and reduction units (Figure 4B, C). The positioning unit uses a registration strategy and provides temporary fixation for fractures. Two flatbeds, called fracture holders, are arranged in series, connecting the two major long-bone fracture fragments for temporary fixation like external fixators (Figure 3B, green and orange; Figure 4C). Two side bars fix the holders together Copyright © 2014 John Wiley & Sons, Ltd.
as the positioning unit, which is relatively independent and can be disassembled and separated from the reduction system. The reduction unit consisted of the core hexapod mechanism below the positioning unit and fracture (Figure 4C, D). The holders and two rings of the hexapod mechanism could be assembled as a solid separate unit called the upper and lower components. In the lower components, also called the static components, the fracture holder stays parallel to the lower ring (Figure 5A, green and blue). Similarly, in the upper component, also called the kinetic/dynamic component, the holder is parallel to the upper ring (Figure 5B, orange and red). This construction is the key configuration change: the parallel configuration of the hexapod mechanism (Figure 3A) evolved to a series connection (Figure 3B) and the previous 6-DOF hexapod kinematics movement was changed to a new form. Fracture movements could be split into six simple movements in the 6-DOF 3D coordinate system (Figure 3C–F). The Z axis was defined as the longitudinal axis of the bone. The X and Y axes were perpendicular to the Z axis and to each other. The 6-DOFmovements were defined as translations along and rotations around the three axes, specified by angulation (rotation around the X and Y axes), translation, length discrepancy and rotation (rotation around the Z axis). In the previous version of this system, the two main fragments of each long-bone fracture were validly connected to the two rings and the 6-DOF relative movements of the rings were directly converted to 6-DOF movements of the two main fragments (Figure 1). In this new system version, static components contain the core hexapod mechanism for execution of the 6-DOF movements, which translated to relevant movements of the two holders in series. Two 6-DOF movements of the hexapod device and the holders with major fracture fragments were vertically adapted to each other. For example, Z axis length variations of the hexapod mechanism would convert to translation variance in X axis holders. In the previous version, the two major fracture fragments were fixed onto the mechanism before a highresolution computed tomography (CT) scan. The whole system was large and heavy to move, which caused patients too much discomfort and might not fit into the CT machine or could jam during procedures. In addition, the reduction mechanism and fractures had to remain attached together during the entire procedure. However, in this new version, the fracture tray was made relatively independent. The positioning unit can be disassembled from the reduction unit. In the clinical setting, after hospitalization, fractures are attached to the holders of the positioning unit immediately and secured through temporary fixation and Schanz nails. Afterwards, high-resolution CT scans are taken of both the broken and unbroken legs. This new Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
353
Advancing CAOS with a hexapod device
Figure 4. (A–D) Details of the new construction: α, fracture holder; β, side bar
positioning unit construction is much simpler, can save a lot of space for reduction procedures and gives patients more comfort. After the CT scans, the holders are rigidly attached to the reduction unit and components. Next, the hydraulic drive units are locked. After disassembling the two side bars, the lower and upper components become two relatively dependent parts that are connected by six parallel hydraulic drive units. Thereafter, the imaging process and pre-operative planning is conducted, followed by the reduction procedure.
Neutral configuration The core hexapod mechanism of the reduction unit has a neutral configuration, with all six internal cylinders 427 mm in length. In the neutral configuration, the two fracture holders of the positioning unit can be assembled into corresponding components for the subsequent reduction. The neutral configuration is the exact position where the reduction procedure starts. After the procedures, the Copyright © 2014 John Wiley & Sons, Ltd.
reduction unit would be in a specific pose and the positioning unit would be replaced by traditional hybrid external fixators and separated from the reduction unit. At that point, fractures are permanently fixed. Then the procedure is over and the reduction unit automatically resets to its neutral configuration via the control panel, ready for another procedure.
Registration strategy We have also introduced a new registration strategy on the positioning unit. The key point concerning the reduction procedure in this system lies in the position and orientation changes to the hexapod device, realized by adjusting the length of six hydraulic cylinders, determined by the 12 cylinder endpoints. However, it was difficult to locate the endpoints because of the complicated structure of the ball joints in the 3D image. Previously, we created a location ball structure to locate the endpoints for conformation and determining the mechanism Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
354
Figure 5. Steps involved in the fracture reduction procedure in the new construction
transformation. Because there is a total of 12 location balls to process in the 3D image (Figure 1), this task would take considerable work and time and potentially introduce variability. To address this issue, we designed a positioning unit system to facilitate and simplify the process by introducing a new algorithm for the registration and reduction procedures. After the CT scan, the positioning unit assembled the reduction unit into its neutral configuration. The two fracture holders are tightly attached with the static and dynamic rings together, to build the upper and lower components. These two constructions are then unchangeable during the entire registration and reduction procedures. This arrangement, with the positioning unit outside of the reduction unit, is more convenient and precise. Because four single points in space determine a spatial position and attitude, four specific marking points must be chosen on the holder corners to act as feature points that are easy to locate, to determine the spatial configuration of the fractures and practical hexapod construction illustrated in Figure 4B (green square). Strikingly, the hexapod device is no longer necessary for the preoperative CT scan. The CT scans of patients can be taken with the fractures attached to the positioning unit only, and the 3D image involved contains only the broken and unbroken limbs and the positioning unit. This set-up facilitates and simplifies the image-analysis process and Copyright © 2014 John Wiley & Sons, Ltd.
reduction procedure. Unlike the previous system version, here we only need to locate four non-specific marking points on each holder. After attaching the holders to the rings, the surgeon can locate the position and altitude of the hexapod device through the image-analysis process. This registration strategy may be more comfortable for patients and simplify the registration and reduction procedures. A custom C++ and Visualization Toolkit (VTK) software algorithm was programmed to account for the new registration strategy and the changes made to the reduction configuration.
Safety strategy Working space analysis The working space of the construction for the conversion of the coordinates and 6-DOF movement mode was analysed. The working space determines the extent of fracture deformity to be corrected. According to hexapod kinematics, working space is mainly affected by three main factors, the diameter of the rings, the nominal length and the variation length of the actuators. For the new configuration, we made several major changes. Both rings in the previous version had the same radius. However, to save space for the holders in the new version, the ring in the upper component was made slightly Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
Advancing CAOS with a hexapod device
smaller. With two attached components, the working space of the hexapod device determines that of the holders. Their working space is analysed below. The core structure of the hexapod device has two rings, radius 110 and 195 mm, for the upper and lower components, respectively, and six actuators, 331 mm in basic length with a 200 mm length variation. In this configuration especially, it has a working space with –220 to +220 mm of translation and –43° to +43° of rotation in the coronal axis, –20 to +250 mm of translation and –35° to +35° of rotation in the sagittal axis, and –95 to +95 mm of translation and –130° to +130° of rotation in the vertical axis (Figure 3). Compared with the previous version, the working space of the new version is significantly increased. The fracture holder motions come from the hexapod device movements. Two 6-DOF movements of the hexapod device and the holders with major fracture fragments were vertically adapted to each other. The mechanism itself and the motion transfer from the reduction unit to the positioning unit increases the complexity of the practical working space of the fracture holders. For example, Tz translations in the fracture holders derive from Tx translations of the hexapod device and Ty rotation in the fracture holders derives from Tz rotations in the hexapod device, e.g. a translation of 1 mm (Tz) in the fracture holders would cause a 1 mm (Tx) translation in the hexapod device. Thus, the virtual working space of the fractures holders is –220 to +220 mm of translation and –43° to +43°of rotation in the coronal axis, –20 to +250 mm of translation and –35° to +35°of rotation in the vertical axis, and –95 to +95 mm of translation and –130° to +130° of rotation in the sagittal axis, which is determined by that of the hexapod device (Figure 3). The virtual working space in the new design is also significantly improved and can cover most deformations. Even for degrees of deformity fracture that exceed the working space, appropriate pre-operative traction can be added to reduce the degree of deformity. Pre-operative trajectory planning Cortical abutment and bone spike overlap are common in fractures. The direct reduction trajectory has a high risk of spike collision. For a smooth and automatic reduction procedure, we introduced a pre-operative trajectory-planning algorithm to avoid iatrogenic trauma during the entire procedure, based on the shortest linear distance and safety strategy. Three fundamental principles were used: (a) necessary distraction; (b) minimization of the required forces induced by moving the fragment (soft-tissue balance); and (c) avoidance of unnecessary movements. All planned procedures conformed to these principles. For standardization, we created a novel long-bone fracture classification system based on fracture pattern. Copyright © 2014 John Wiley & Sons, Ltd.
355
Distraction in advance is intended to regain a broken limb’s normal length and provide extra room for deformity correction. Beyond regaining the limb length, overdistraction of the fracture fragments by 5–10 mm is recommended (19). In addition, Seide et al. (20) recommended that fractures should first be distracted to a slight over-distraction by ca. 2 mm of least reduction room. With the least reduction room 2 mm in the distraction externum 10 mm, the 8 mm of overlap length after the procedure could makes a completely different reduction strategy. There is up to 8 mm of room for the overlapping parts of the main fracture fragments spike. Thus, the fracture classifications we created were mainly based on the overlap length after reduction. We divided the femoral shaft fractures into three main types: A, no overlap length; B, overlap length < 8 mm; and C, overlap ≥ 8 mm. For type A and B fractures, the procedures are relatively easy to perform. The basic treatment steps include: 1, careful evaluation, classification and measurement of the overlap length; 2, adequate over-distraction (no more than 10 + 2 mm for the least reduction room); 3, alignment correction; and 4, contraction and compression. However, in type C fractures with overlap length ≥ 8 mm, simultaneous alignment is not possible. The initial overdistraction must be limited within 10 mm, so there may not be enough room for simultaneous alignment; bone spikes could collide. Fractures with face-to-face surfaces are classified as subtype C1 and back-to-back as subtype C2. For fractures of type C1, the over-distraction, rotation and angulation correction processes listed above would manoeuvre the two fracture surfaces into a face-to-face configuration, and the inclined surface plane might provide enough room for subsequent translation correction; but for C2 fractures, the back-to-back bone spikes need a revolution to bypass each other before a transformation from the back-to-back configuration into a face-to-face one. Contraction and compression would then finish the entire procedure. Closed-loop control system We used a personal computer (Intel® Core TM 2, 3.00 GHz) as a controller and ran a custom software program in VC++ to prepare the 3D reconstruction, to register the CT image data, to create the signal procedure for the sensor and to plan the trajectory and function of the motion control. A six-axis motion control card (ADT-8960) was used for communication, based on the PCI bus, to control the stepper motors motion by providing pulse signals. In order to achieve synchronous motions of the external and internal hydraulic cylinders, we introduced a doubleposition, closed-loop control system. The encoder signals were fed back to the motor driver to achieve an inner position closed loop to decrease the errors caused by stepper-motor motion (Figure 4). The linear displacement Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
356
transducer that was fixed on the internal hydraulic cylinders in the hexapod mechanism fed back displacement signals to the PC via the data acquisition card, to realize the external position closed-loop control for decreasing errors caused by the hybrid-driven system based on the hydraulics and motor. Indeed, the double-position closed-loop control guaranteed the position accuracy of the robot. Taking the 6-DOF movements of the hexapod device into account, we used a binocular visual tracker (Micron Tracker MTC3.0, Claron Technology Inc., Toronto, Canada) to test the precision of the hexapod mechanism. Within the range of the working space, translation errors were < 2 mm and angulation errors were < 1.2°. Fractures that heal with an angular deformity > 5° in any anatomical plane, or a translational deformity > 10 mm are regarded as malunions (21). These test results showed that the hydraulic drive parts and reduction construction could provide excellent precision, while occupying less space compared with electric telescopic struts. Other improvements We also added several security measures. Emergency brakes were installed in both the software and the control box. Mechanical sensors were fixed on the positioning unit for detecting soft-tissue interference.
Results To verify the feasibility and accuracy of reduction, we used a pair of bovine femurs as a fracture model, and one of the bovine femurs was broken to simulate a transverse fracture (Figure 5). Before modelling, CT scans of both intact femurs were taken for data back-up. These CT data were used to reconstruct the original 3D image for residual deformity measurement and analysis. Figure 5 shows the routine steps of the reduction procedure for a new fracture reduction: 1, attach the fractures rigidly to the holders of the positioning unit for temporary fixation; 2, take a pre-operative high-resolution CT scan for both limbs; 3, load the DICOM CT data into the software program for 3D image reconstruction, registration, image-analysis processing and pre-operative planning; 4, perform the reduction procedure; 5, after the reduction, replace the fracture holders of the positioning unit with traditional hybrid external fixators; and 6, take another CT scan for postoperative precision testing. Using equation (2), the maximum force and velocity of hybrid-driven system were calculated as F = 1243 N and V = 10 mm/s. Based on these results, the output force and velocity met the demands of the reduction system, as the system needed a lower velocity for safety. Within the range of the working space, translation errors of the Copyright © 2014 John Wiley & Sons, Ltd.
hexapod device measured from MicronTracker were 2 mm and angulation errors were < 1.2°. The sites where we choose to insert Schanz nails for fixation were away from the peripheral nerves, arteriovenous system and tendons, in case of iatrogenic injury and limitation of activity, such as in the lateral femur or medial tibia. Each major fragment requires at least two Schanz nails for stabilization to prevent rotation. Reduction procedure experiments were carried out and the precision of this reduction prototype was tested seven times on this in vitro bovine femoral fracture model. All the procedures followed the routine steps shown in Figure 5. After a procedure, the residual deformity was measured by the same operator under direct visualization of the fracture sites. The four parameters of axial deflection, translation, angulation and rotation were collected and used to evaluate precision (Tables 1 and 2). Based on a definition that reduction with an angular deformity > 5° in any anatomical plane, or a translational deformity > 10 mm, are regarded as malunions (21), the reductions achieved excellent union. In addition, the translation and rotation corrections were better than the precision test results from the previous prototype version (p < 0.05) (13). The hexapod device was able to return to its neutral position after each reduction procedure. During the seven tests, several path points were each randomly and repeatedly chosen to determine precision intra-operatively. Two rough parameters were collected, displacement and angulation deviation, and compared with the corresponding points in the trajectory planning software window. All displacement deviations were < 1.8 (average < 1) mm and angulation deviations were < 1.5°. No scratches or jamming of the prototype occurred.
Discussion The operative realignment of mechanical axis deviations (MAD) is necessary to prevent early joint degeneration and to reach normal load-bearing capacity. However, Table 1. Practical residual deformity value collected from postoperative measurements of seven precision tests Test No. 1 2 3 4 5 6 7
Axial deflection (mm)
Translation (mm)
1.1 2.0 0.5 0.9 0.8 2.1 1.2
0.6 1.9 1.3 0.6 0.7 1.2 1.0
Angulation Rotation (°) (°) 4.2 2.6 3.0 2.9 0.6 2.1 2.3
2.5 1.2 3.2 1.7 2.1 4.1 2.4
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
357
Advancing CAOS with a hexapod device Table 2. Results from bovine femoral repositions without soft tissues (n = 7 times) Parameters Axial deflection (mm) Translation (mm) Angulation (°) Rotation (°)
Average
SD
Minimum
Maximum
1.23 1.04 2.53 2.46
0.60 0.47 1.09 0.96
0.5 0.6 0.6 1.2
2.1 1.9 4.2 4.1
achieving perfect reduction accuracy and allowing early functional exercise using the AO technique, or using soft tissue preservation to achieve good healing outcomes for fracture treatment, presents a difficult dilemma. How to best combine the respective merits of the different approaches is an important scientific concern. The AO technique reduces fractures through open reduction and internal fixation (ORIF) to reconstruct the MAD, which can give rise to soft tissue and periosteum damage and haematoma drainage around the fracture site. The priority of the BO technique is avoiding such damage by preserving the native biological environment. Several complications, such as radiative overexposure, poor accuracy and high surgeons’ experience requirements, restrict the practical application of these techniques. For example, using a fluoroscopic technique, good reduction can be achieved, at the cost of excessive radiative exposure to surgeons, through a minimally invasive procedure that requires an experienced surgeon. Biological preservation does not necessarily mean a small incision and minimizing MAD is difficult, as is performing the minimally invasive percutaneous plate osteosynthesis technique (22). In recent years, CAOS was developed to improve the accuracy of surgical procedures (23,24). CAOS-based minimally invasive techniques could provide better healing outcomes by maintaining the soft tissue attachments and blood supply, and minimizing MAD through computational techniques. CAOS technology has developed several algorithms and prototypes. However, no current prototype has demonstrated all the potential benefits and limitations. The da Vinci surgical robot is the best example of this kind of computer-assisted surgical tool (25). The role of such technology is limited at present, as surgeons must be proficient in traditional surgical approaches before mastering and using this technology and equipment. To achieve better precision and lower experience requirements and cost, we developed a reduction system based on a mirror registration technique and hexapod kinematics. The physical kinematic features of the human skeletal system create human body mirror symmetry. Two mirror-symmetrical identical bones should share the same physical parameters (15). Through computer-imaging processes, we obtained precise anatomical reduction parameters to assist fracture treatment. Copyright © 2014 John Wiley & Sons, Ltd.
To execute the reduction procedure, a custom-made reduction construction prototype was designed, manufactured and improved (13). In the present study we proposed and tested two major improvements in the construction of a new reduction mechanism, a hybrid-driven system and a series and parallel configuration. The present reduction system adopted a hybrid-driven system, based on a motor drive and hydraulic transmission instead of the direct driving mode that was used in the previous prototype version. This system provides more precision and occupies less space. In this new prototype, the electro-hydraulic hybrid-driven actuators are larger and provide more power, but occupy a relatively small space. The precision tests of the hybriddriven system also reported good results. Another improvement to the prototype construction is the series and parallel configuration. The serial mechanism has a large workspace and a flexible mechanism but poor load capacity, and occupies a large operative workspace. Compared with this serial mechanism, the parallel mechanism has higher position accuracy and greater load capacity and rigidity, but a smaller workspace. The features of each mechanism are complementary to one another. Therefore, we combined them into a series and parallel configuration (Figure 4): the parallel part is the hexapod device, as a reduction unit, and the series part is the positioning unit, with the fracture holders for temporary fixation and registration. The two fracture holders can be disassembled from the upper and lower components, and the fractures are attached to the holders using temporary fixation. As an improvement, we now use registration points on the holders themselves, so that the hexapod device does not need to be attached during the CT scan. Compared with the previous version, patients will be much more comfortable during the CT scan, with only the holders in place. This change also simplifies analysis of the CT data compared with the previous version, because the hexapod device is not included and the 3D images are smaller and contain fewer artifacts that could influence registration and pre-operative planning. In the current version, we present a new registration strategy. In the 3D reconstruction image, instead of the previous method of identifying balls, a series of feature points were taken as markers that were easily identified on the fracture holders to perform the conversion of the coordinate system. This change improves the convenience, flexibility, ease of operation and portability associated with the system. After executing the reduction plan, residual deformity could remain in some cases during clinical practice. The post-reduction CT scans are meant to test the procedure precision and determine any remedial reduction procedures. We discussed additional safety strategies to guarantee a smooth and automatic reduction procedure. The working space defines the degree of deformity of a fracture. In this Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
358
new construction, we used a larger configuration with more working space, which could deal with more complicated fractures. If deformity exceeds the working space, a pre-operative rough alignment might be necessary to decrease the degree of deformity before the reduction procedure. Of all the safety strategies discussed, we considered trajectory planning to be the top priority pre-operatively. During any reduction procedure, peaks of the bone fragment could collide when using a direct trajectory. Therefore, a precise pre-operative trajectory-planning analysis is needed to avoid iatrogenic trauma. The intra-operative reduction precisions we tested were varied, but the displacement deviations were < 1.8 (average < 1) mm and angulation deviations were < 1.5°. Graham (26) has reported that a simple midshaft fracture had a maximum resulting force of 428 N. Owing to the strong power of muscles, mechanical sensors would need, to a relatively high threshold, to avoid soft-tissue iatrogenic injury. In this case, iatrogenic fracture from a spike collision could be avoided through careful pre-operative trajectory planning instead of mechanical sensors. Fracture pattern classification and reduction trajectory preplanning and analysis should be performed independent of which reduction technique will be used. For clinical practice, we suggest a minimum 1 mm gap between the edges of fracture spikes in pre-operative trajectory planning, in case of intra-operative collision below the threshold. We maintained our closed-loop control system for security and to improve precision. The encoder signals were fed back to the motor driver to achieve an inner position closed loop, to decrease the errors caused by the hybrid-driven system based on the hydraulic and motor. If a jam happens during the reduction procedure as a result of soft-tissue involvement, the hexapod device cannot reach its final position and the sensors on actuators would feed back to the motors to continue adjustment until the final position is reached. Among the improvements we made, the simplification of the configuration, the registration strategy and the closed-loop control system contribute to the improvement in precision. A positioning unit was designed to fix the fracture that simplified the system. The CT scans are conducted with patients wearing the positioning unit only, so fewer CT data were collected and the data were easy to process. In addition, four specific marking points on the fracture holder corners were taken as feature points in place of the location balls structure, which also simplified the course of registration. The closed-loop control system could also compensate for residual deformity to improve precision. Two side bars and two fracture holders are involved in the positioning unit and after each procedure the reduction unit would return to a specific pose, and the Copyright © 2014 John Wiley & Sons, Ltd.
positioning unit was then replaced by traditional hybrid external fixators and disassembled from the reduction unit. The configuration of the positioning unit with four parts and the disassembly are complicated and could lead to precision loss. As a future improvement, simplified unilateral serial external fixators with a universal joint and a telescopic sleeve could be introduced. Such a new fixator would include special connectors with the reduction unit and registration marks. Patients would have their fractures attached to the new fixators, with the universal joint and telescopic sleeve locked before the CT scan, and be connected with the reduction unit. During the reduction procedure, the new fixators would be unlocked and disassembled from the reduction unit. After the procedure, they would be locked and used as traditional external fixators. This configuration would further simplify the system. We will also try to expand the capacity of this hexapod reduction system to treat various long-bone fractures. In conclusion, the hexapod reduction system developed in this study allows surgeons the ability to perform fracture reduction in an objective, efficient and accurate manner, while minimizing the radiation exposure to both the patients and the surgical team. Using the mirror image of the unbroken limb as a template to guide the reduction procedure, surgeons can concentrate on preoperative planning and supervise the reduction procedures. The hexapod reduction system may be able to reduce long-bone diaphyseal fractures more precisely and anatomically, and also minimize the extent of periosteum and soft-tissue damage around the fracture site.
Acknowledgements This project was funded by the National High Technology Research and Development Programme 863 (Grant No. 2012AA041604), the National Key Technology R&D Programme (Grant No. 2012BAI14B02), the Military Medical Research Programme (Grant No. BWS11J113) and the National Natural Science Fund (Grant No. 31271000).
Conflict of interest The authors have stated explicitly that there are no conflicts of interest in connection with this article.
Funding No specific funding. Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
Advancing CAOS with a hexapod device
References 1. Leloup T, El Kazzi W, Schuind F, et al. Conception of a navigation system controlling diaphyseal fracture reduction treated with external fixation. Int J Med Rob 2009; 5: 99–109. 2. Stephen DJ, Kreder HJ, Schemitsch EH, et al. Femoral intramedullary nailing: comparison of fracture-table and manual traction. a prospective, randomized study. J Bone Joint Surg Am 2002; 84A: 1514–1521. 3. Weil YA, Greenberg A, Khoury A, et al. Computerized navigation for length and rotation control in femoral fractures: a preliminary clinical study. J Orthop Trauma 2014(Feb); 28(2): e27–e33. 4. Bouazza-Marouf K, Browbank I, Hewit JR. Robotic-assisted internal fixation of femoral fractures. Proc Inst Mech Eng H 1995; 209: 51–58. 5. Merloz P, Tonetti J, Pittet L, et al. Pedicle screw placement using image guided techniques. Clin Orthop Relat Res 1998 (Sep); 354: 39–48. 6. Koo TK, Chao EY, Mak AF. Development and validation of a new approach for computer-aided long bone fracture reduction using unilateral external fixator. J Biomech 2006; 39: 2104–2112. 7. Seide K, Faschingbauer M, Wenzl ME, et al. A hexapod robot external fixator for computer assisted fracture reduction and deformity correction. Int J Med Rob 2004; 1: 64–69. 8. Stewart D. A platform with six degrees of freedom. Proc Inst Mech Eng 1965; 180: 371–378. 9. Taylor JC. Perioperative planning for two- and three-plane deformities. Foot Ankle Clin 2008; 13: 69–121, vi. 10. Al-Sayyad MJ. Taylor spatial frame in the treatment of pediatric and adolescent tibial shaft fractures. J Pediatr Orthop 2006; 26: 164–170. 11. Sala F, Elbatrawy Y, Thabet AM, et al. Taylor spatial frame fixation in patients with multiple traumatic injuries: study of 57 long-bone fractures. J Orthop Trauma 2013; 27: 442–450. 12. Dammerer D, Kirschbichler K, Donnan L, et al. Clinical value of the Taylor spatial frame: a comparison with the Ilizarov and Orthofix fixators. J Child Orthop 2011; 5: 343–349.
Copyright © 2014 John Wiley & Sons, Ltd.
359 13. Tang P, Hu L, Du H, et al. Novel 3D hexapod computer-assisted orthopaedic surgery system for closed diaphyseal fracture reduction. Int J Med Rob 2012; 8: 17–24. 14. Townsend G, Richards L. Twins and twinning, dentists and dentistry. Aust Dent J 1990; 35: 317–327. 15. Hu L, Zhang J, Li C, et al. A femur fracture reduction method based on anatomy of the contralateral side. Comput Biol Med 2013; 43: 840–846. 16. Weng YYN, Pierson R. Three-dimensional surface reconstruction using optical flow for medical imaging. IEEE Trans Med Imaging 1997; 16: 630–641. 17. Lorensen WE, William EL, Harvey EC. Marching cubes. A high resolution 3D surface construction algorithm. Comput Graphics 1987; 21: 163–169. 18. Paul JB, Neil DM. A method for registration of 3D shapes. IEEE Trans Pattern Anal Mach Intell 1992; 14: 239–256. 19. Starr BR, Bucholz RW, Heckman JD. Fractures of the shaft of the femur. In Rockwood and Green’s Fractures in Adults, Bucholz RW, Heckman JD (eds). Lippincott, Williams & Wilkins: Philadelphia, PA, 2002; 1683–1730. 20. Seide K, Wolter D, Kortmann HR. Fracture reduction and deformity correction with the hexapod Ilizarov fixator. Clin Orthop Relat Res 1999(Jun); 363: 186–195. 21. Koo TK, Mak AF. A knowledge-based computer-aided system for closed diaphyseal fracture reduction. Clin Biomech 2007; 22: 884–893. 22. Krettek C, Muller M, Miclau T. Evolution of minimally invasive plate osteosynthesis (MIPO) in the femur. Injury 2001; 32(suppl 3): SC14–SC23. 23. Renkawitz T, Tingart M, Grifka J, et al. Computer-assisted total hip arthroplasty: coding the next generation of navigation systems for orthopedic surgery. Expert Rev Med Devices 2009; 6: 507–514. 24. Jackson DW, Simon TM. History of computer-assisted orthopedic surgery (CAOS) in sports medicine. Sports Med Arthrosc 2008; 16: 62–66. 25. Sterrett SP, Jarrard DF. Robotic assisted laparoscopic radical prostatectomy: evolution and outcomes. Minerva Urol Nefrol 2008; 60: 31–39. 26. Graham AE, Xie SQ, Aw KC, et al. Bone–muscle interaction of the fractured femur. J Orthop Res 2008; 26: 1159–1165.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
ARTICLE
Advancing computer-assisted orthopaedic surgery using a hexapod device for closed diaphyseal fracture reduction
Hailong Du1 Lei Hu2 Changsheng Li2 Tianmiao Wang2 Lu Zhao2 Yang Li2 Zhi Mao1 Daohong Liu1 Lining Zhang1 Chunqing He1 Licheng Zhang1 Hongping Hou1 Lihai Zhang1*† Peifu Tang1*† 1
Department of Orthopaedics, Chinese PLA General Hospital, Beijing, People’s Republic of China
2
Robotics Institute, Beihang University, Beijing, People’s Republic of China *Correspondence to: P. Tang, L. Zhang, Department of Orthopaedics, Chinese PLA General Hospital, 28 Fuxing Road, Haidian District, Beijing 100853, People’s Republic of China. E-mail: [email protected]; [email protected]
†
These authors contributed equally to this study.
Accepted: 13 August 2014
Copyright © 2014 John Wiley & Sons, Ltd.
Abstract Background Surgical complications such as healing problems, in fractures treated using the Arbeitsgemeinschaft für Osteosynthesefragen (AO) technique, present functional and economic challenges to patients and treatment dilemmas for surgeons. Computer-assisted orthopaedic surgery using minimally invasive techniques focused on biological osteosynthesis is a novel direction for fracture treatment. Method We modified the hexapod computer-assisted fracture reduction system by introducing a new reduction strategy, building a new system configuration and upgrading the corresponding software. We then validated the entire system, using a fracture model of bovine femur. Results Precision tests were performed seven times on a bovine femur with a transverse fracture. Residual deviation was 1.23 ± 0.60 mm in axial deflection, 1.04 ± 0.47 mm in translation, 2.34 ± 1.79° in angulation and 2.83 ± 0.96° in rotation. Conclusion Our new reduction system described here is detachable, flexible and more precise in coordinate transformations. The detachable, modular design will allow for more analogous applications in the future. Copyright © 2014 John Wiley & Sons, Ltd. Keywords CAOS
hexapod; computer assisted orthopaedic surgery; diaphyseal fracture;
Introduction In orthopaedics, the Arbeitsgemeinschaft für Osteosynthesefragen (AO) technique maintains the original skeletal geometry and provides anatomical reduction and solid fixation for fracture treatment. However, healing problems in the soft tissue and blood supply can impair drainage and cause hematoma. Biological osteosynthesis (BO) aims to preserve the biology and create solid fracture unions, but requires radiative over-exposure (1) and an experienced surgeon and still shows poor accuracy and security, with rotational residuals always remaining (2,3). Owing to advances in computer
Advancing CAOS with a hexapod device
and engineering technology, computer-assisted orthopaedic surgery (CAOS) is attracting increasing clinical attention, is widely used in the orthopaedic field and shows a promising future. CAOS is performed by digitalizing the patient’s anatomy, visualizing the fractures and local physical characteristics, digitally combining the images and integrating the surgical instruments into the digitized image background. This allows surgeons to perform any procedure in an improved, practical and real-time visual environment in a minimally invasive way, with increased accuracy and decreased radiation exposure. CAOS was first described as a hypothesis by BouazzaMaroufin 1995 (4). In the late 1990s, CAOS was first introduced into practical application (5) and was then rapidly developed. Koo et al. (6) and Seide et al. (7) developed the algorithm and prototype, respectively, to perform closed fracture reduction. The hexapod robot kinematics that Seide et al. used was first described in engineering terms as the Stewart platform by Stewart in 1965 (8), and then in medicine for limb lengthening and deformity correction by Taylor in 1994 (9) as the Taylor spatial frame. The hexapod device has several advantages, such as ease of application, computer accuracy for the correction of six degrees of freedom (6-DOF) deformities, and the ability to perform residual serial postoperative adjustments and manipulate fractures into excellent alignment in an
349
outpatient, less invasive, manner through struts adjustments only (10–12). Using the hexapod kinematics and computer technology, we designed a novel 3D hexapod CAOS system for diaphyseal fracture reduction of long bones that can provide precise parameters and convenient procedures for fracture reduction and deformity correction compared to conventional techniques (13). A novel reduction strategy, a registration and reduction technique using mirror imaging, for one-legged fracture (14,15) was also created. To address several problems we encountered in the last version (13), we advanced the system and performed serial specific tests to determine its precision, practicality and feasibility. We had endeavoured to make precise configuration changes during reduction procedures, so we used electric telescopic struts with DC motors as the actuators, whose length did not vary freely. However, when released after the procedures, the neutral position could not be regained easily. Also, the DC motors mounted on the actuators occupied too much space, limiting the surgeons’ operation and risking jamming between adjacent motors; this would be a major inconvenience, a waste of time and a burden on the surgical crew. The other problem was that the system operated with the patient’s legs placed through two rings, which could make the patients uncomfortable, jam during the procedure or occupy space during CT scans (Figure 1).
Figure 1. (A, B) Twelve specific location balls were fixed on each ring for registration. (C, D) The cadaver precision tests showed acceptable results Copyright © 2014 John Wiley & Sons, Ltd.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
350
Materials and methods Reduction strategy We developed a novel reduction strategy, a registration and reduction technique using Townsend and Richards (14) mirror imaging for one-legged fractures on the hexapod system. The physical kinematics features of the human skeletal system create human body mirror symmetry. Two mirror-symmetrical identical bones share the same physical parameters (15). The non-fractured mirror image leg should ideally match the prefractured injured leg. Thus, the mirror image can provide precise parameters for anatomical fracture reduction. Furthermore, we introduced 3D image processing techniques and programmed the corresponding software (Figure 2). For registration, mirror transformations of the contralateral long-bone models were used as reference standards for reduction. Two major fragments of long-bone fractures were made to coincide with those of the contralateral bones, and the reduction trajectory was planned according to the relative displacement transformation matrix after coincidence. The reduction procedure was then performed by the reduction unit. During image processing, we introduced a threedimensional (3D) surface stereolithography (STL) model of the fractures as our reconstruction method, based on computed tomography (CT) data from Mimics software v. 10.01 (Materialise NV, Leuven, Belgium), using the surface rendering algorithm of marching cubes (16,17) to accurately measure and analyse the geometric parameters of the bones. The 3D image was reconstructed from DICOM slices of high-resolution CT scans made of point cloud data, whose surface consists of characteristics of the femur anatomy. We considered the 3D surface STL models of the contralateral long bone after mirror
transformation to be essentially identical to the fractured long bones after reduction. Two shape-specific sections were selected for registration with the contralateral 3D image. The mapping matrix ensured that the two point clouds were the minimum distance apart, enabling the reduction robot to move the fractured point cloud to the normal femur point cloud. ICP algorithms are commonly used to compute the transform matrix (18), and we adopted the following data processes, based on an ICP algorithm, to calculate the mapping matrix for the two point clouds. The mathematical model for registration, based on the ICP algorithm, was as follows: f ¼ min
Xn i¼1
kpi ðR qi þ SÞk2 ; i ¼ 1; 2; ⋯; n
(1)
where pi stands for the points in the specific region of the two fracture fragments, qi stands for the points in the specific region of the two contralateral fragments, and R and S stand for the rotation and translation matrix after the registration of fractures. The parameters for automatic registration were obtained by finding optimal solution of this equation. If deviations arose because of the automatic registration, manual trimming was needed. For selection, priority should be given to the region with the least impurity on the 3D surface of the STL model of pi and qi.
Construction of a novel hexapod reduction configuration We designed the original hexapod configuration based on a Taylor spatial frame, which can perform any six-degree of freedom (6-DOF) movements without loss of stability or the need to change parts during any reduction
Figure 2. Operator interface of custom software program; a displaying and operating window is on the left and four categories of menu are integrated and listed on the right Copyright © 2014 John Wiley & Sons, Ltd.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
351
Advancing CAOS with a hexapod device
procedure for long-bone fractures. The two major fracture fragments were tightly attached to the proximal and distal rings and then the fracture was reduced, along with configuration changes of the hexapod device. Twelve balls at specific locations were fixed on each side of the rings for registration (Figure 1A, B). To address the problems described above, we made several improvements through a redesign after deep discussions. Figure 3A shows the new fracture reduction device in its neutral configuration, which maintains all of the original hexapod advantages.
Hybrid-driven system A hybrid-driven system, based on a motor drive and a hydraulic transmission, was used in place of the electric driving mode, and consisted of a motor, a screw and two groups of hydraulic cylinders. Six electric telescopic actuators used previously in the reduction mechanism were replaced by a group of six hydraulic cylinders as an internal hydraulic cylinder set (Figure 3A, yellow) that provided more non-working freedom and higher stability and precision during all procedures. An identical external
Figure 3. (A, B) Key hexapod construction and ordinate transformation. (C–F) Six degrees of freedom movements: (C) translation movements along the X, Y axes; (D) translation movement along the Z axis; (E) rotation movement around the X, Y axes; (F) rotation movements around the Z axis. (G, H) Upper and lower components in the series and parallel configurations: (G) parts in blues and green demonstrate lower or static components; (H) parts in red and orange demonstrate upper or dynamic components Copyright © 2014 John Wiley & Sons, Ltd.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
352
hydraulic cylinder set in the control box was connected with the internal set by hydraulic pipes and sealed. The adjustment of the hydraulic cylinder on the mechanism was controlled by driving the external hydraulic cylinder via a stepper motor drive and hydraulic transmission. Non-compressed liquid was infused into the cylinder and pipe. The rotary motion of the servo-motors was converted into reciprocating motion of the hydraulic cylinder, which transferred the power to the internal hydraulic cylinder set. The biggest advantages of this electro-hydraulic hybriddriven system are that, when released, the length of the hydraulic cylinders can be adjusted freely and randomly and, when validated, the length can be solidly fixed and precisely adjusted. This mechanical part makes the system more reliable and precise, with higher freedom of movement, while simultaneously requiring less space and facilitating the entire reduction mechanism and surgical procedure. In order to improve transfer efficiency for long-standing depreciation, the single-thread triangular screw in the last version was replaced by ball screws (lead, P = 5 mm; transmission efficiency, η = 90%). Neglecting the influence of transmission efficiency of the reduction gearbox, the output maximum force and velocity of each hydraulic cylinder can be found using the following formula: F¼
2 πMiη n ; V¼ P P 60i
(2)
Series and parallel configurations In addition to the hydraulic drive units, a new series and parallel configuration was introduced. In the previous version, the mechanism configuration was designed directly from the hexapod robot kinematics and was characterized by a parallel robot manipulator or Stewart platform, which worked with the patients’ broken legs in the rings. This configuration could jam during the reduction procedure and make the patients uncomfortable with parts of the mechanism between their legs. Thus, we designed a new series and parallel configuration, which extends a specific positioning unit outside the hexapod device. The reduction mechanism includes two major units, the positioning and reduction units (Figure 4B, C). The positioning unit uses a registration strategy and provides temporary fixation for fractures. Two flatbeds, called fracture holders, are arranged in series, connecting the two major long-bone fracture fragments for temporary fixation like external fixators (Figure 3B, green and orange; Figure 4C). Two side bars fix the holders together Copyright © 2014 John Wiley & Sons, Ltd.
as the positioning unit, which is relatively independent and can be disassembled and separated from the reduction system. The reduction unit consisted of the core hexapod mechanism below the positioning unit and fracture (Figure 4C, D). The holders and two rings of the hexapod mechanism could be assembled as a solid separate unit called the upper and lower components. In the lower components, also called the static components, the fracture holder stays parallel to the lower ring (Figure 5A, green and blue). Similarly, in the upper component, also called the kinetic/dynamic component, the holder is parallel to the upper ring (Figure 5B, orange and red). This construction is the key configuration change: the parallel configuration of the hexapod mechanism (Figure 3A) evolved to a series connection (Figure 3B) and the previous 6-DOF hexapod kinematics movement was changed to a new form. Fracture movements could be split into six simple movements in the 6-DOF 3D coordinate system (Figure 3C–F). The Z axis was defined as the longitudinal axis of the bone. The X and Y axes were perpendicular to the Z axis and to each other. The 6-DOFmovements were defined as translations along and rotations around the three axes, specified by angulation (rotation around the X and Y axes), translation, length discrepancy and rotation (rotation around the Z axis). In the previous version of this system, the two main fragments of each long-bone fracture were validly connected to the two rings and the 6-DOF relative movements of the rings were directly converted to 6-DOF movements of the two main fragments (Figure 1). In this new system version, static components contain the core hexapod mechanism for execution of the 6-DOF movements, which translated to relevant movements of the two holders in series. Two 6-DOF movements of the hexapod device and the holders with major fracture fragments were vertically adapted to each other. For example, Z axis length variations of the hexapod mechanism would convert to translation variance in X axis holders. In the previous version, the two major fracture fragments were fixed onto the mechanism before a highresolution computed tomography (CT) scan. The whole system was large and heavy to move, which caused patients too much discomfort and might not fit into the CT machine or could jam during procedures. In addition, the reduction mechanism and fractures had to remain attached together during the entire procedure. However, in this new version, the fracture tray was made relatively independent. The positioning unit can be disassembled from the reduction unit. In the clinical setting, after hospitalization, fractures are attached to the holders of the positioning unit immediately and secured through temporary fixation and Schanz nails. Afterwards, high-resolution CT scans are taken of both the broken and unbroken legs. This new Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
353
Advancing CAOS with a hexapod device
Figure 4. (A–D) Details of the new construction: α, fracture holder; β, side bar
positioning unit construction is much simpler, can save a lot of space for reduction procedures and gives patients more comfort. After the CT scans, the holders are rigidly attached to the reduction unit and components. Next, the hydraulic drive units are locked. After disassembling the two side bars, the lower and upper components become two relatively dependent parts that are connected by six parallel hydraulic drive units. Thereafter, the imaging process and pre-operative planning is conducted, followed by the reduction procedure.
Neutral configuration The core hexapod mechanism of the reduction unit has a neutral configuration, with all six internal cylinders 427 mm in length. In the neutral configuration, the two fracture holders of the positioning unit can be assembled into corresponding components for the subsequent reduction. The neutral configuration is the exact position where the reduction procedure starts. After the procedures, the Copyright © 2014 John Wiley & Sons, Ltd.
reduction unit would be in a specific pose and the positioning unit would be replaced by traditional hybrid external fixators and separated from the reduction unit. At that point, fractures are permanently fixed. Then the procedure is over and the reduction unit automatically resets to its neutral configuration via the control panel, ready for another procedure.
Registration strategy We have also introduced a new registration strategy on the positioning unit. The key point concerning the reduction procedure in this system lies in the position and orientation changes to the hexapod device, realized by adjusting the length of six hydraulic cylinders, determined by the 12 cylinder endpoints. However, it was difficult to locate the endpoints because of the complicated structure of the ball joints in the 3D image. Previously, we created a location ball structure to locate the endpoints for conformation and determining the mechanism Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
354
Figure 5. Steps involved in the fracture reduction procedure in the new construction
transformation. Because there is a total of 12 location balls to process in the 3D image (Figure 1), this task would take considerable work and time and potentially introduce variability. To address this issue, we designed a positioning unit system to facilitate and simplify the process by introducing a new algorithm for the registration and reduction procedures. After the CT scan, the positioning unit assembled the reduction unit into its neutral configuration. The two fracture holders are tightly attached with the static and dynamic rings together, to build the upper and lower components. These two constructions are then unchangeable during the entire registration and reduction procedures. This arrangement, with the positioning unit outside of the reduction unit, is more convenient and precise. Because four single points in space determine a spatial position and attitude, four specific marking points must be chosen on the holder corners to act as feature points that are easy to locate, to determine the spatial configuration of the fractures and practical hexapod construction illustrated in Figure 4B (green square). Strikingly, the hexapod device is no longer necessary for the preoperative CT scan. The CT scans of patients can be taken with the fractures attached to the positioning unit only, and the 3D image involved contains only the broken and unbroken limbs and the positioning unit. This set-up facilitates and simplifies the image-analysis process and Copyright © 2014 John Wiley & Sons, Ltd.
reduction procedure. Unlike the previous system version, here we only need to locate four non-specific marking points on each holder. After attaching the holders to the rings, the surgeon can locate the position and altitude of the hexapod device through the image-analysis process. This registration strategy may be more comfortable for patients and simplify the registration and reduction procedures. A custom C++ and Visualization Toolkit (VTK) software algorithm was programmed to account for the new registration strategy and the changes made to the reduction configuration.
Safety strategy Working space analysis The working space of the construction for the conversion of the coordinates and 6-DOF movement mode was analysed. The working space determines the extent of fracture deformity to be corrected. According to hexapod kinematics, working space is mainly affected by three main factors, the diameter of the rings, the nominal length and the variation length of the actuators. For the new configuration, we made several major changes. Both rings in the previous version had the same radius. However, to save space for the holders in the new version, the ring in the upper component was made slightly Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
Advancing CAOS with a hexapod device
smaller. With two attached components, the working space of the hexapod device determines that of the holders. Their working space is analysed below. The core structure of the hexapod device has two rings, radius 110 and 195 mm, for the upper and lower components, respectively, and six actuators, 331 mm in basic length with a 200 mm length variation. In this configuration especially, it has a working space with –220 to +220 mm of translation and –43° to +43° of rotation in the coronal axis, –20 to +250 mm of translation and –35° to +35° of rotation in the sagittal axis, and –95 to +95 mm of translation and –130° to +130° of rotation in the vertical axis (Figure 3). Compared with the previous version, the working space of the new version is significantly increased. The fracture holder motions come from the hexapod device movements. Two 6-DOF movements of the hexapod device and the holders with major fracture fragments were vertically adapted to each other. The mechanism itself and the motion transfer from the reduction unit to the positioning unit increases the complexity of the practical working space of the fracture holders. For example, Tz translations in the fracture holders derive from Tx translations of the hexapod device and Ty rotation in the fracture holders derives from Tz rotations in the hexapod device, e.g. a translation of 1 mm (Tz) in the fracture holders would cause a 1 mm (Tx) translation in the hexapod device. Thus, the virtual working space of the fractures holders is –220 to +220 mm of translation and –43° to +43°of rotation in the coronal axis, –20 to +250 mm of translation and –35° to +35°of rotation in the vertical axis, and –95 to +95 mm of translation and –130° to +130° of rotation in the sagittal axis, which is determined by that of the hexapod device (Figure 3). The virtual working space in the new design is also significantly improved and can cover most deformations. Even for degrees of deformity fracture that exceed the working space, appropriate pre-operative traction can be added to reduce the degree of deformity. Pre-operative trajectory planning Cortical abutment and bone spike overlap are common in fractures. The direct reduction trajectory has a high risk of spike collision. For a smooth and automatic reduction procedure, we introduced a pre-operative trajectory-planning algorithm to avoid iatrogenic trauma during the entire procedure, based on the shortest linear distance and safety strategy. Three fundamental principles were used: (a) necessary distraction; (b) minimization of the required forces induced by moving the fragment (soft-tissue balance); and (c) avoidance of unnecessary movements. All planned procedures conformed to these principles. For standardization, we created a novel long-bone fracture classification system based on fracture pattern. Copyright © 2014 John Wiley & Sons, Ltd.
355
Distraction in advance is intended to regain a broken limb’s normal length and provide extra room for deformity correction. Beyond regaining the limb length, overdistraction of the fracture fragments by 5–10 mm is recommended (19). In addition, Seide et al. (20) recommended that fractures should first be distracted to a slight over-distraction by ca. 2 mm of least reduction room. With the least reduction room 2 mm in the distraction externum 10 mm, the 8 mm of overlap length after the procedure could makes a completely different reduction strategy. There is up to 8 mm of room for the overlapping parts of the main fracture fragments spike. Thus, the fracture classifications we created were mainly based on the overlap length after reduction. We divided the femoral shaft fractures into three main types: A, no overlap length; B, overlap length < 8 mm; and C, overlap ≥ 8 mm. For type A and B fractures, the procedures are relatively easy to perform. The basic treatment steps include: 1, careful evaluation, classification and measurement of the overlap length; 2, adequate over-distraction (no more than 10 + 2 mm for the least reduction room); 3, alignment correction; and 4, contraction and compression. However, in type C fractures with overlap length ≥ 8 mm, simultaneous alignment is not possible. The initial overdistraction must be limited within 10 mm, so there may not be enough room for simultaneous alignment; bone spikes could collide. Fractures with face-to-face surfaces are classified as subtype C1 and back-to-back as subtype C2. For fractures of type C1, the over-distraction, rotation and angulation correction processes listed above would manoeuvre the two fracture surfaces into a face-to-face configuration, and the inclined surface plane might provide enough room for subsequent translation correction; but for C2 fractures, the back-to-back bone spikes need a revolution to bypass each other before a transformation from the back-to-back configuration into a face-to-face one. Contraction and compression would then finish the entire procedure. Closed-loop control system We used a personal computer (Intel® Core TM 2, 3.00 GHz) as a controller and ran a custom software program in VC++ to prepare the 3D reconstruction, to register the CT image data, to create the signal procedure for the sensor and to plan the trajectory and function of the motion control. A six-axis motion control card (ADT-8960) was used for communication, based on the PCI bus, to control the stepper motors motion by providing pulse signals. In order to achieve synchronous motions of the external and internal hydraulic cylinders, we introduced a doubleposition, closed-loop control system. The encoder signals were fed back to the motor driver to achieve an inner position closed loop to decrease the errors caused by stepper-motor motion (Figure 4). The linear displacement Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
356
transducer that was fixed on the internal hydraulic cylinders in the hexapod mechanism fed back displacement signals to the PC via the data acquisition card, to realize the external position closed-loop control for decreasing errors caused by the hybrid-driven system based on the hydraulics and motor. Indeed, the double-position closed-loop control guaranteed the position accuracy of the robot. Taking the 6-DOF movements of the hexapod device into account, we used a binocular visual tracker (Micron Tracker MTC3.0, Claron Technology Inc., Toronto, Canada) to test the precision of the hexapod mechanism. Within the range of the working space, translation errors were < 2 mm and angulation errors were < 1.2°. Fractures that heal with an angular deformity > 5° in any anatomical plane, or a translational deformity > 10 mm are regarded as malunions (21). These test results showed that the hydraulic drive parts and reduction construction could provide excellent precision, while occupying less space compared with electric telescopic struts. Other improvements We also added several security measures. Emergency brakes were installed in both the software and the control box. Mechanical sensors were fixed on the positioning unit for detecting soft-tissue interference.
Results To verify the feasibility and accuracy of reduction, we used a pair of bovine femurs as a fracture model, and one of the bovine femurs was broken to simulate a transverse fracture (Figure 5). Before modelling, CT scans of both intact femurs were taken for data back-up. These CT data were used to reconstruct the original 3D image for residual deformity measurement and analysis. Figure 5 shows the routine steps of the reduction procedure for a new fracture reduction: 1, attach the fractures rigidly to the holders of the positioning unit for temporary fixation; 2, take a pre-operative high-resolution CT scan for both limbs; 3, load the DICOM CT data into the software program for 3D image reconstruction, registration, image-analysis processing and pre-operative planning; 4, perform the reduction procedure; 5, after the reduction, replace the fracture holders of the positioning unit with traditional hybrid external fixators; and 6, take another CT scan for postoperative precision testing. Using equation (2), the maximum force and velocity of hybrid-driven system were calculated as F = 1243 N and V = 10 mm/s. Based on these results, the output force and velocity met the demands of the reduction system, as the system needed a lower velocity for safety. Within the range of the working space, translation errors of the Copyright © 2014 John Wiley & Sons, Ltd.
hexapod device measured from MicronTracker were 2 mm and angulation errors were < 1.2°. The sites where we choose to insert Schanz nails for fixation were away from the peripheral nerves, arteriovenous system and tendons, in case of iatrogenic injury and limitation of activity, such as in the lateral femur or medial tibia. Each major fragment requires at least two Schanz nails for stabilization to prevent rotation. Reduction procedure experiments were carried out and the precision of this reduction prototype was tested seven times on this in vitro bovine femoral fracture model. All the procedures followed the routine steps shown in Figure 5. After a procedure, the residual deformity was measured by the same operator under direct visualization of the fracture sites. The four parameters of axial deflection, translation, angulation and rotation were collected and used to evaluate precision (Tables 1 and 2). Based on a definition that reduction with an angular deformity > 5° in any anatomical plane, or a translational deformity > 10 mm, are regarded as malunions (21), the reductions achieved excellent union. In addition, the translation and rotation corrections were better than the precision test results from the previous prototype version (p < 0.05) (13). The hexapod device was able to return to its neutral position after each reduction procedure. During the seven tests, several path points were each randomly and repeatedly chosen to determine precision intra-operatively. Two rough parameters were collected, displacement and angulation deviation, and compared with the corresponding points in the trajectory planning software window. All displacement deviations were < 1.8 (average < 1) mm and angulation deviations were < 1.5°. No scratches or jamming of the prototype occurred.
Discussion The operative realignment of mechanical axis deviations (MAD) is necessary to prevent early joint degeneration and to reach normal load-bearing capacity. However, Table 1. Practical residual deformity value collected from postoperative measurements of seven precision tests Test No. 1 2 3 4 5 6 7
Axial deflection (mm)
Translation (mm)
1.1 2.0 0.5 0.9 0.8 2.1 1.2
0.6 1.9 1.3 0.6 0.7 1.2 1.0
Angulation Rotation (°) (°) 4.2 2.6 3.0 2.9 0.6 2.1 2.3
2.5 1.2 3.2 1.7 2.1 4.1 2.4
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
357
Advancing CAOS with a hexapod device Table 2. Results from bovine femoral repositions without soft tissues (n = 7 times) Parameters Axial deflection (mm) Translation (mm) Angulation (°) Rotation (°)
Average
SD
Minimum
Maximum
1.23 1.04 2.53 2.46
0.60 0.47 1.09 0.96
0.5 0.6 0.6 1.2
2.1 1.9 4.2 4.1
achieving perfect reduction accuracy and allowing early functional exercise using the AO technique, or using soft tissue preservation to achieve good healing outcomes for fracture treatment, presents a difficult dilemma. How to best combine the respective merits of the different approaches is an important scientific concern. The AO technique reduces fractures through open reduction and internal fixation (ORIF) to reconstruct the MAD, which can give rise to soft tissue and periosteum damage and haematoma drainage around the fracture site. The priority of the BO technique is avoiding such damage by preserving the native biological environment. Several complications, such as radiative overexposure, poor accuracy and high surgeons’ experience requirements, restrict the practical application of these techniques. For example, using a fluoroscopic technique, good reduction can be achieved, at the cost of excessive radiative exposure to surgeons, through a minimally invasive procedure that requires an experienced surgeon. Biological preservation does not necessarily mean a small incision and minimizing MAD is difficult, as is performing the minimally invasive percutaneous plate osteosynthesis technique (22). In recent years, CAOS was developed to improve the accuracy of surgical procedures (23,24). CAOS-based minimally invasive techniques could provide better healing outcomes by maintaining the soft tissue attachments and blood supply, and minimizing MAD through computational techniques. CAOS technology has developed several algorithms and prototypes. However, no current prototype has demonstrated all the potential benefits and limitations. The da Vinci surgical robot is the best example of this kind of computer-assisted surgical tool (25). The role of such technology is limited at present, as surgeons must be proficient in traditional surgical approaches before mastering and using this technology and equipment. To achieve better precision and lower experience requirements and cost, we developed a reduction system based on a mirror registration technique and hexapod kinematics. The physical kinematic features of the human skeletal system create human body mirror symmetry. Two mirror-symmetrical identical bones should share the same physical parameters (15). Through computer-imaging processes, we obtained precise anatomical reduction parameters to assist fracture treatment. Copyright © 2014 John Wiley & Sons, Ltd.
To execute the reduction procedure, a custom-made reduction construction prototype was designed, manufactured and improved (13). In the present study we proposed and tested two major improvements in the construction of a new reduction mechanism, a hybrid-driven system and a series and parallel configuration. The present reduction system adopted a hybrid-driven system, based on a motor drive and hydraulic transmission instead of the direct driving mode that was used in the previous prototype version. This system provides more precision and occupies less space. In this new prototype, the electro-hydraulic hybrid-driven actuators are larger and provide more power, but occupy a relatively small space. The precision tests of the hybriddriven system also reported good results. Another improvement to the prototype construction is the series and parallel configuration. The serial mechanism has a large workspace and a flexible mechanism but poor load capacity, and occupies a large operative workspace. Compared with this serial mechanism, the parallel mechanism has higher position accuracy and greater load capacity and rigidity, but a smaller workspace. The features of each mechanism are complementary to one another. Therefore, we combined them into a series and parallel configuration (Figure 4): the parallel part is the hexapod device, as a reduction unit, and the series part is the positioning unit, with the fracture holders for temporary fixation and registration. The two fracture holders can be disassembled from the upper and lower components, and the fractures are attached to the holders using temporary fixation. As an improvement, we now use registration points on the holders themselves, so that the hexapod device does not need to be attached during the CT scan. Compared with the previous version, patients will be much more comfortable during the CT scan, with only the holders in place. This change also simplifies analysis of the CT data compared with the previous version, because the hexapod device is not included and the 3D images are smaller and contain fewer artifacts that could influence registration and pre-operative planning. In the current version, we present a new registration strategy. In the 3D reconstruction image, instead of the previous method of identifying balls, a series of feature points were taken as markers that were easily identified on the fracture holders to perform the conversion of the coordinate system. This change improves the convenience, flexibility, ease of operation and portability associated with the system. After executing the reduction plan, residual deformity could remain in some cases during clinical practice. The post-reduction CT scans are meant to test the procedure precision and determine any remedial reduction procedures. We discussed additional safety strategies to guarantee a smooth and automatic reduction procedure. The working space defines the degree of deformity of a fracture. In this Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
H. Du et al.
358
new construction, we used a larger configuration with more working space, which could deal with more complicated fractures. If deformity exceeds the working space, a pre-operative rough alignment might be necessary to decrease the degree of deformity before the reduction procedure. Of all the safety strategies discussed, we considered trajectory planning to be the top priority pre-operatively. During any reduction procedure, peaks of the bone fragment could collide when using a direct trajectory. Therefore, a precise pre-operative trajectory-planning analysis is needed to avoid iatrogenic trauma. The intra-operative reduction precisions we tested were varied, but the displacement deviations were < 1.8 (average < 1) mm and angulation deviations were < 1.5°. Graham (26) has reported that a simple midshaft fracture had a maximum resulting force of 428 N. Owing to the strong power of muscles, mechanical sensors would need, to a relatively high threshold, to avoid soft-tissue iatrogenic injury. In this case, iatrogenic fracture from a spike collision could be avoided through careful pre-operative trajectory planning instead of mechanical sensors. Fracture pattern classification and reduction trajectory preplanning and analysis should be performed independent of which reduction technique will be used. For clinical practice, we suggest a minimum 1 mm gap between the edges of fracture spikes in pre-operative trajectory planning, in case of intra-operative collision below the threshold. We maintained our closed-loop control system for security and to improve precision. The encoder signals were fed back to the motor driver to achieve an inner position closed loop, to decrease the errors caused by the hybrid-driven system based on the hydraulic and motor. If a jam happens during the reduction procedure as a result of soft-tissue involvement, the hexapod device cannot reach its final position and the sensors on actuators would feed back to the motors to continue adjustment until the final position is reached. Among the improvements we made, the simplification of the configuration, the registration strategy and the closed-loop control system contribute to the improvement in precision. A positioning unit was designed to fix the fracture that simplified the system. The CT scans are conducted with patients wearing the positioning unit only, so fewer CT data were collected and the data were easy to process. In addition, four specific marking points on the fracture holder corners were taken as feature points in place of the location balls structure, which also simplified the course of registration. The closed-loop control system could also compensate for residual deformity to improve precision. Two side bars and two fracture holders are involved in the positioning unit and after each procedure the reduction unit would return to a specific pose, and the Copyright © 2014 John Wiley & Sons, Ltd.
positioning unit was then replaced by traditional hybrid external fixators and disassembled from the reduction unit. The configuration of the positioning unit with four parts and the disassembly are complicated and could lead to precision loss. As a future improvement, simplified unilateral serial external fixators with a universal joint and a telescopic sleeve could be introduced. Such a new fixator would include special connectors with the reduction unit and registration marks. Patients would have their fractures attached to the new fixators, with the universal joint and telescopic sleeve locked before the CT scan, and be connected with the reduction unit. During the reduction procedure, the new fixators would be unlocked and disassembled from the reduction unit. After the procedure, they would be locked and used as traditional external fixators. This configuration would further simplify the system. We will also try to expand the capacity of this hexapod reduction system to treat various long-bone fractures. In conclusion, the hexapod reduction system developed in this study allows surgeons the ability to perform fracture reduction in an objective, efficient and accurate manner, while minimizing the radiation exposure to both the patients and the surgical team. Using the mirror image of the unbroken limb as a template to guide the reduction procedure, surgeons can concentrate on preoperative planning and supervise the reduction procedures. The hexapod reduction system may be able to reduce long-bone diaphyseal fractures more precisely and anatomically, and also minimize the extent of periosteum and soft-tissue damage around the fracture site.
Acknowledgements This project was funded by the National High Technology Research and Development Programme 863 (Grant No. 2012AA041604), the National Key Technology R&D Programme (Grant No. 2012BAI14B02), the Military Medical Research Programme (Grant No. BWS11J113) and the National Natural Science Fund (Grant No. 31271000).
Conflict of interest The authors have stated explicitly that there are no conflicts of interest in connection with this article.
Funding No specific funding. Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
Advancing CAOS with a hexapod device
References 1. Leloup T, El Kazzi W, Schuind F, et al. Conception of a navigation system controlling diaphyseal fracture reduction treated with external fixation. Int J Med Rob 2009; 5: 99–109. 2. Stephen DJ, Kreder HJ, Schemitsch EH, et al. Femoral intramedullary nailing: comparison of fracture-table and manual traction. a prospective, randomized study. J Bone Joint Surg Am 2002; 84A: 1514–1521. 3. Weil YA, Greenberg A, Khoury A, et al. Computerized navigation for length and rotation control in femoral fractures: a preliminary clinical study. J Orthop Trauma 2014(Feb); 28(2): e27–e33. 4. Bouazza-Marouf K, Browbank I, Hewit JR. Robotic-assisted internal fixation of femoral fractures. Proc Inst Mech Eng H 1995; 209: 51–58. 5. Merloz P, Tonetti J, Pittet L, et al. Pedicle screw placement using image guided techniques. Clin Orthop Relat Res 1998 (Sep); 354: 39–48. 6. Koo TK, Chao EY, Mak AF. Development and validation of a new approach for computer-aided long bone fracture reduction using unilateral external fixator. J Biomech 2006; 39: 2104–2112. 7. Seide K, Faschingbauer M, Wenzl ME, et al. A hexapod robot external fixator for computer assisted fracture reduction and deformity correction. Int J Med Rob 2004; 1: 64–69. 8. Stewart D. A platform with six degrees of freedom. Proc Inst Mech Eng 1965; 180: 371–378. 9. Taylor JC. Perioperative planning for two- and three-plane deformities. Foot Ankle Clin 2008; 13: 69–121, vi. 10. Al-Sayyad MJ. Taylor spatial frame in the treatment of pediatric and adolescent tibial shaft fractures. J Pediatr Orthop 2006; 26: 164–170. 11. Sala F, Elbatrawy Y, Thabet AM, et al. Taylor spatial frame fixation in patients with multiple traumatic injuries: study of 57 long-bone fractures. J Orthop Trauma 2013; 27: 442–450. 12. Dammerer D, Kirschbichler K, Donnan L, et al. Clinical value of the Taylor spatial frame: a comparison with the Ilizarov and Orthofix fixators. J Child Orthop 2011; 5: 343–349.
Copyright © 2014 John Wiley & Sons, Ltd.
359 13. Tang P, Hu L, Du H, et al. Novel 3D hexapod computer-assisted orthopaedic surgery system for closed diaphyseal fracture reduction. Int J Med Rob 2012; 8: 17–24. 14. Townsend G, Richards L. Twins and twinning, dentists and dentistry. Aust Dent J 1990; 35: 317–327. 15. Hu L, Zhang J, Li C, et al. A femur fracture reduction method based on anatomy of the contralateral side. Comput Biol Med 2013; 43: 840–846. 16. Weng YYN, Pierson R. Three-dimensional surface reconstruction using optical flow for medical imaging. IEEE Trans Med Imaging 1997; 16: 630–641. 17. Lorensen WE, William EL, Harvey EC. Marching cubes. A high resolution 3D surface construction algorithm. Comput Graphics 1987; 21: 163–169. 18. Paul JB, Neil DM. A method for registration of 3D shapes. IEEE Trans Pattern Anal Mach Intell 1992; 14: 239–256. 19. Starr BR, Bucholz RW, Heckman JD. Fractures of the shaft of the femur. In Rockwood and Green’s Fractures in Adults, Bucholz RW, Heckman JD (eds). Lippincott, Williams & Wilkins: Philadelphia, PA, 2002; 1683–1730. 20. Seide K, Wolter D, Kortmann HR. Fracture reduction and deformity correction with the hexapod Ilizarov fixator. Clin Orthop Relat Res 1999(Jun); 363: 186–195. 21. Koo TK, Mak AF. A knowledge-based computer-aided system for closed diaphyseal fracture reduction. Clin Biomech 2007; 22: 884–893. 22. Krettek C, Muller M, Miclau T. Evolution of minimally invasive plate osteosynthesis (MIPO) in the femur. Injury 2001; 32(suppl 3): SC14–SC23. 23. Renkawitz T, Tingart M, Grifka J, et al. Computer-assisted total hip arthroplasty: coding the next generation of navigation systems for orthopedic surgery. Expert Rev Med Devices 2009; 6: 507–514. 24. Jackson DW, Simon TM. History of computer-assisted orthopedic surgery (CAOS) in sports medicine. Sports Med Arthrosc 2008; 16: 62–66. 25. Sterrett SP, Jarrard DF. Robotic assisted laparoscopic radical prostatectomy: evolution and outcomes. Minerva Urol Nefrol 2008; 60: 31–39. 26. Graham AE, Xie SQ, Aw KC, et al. Bone–muscle interaction of the fractured femur. J Orthop Res 2008; 26: 1159–1165.
Int J Med Robotics Comput Assist Surg 2015; 11: 348–359. DOI: 10.1002/rcs
