SEARCH

SEARCH BY CITATION

Keywords:

  • robot-assisted surgery;
  • three-dimensional visualization;
  • robotic fracture reduction;
  • analysis of precision;
  • femur fracture

Abstract

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Closed fracture reduction can be a challenging task. Robot-assisted reduction of the femur is a newly developed technique that could minimize potential complications and pitfalls associated with fracture reduction and fixation. We conducted an experimental study using 11 human cadaver femora with intact soft tissues. We compared robot-assisted fracture reduction using 3D visualization with manual reduction, using 2D fluoroscopy. The main outcome measure was the accuracy of reduction. The manual reductions were done by an experienced orthopedic trauma surgeon, whereas the robot-assisted reductions were done by surgeons of different experience. The robot-assisted group showed significantly less postreduction malalignment (p < 0.05) for internal/external rotation (2.9° vs. 8.4°) and for varus/valgus alignment (1.1° vs. 2.5°). However, the reduction time was significantly (p < 0.01) longer (6:14 min vs. 2:16 min). The higher precision associated with robot-assisted fracture reduction makes this technique attractive and further research and development worthwhile. In particular, less experienced surgeons may benefit from this new technique. © 2010 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 28:1240–1244, 2010

Closed reduction and minimally invasive stabilization techniques help preserve the biologic environment of fractures, leading to enhanced union rates and decreased infection rates.1–4 However, closed reduction can be technically demanding. Three main problems exist: (1) manipulation of the fragments without exposing the fracture site; (2) assessment of the correct reduction; and (3) temporary stabilization.5, 6 These problems are especially evident in femoral shaft fractures. The thick muscle envelope leads to high counteracting forces and torques, and the central position of the shaft makes direct manipulation of the fragments difficult.7–9 In addition, 2D fluoroscopy provides suboptimal radiographic visualization with regards to the precision of reduction, particularly considering the tubular shape of the femoral shaft. Even intramedullary nails, which can be used as self-aligning implants, cause malalignment, especially rotationally.4, 10–13

New devices and techniques have recently been developed to support fracture reduction. From devices like the AO femoral distractor, development has lead to the implementation of navigation systems to aid the surgeon, for example, by enhanced visualization.7, 14–21 But navigation does not solve the precise manipulation of the bone fragments and their retention in the reduced position until fitting the fracture fixation device. A robot application may be expedient.

The implementation of a robotic system into the process of fracture reduction meets two challenges. It facilitates the assessment of the fracture by intelligent and interactive visualization and enables intuitive fracture manipulation. Further, the robots' ability of rigid fracture retention after reduction coincides with temporary fixation.

The idea of robot-assisted fracture reduction was first described in 1995 by Bouazza-Marouf et al.22 Ten years later the first experiments using synthetic composite bones were presented23 demonstrating robot-assisted fracture reduction to be highly accurate and to require less radiation. A subsequent study on human cadavers showed a decreased radiation time. Nevertheless the reduction accuracy could not be improved significantly compared to the manual reduction.24 Reduction control in the first study was facilitated by two orthogonal cameras, whereas in the second study a 2D image intensifier was used. These findings demonstrated that visualization is a crucial factor in achieving accurate results. Since the cameras only showed a surface image of the fragments, we postulate that a significant improvement of the final fracture alignment by 3D fragment visualization is one element of the robotic setup. Further setup elements like the ability of rigid retention and the feedback of forces at the fracture site contribute to this improvement.

The purpose of our study was to compare manual fracture reduction using 2D fluoroscopy with robot-assisted reduction using 3D fracture visualization software. The focus of imaging was on the fracture site. Because of the limited scanning volume (12 × 12 cm2) of intraoperative 3D fluoroscopy, it was impossible to represent the entire femur. A recent publication emphasized the importance of natural forms of human communication in a human/machine interface, described as “AHMI—Advanced Human Machine Interface.”25 To meet one of the demands of natural human environmental perception, we chose to implement an interactive 3D surface model.

The revolutionary aspect of this method is that a (limited) virtual direct view of the fracture site is used to align the entire femur, disregarding the proximal and distal aspects of the bone. This contrasts with the conventional method, paying less attention to the fracture site rather than to axes, length, and rotation of the femur. We hypothesized that robot-assisted fracture reduction with 3D visualization would be more accurate and faster than the conventional method and that a homogenous result among the operating surgeons would be achieved.

METHODS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We established an experimental study on human cadavers with artificial fractures of the femur shaft. These fractures were reduced by robot-assistance with 3D fracture visualization in group one and manually using 2D visualization in group two. Our main outcome was the difference in postreduction alignment between groups. Moreover, we looked for interindividual differences within the robotic group. Secondarily, the reduction time between both groups was compared.

Robot Set-Up

We used the clean room version of a 6 degree-of-freedom industrial robot (RX 90c, Stäubli Tec-Systems, Faverges, France) as a manipulator. The manipulator worked in a master-slave mode, which means that the surgeon directly controlled the motion of the robot. The robot was equipped with a force/torque sensor (FT Delta SI-660-60; Schunk, Lauffen, Germany) and connected via an ISA card to a standard PC (Pentium 4 2.8 GHz) running Microsoft Windows 2000. The surgeon's main input device was a standard computer joystick with force feedback (Microsoft, SideWinder ForceFeedback 2). An optical navigation system (VectorVision, BrainLAB, Munich, Germany) was interfaced via a second Ethernet TCP/IP socket connection to the controller PC.

We used the same technical set-up for the robot that was described before with the following modifications:24 The 2D image intensifier was changed for a 3D image intensifier (Siremobil Iso-C-3D, Siemens AG, Medical Solns, Erlangen, Germany). This image intensifier was equipped with the localization crown for navigated applications. For 3D referencing, localization, and tracking, we used the above-mentioned navigation system. Microsoft Windows along with a custom software designed for segmentation, visualization, and interactive robot control was used for data processing. All components of the setup were linked together with a simple Ethernet network connection for data transfer.

The Siemens Iso-C-3D acquired a 3D DICOM data set first, which was simultaneously localized by the BrainLAB navigation system. Thereby, the transformation of the DICOM coordinates to the DRB, rigidly fixed to the proximal femur fragment, was computed. These topological referenced data were then transferred to the PC that controls the robot. Thus a threshold-based segmentation of the major fragments was computed. Subsequently, the 3D surface models were reconstructed using the marching cube algorithm. The main axis of the proximal bone fragment was identified for later visualization. The 3D surface model was available to the surgeon on the monitor of the robot controlling PC (Fig. 1A). The surgeon could choose the viewing direction by pushing or pulling the top button of the joystick. The 3D surface model then rotated interactively around its main axis (Fig. 1B). This allowed the surgeon to explore the anatomy of the fracture site before, during, and after the reduction. By pressing the appropriate button, the joystick could be set to either command, translational, or rotational movements. By intuitively moving the joystick, the distal fragment was manipulated by the robot. Information about resultant forces and torques at the fracture site were transferred back to the surgeon from the sensor to the force-feedback joystick.

thumbnail image

Figure 1. (A) The surgeon interacts via the robot-controlling PC. The input device is a force-feedback joystick. Forces, tensions, and bone collisions are transferred back to the surgeon. (B) Screenshot of the 3D visualization of the fracture site. The surgeon chooses the viewing direction by pushing or pulling the top button of the joystick. The 3D surface model then rotates interactively around its main axis.

Download figure to PowerPoint

In case of an excessive force or torque above a preset threshold value, the software stopped the robot. Also, we used a pneumatic safeguard to obtain redundancy in the safety aspects. Whenever the force or torque exceeded the preset threshold (500 N and 80 Nm, respectively), the safeguard caused an emergency stop of the robot controller. These thresholds were set close to the expected maximum forces and torques to protect patient, personnel, and equipment.

We utilized high forces for our manipulations at the thigh with its strong soft-tissue envelope. The software reduced the speed linearly as the force increased above 110 N. The lowest speed was reached at a force of 250 N. Manipulation was still possible, as far as the intended movements did not lead to an additional increase in the force. In a previous study, we evaluated increasing forces during manual femoral fracture reduction. Maximum forces up to 411 N and torques to 74 Nm could be monitored,8 but these excessive forces caused by sudden jerky movements.

Pretesting

Robotic reduction trials were performed on 14 human cadaver femora. The reductions were done by four surgeons of diverse experience, from 1st year of clinical practice to 6th year of surgical training.

Specimen Preparation

This study was approved by our Institutional Review Board. The minimum sample size was calculated based on previous studies.23, 24 We expected the highest differences for malrotation. The mean rotation malalignment with a similar manual reduction technique was 5.1°,24 whereas the mean difference for the robot-assisted reduction with surface visualization was 0.8°.23 With a power of 0.8 and a p-value of 0.05, the minimal sample size was 12. We used 7 embalmed cadaver specimens (14 femurs); 3 other femora were not selected due to prior total hip arthroplasty (THA) or proximal femoral nailing. So 11 out of the 14 femora could be used for the study. This set the power on 0.789 instead of 0.8.

To quantify the femoral alignment after reduction, we used the “reverse fracture model.” A similar model was previously described for pelvic reduction.26 Our model calculates the relative transformation between the proximal and the distal segments of the femur using two reference bases of an optical navigation system. The bases are mounted onto the femur before the bone is fractured. This transformation is saved as the reference transformation for the intact femur. This model allows quantification of the reduction alignment in relation to the intact phase. The reference bases could be removed and remounted in the same position and orientation. The maximum angular error was 1.8°. The description and validation of this model was previously established.24 The femur was then fractured by 3-point bending; fractures where classified according to the AO classification system.27

Reduction Test

All 11 femora were reduced by the 4 trained surgeons using the robotic setup (Fig. 2). When the surgeon was satisfied with the reduction result of both main fragments on the computer screen, the trial was stopped. The relative residual malalignment was recorded as a 3D matrix and was the main outcome measure. Also, we recorded the reduction times. In the control group, the same 11 femora were reduced manually. Percutaneous joysticks were used for manipulation of both main fragments.7 These reductions were performed by an orthopedic trauma surgeon with 11 years experience. The 2D mode of the image intensifier was used.

thumbnail image

Figure 2. The robot manipulates the distal femoral fragment of the specimen percutaneously. The image dataset is provided by a 3D isocentric fluoroscope. Tracking of the femoral fragments and the robot is performed by a navigation system.

Download figure to PowerPoint

The data were evaluated using SPSS (Version 11.5, SPSS, Inc., Chicago, IL). Two-way ANOVA was used to compare postreduction malalignment between groups. The null hypothesis was that no difference would be found in postreduction angulation and reduction time. Within the robotic group, the null hypothesis was that no differences would be found in postreduction angulation between the surgeon groups. A one-way ANOVA was used to compare the reduction time between groups. The null hypothesis was that no difference would be found in the reduction time. Significance was set at p < 0.05.

RESULTS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The postreduction malalignment in the robot-assisted group was significantly lower (p < 0.05). Eight fractures were type A, and three were type B.27 The robotic reductions were successfully executed in all 11 femora; no emergency stop of the robot was recorded. Multiple decided comparisons showed the differences for rotation around the shaft axis and for rotation in the coronal plane (varus/valgus) (Fig. 3, Table 1). No significant differences among the four surgeons in the robot-assisted group were found. The reduction time was significantly longer (p < 0.01) in the robot-assisted group. The mean reduction time was 6:14 min (±4:52) in the robot-assisted group and 2:16 min (±0:43) in the manual group.

thumbnail image

Figure 3. The postreduction malalignment in the robot-assisted group was significantly lower, considering internal/external rotation and varus/valgus (*p < 0.05; **p < 0.001). Differences considering ante-/recurvature between groups were not significant.

Download figure to PowerPoint

Table 1. Postreduction Malalignment [°]
GroupInternal-/External RotationAnte-/RecurvatureVarus/Valgus
MeanMinMaxMeanMinMaxMeanMinMax
Robot2.9 ± 2.60.116.21.2 ± 1.00.04.81.1 ± 0.90.03.6
Manual8.4 ± 8.50.131.71.9 ± 1.40.05.02.5 ± 2.00.16.3

DISCUSSION

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

This study had several limitations. We chose cadavers embalmed in formalin because of their availability and logistic advantages. Nonembalmed cadavers or living patients might show different characteristics during reduction. However, thus far no study has compared the reduction of differently fixed cadavers or living patients. In our study, the trauma surgeon rated the reduction in the cadavers somewhat easier than reduction of femoral shaft fractures in patients. Only type A and B fractures were presented in this study. Type C fractures, in which no contact between the proximal and distal segments exists, might not profit from 3D visualization. Only one surgeon performed the manual reduction, whereas four different surgeons performed the robotic-assisted reduction. However, we do not expect a more accurate manual fracture reduction, considering data from clinical studies.10, 11, 13

The reduction time in the robot-assisted group was much longer than expected. Although all reductions were successfully performed, the load limit (a recommended nominal load of 60 N and a maximum load of 110 N) of the robot was frequently outrun. When this maximum load was reached, evasive action was necessary to lower the counteracting forces and proceed with reduction; this was the main time-consuming step. Increasing the maximal load capacity would have put the bone, soft-tissues, and the equipment at a higher risk of damage. Measurements of intraoperative forces during fracture reduction showed peak values of >400 N,8 almost four times the load capacity of our robot. Graham et al.,28 using simulation software to represent the complex biological system of bone–muscle interaction of the fractured femur, found forces that rose up to 428 N in a midshaft-fracture, equivalent to in vivo measurements.

The robotic reduction was slower, but more accurate than manual reduction. The above-mentioned study8 did not measure the minimum forces necessary to reduce the fracture, but rather the peak forces that occurred during successful reduction. A principal objective of a robotic application is to obtain low peak forces, almost at the level of the minimal forces during the entire reduction process. Reduced peak forces and precise movements might lower iatrogenic soft-tissue trauma reached by intelligent reduction paths with respect to the anatomical and biomechanical attributes of the soft-tissue envelope. The robot-controlling computer in our setup displayed the forces in a real-time bar diagram, so the surgeon could continuously reevaluate his strategy of reduction and find the ideal path.

The use of a 3D fluoroscope is controversial. We integrated it into the experiment to generate a surface-visualized virtual model, which then could be manipulated intuitively. To enhance the man–machine interface and to obtain the most information from a limited scanning range of ∼12 cm3 at the fracture site, a 3D visualization was implemented.

Robot-assisted fracture reduction is a new topic in orthopedic trauma care. Thus far, only experimental studies by a few groups have been published.23–25, 28–32 Improved precision of reduction is the most discussed objective. Although precision was significantly improved with the robot, malalignment after manual reduction was more than satisfactory. Clinical studies showed higher values of malalignment than those found in our study.7, 14, 16, 33 All of the manual reductions were performed by an experienced trauma surgeon, whereas all robot-assisted fracture reductions were done by surgeons of lesser experience. Experienced surgeons may not profit as much as younger or inexperienced surgeons from robot-assisted reduction. We can also speculate that the younger surgeons are more acquainted with 3D computer animations and telemanipulators and that their acceptance of such new technologies might be higher.

Several critical points need to be addressed prior to practical implementation of robot-assisted fracture reduction. Patient safety remains the first priority. As mentioned above, higher maximum robotic loads increase the risk of damage to bone and soft-tissues. We used a software controlled load cell. In case of software problems the pneumatic safeguard stopped the controller whenever the critical force (i.e., torque) was exceeded.

Ergonomic considerations are important. Placing a device like a serial robot into an OR results in loss of space. Either the system can be setup in a large enough operating theatre or compromises must be made. In particular, keeping sterility is a requirement. The robotic system developed by Graham et al. uses a parallel robotic configuration. The payload to weight (and size) ratio of a parallel system is higher than that of a serial system, which is advantageous considering ergonomics and sterility.25

Financial aspects must also be considered. To avoid increasing time in the operating room, the duration of setup and reduction must be about the same as with conventional procedures. Purchasing and maintenance costs of the robotic system must be weighed against the benefit of increased precision and therefore potentially fewer revision surgeries. Also, the presence of an assistant surgeon might not be necessary when using a robot. Nevertheless, it might be economically unreasonable for hospitals to purchase a robot system exclusively for the reduction of shaft fractures.

Robot-assisted reduction is a new attempt to improve the precision of postreduction alignment. The precision is significantly higher compared to the conventional technique. Before this technique can be applied to patient care, however, further research is necessary, especially on its applicability and safety issues. However, the increased accuracy of fracture reduction makes this new robotic reduction technique an attractive, especially to less experienced surgeons.

Acknowledgements

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Funding was provided by Deutsche-Forschungsgemeinschaft (German Research Foundation). The authors thank BrainLab, who provided an interface to their navigation system. We also thank the anatomy department of the Hannover Medical School for providing specimens.

REFERENCES

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Ganz R, Mast JW, Weber BG, et al. 1991. Clinical aspects of biological plating. Injury 22(Suppl 1): 45.
  • 2
    Gerber C, Mast JW, Ganz R. 1990. Biological internal fixation of fractures. Arch Orthop Trauma Surg 109: 295303.
  • 3
    Winquist RA, Hansen ST, Jr. 1980. Comminuted fractures of the femoral shaft treated by intramedullary nailing. Orthop Clin North Am 11: 633648.
  • 4
    Wolinsky PR, McCarty E, Shyr Y, et al. 1999. Reamed intramedullary nailing of the femur: 551 cases. J Trauma 46: 392399.
  • 5
    Krettek C, Muller M, Miclau T. 2001. Evolution of minimally invasive plate osteosynthesis (MIPO) in the femur. Injury 32(Suppl 3): SC14SC23.
  • 6
    Krettek C, Schandelmaier P, Miclau T, et al. 1997. Transarticular joint reconstruction and indirect plate osteosynthesis for complex distal supracondylar femoral fractures. Injury 28(Suppl 1): A31A41.
  • 7
    Georgiadis GM, Burgar AM. 2001. Percutaneous skeletal joysticks for closed reduction of femoral shaft fractures during intramedullary nailing. J Orthop Trauma 15: 570571.
  • 8
    Gosling T, Westphal R, Faulstich J, et al. 2006. Forces and torques during fracture reduction: intraoperative measurements in the femur. J Orthop Res 24: 333338.
  • 9
    Kruger DM, Kayner DC, Hankin FM, et al. 1990. Traction force profiles associated with the use of a fracture table: a preliminary report. J Orthop Trauma 4: 283286.
  • 10
    Jaarsma RL, Pakvis DF, Verdonschot N, et al. 2004. Rotational malalignment after intramedullary nailing of femoral fractures. J Orthop Trauma 18: 403409.
  • 11
    Prevot N, Charissoux JL, Fiorenza F, et al. 1998. Utilisation d'un clou non fendu de Russel-Taylor pour la stabilisation des fractures du femur. A propos de 57 fractures avec 30 etudes tomodensitometriques de la rotation. Rev Chir Orthop Reparatrice Appar Mot 84: 3340.
  • 12
    Winquist RA, Hansen ST, Jr., Clawson DK. 1984. Closed intramedullary nailing of femoral fractures. A report of five hundred and twenty cases. J Bone Joint Surg Am 66: 529539.
  • 13
    Strecker W, Suger G, Kinzl L. 1996. Local complications of intramedullary nailing. Orthopade 25: 274291.
  • 14
    Baumgaertel F, Dahlen C, Stiletto R, et al. 1994. Technique of using the AO-femoral distractor for femoral intramedullary nailing. J Orthop Trauma 8: 315321.
  • 15
    Hofstetter R, Slomczykowski M, Krettek C, et al. 2000. Computer-assisted fluoroscopy-based reduction of femoral fractures and antetorsion correction. Comput Aided Surg 5: 311325.
  • 16
    Kempf I, Grosse A, Beck G. 1985. Closed locked intramedullary nailing. Its application to comminuted fractures of the femur. J Bone Joint Surg Am 67: 709720.
  • 17
    Kendoff D, Citak M, Gardner MJ, et al. 2007. Navigated femoral nailing using noninvasive registration of the contralateral intact femur to restore anteversion. Technique and clinical use. J Orthop Trauma 21: 725730.
  • 18
    Citak M, Kendoff D, Citak M, et al. 2008. Femoral nail osteosynthesis. Mechanical factors influencing the femoral antetorsion. Unfallchirurg 111: 240246.
  • 19
    Hufner T, Kfuri M, Jr., Kendoff D, et al. 2003. Navigated osteosynthesis of the proximal femur. An experimental study. Unfallchirurg 106: 975979.
  • 20
    Kendoff D, Citak M, Gardner MJ, et al. 2007. Navigated femoral nailing using noninvasive registration of the contralateral intact femur to restore anteversion. Technique and clinical use. J Orthop Trauma 21: 725730.
  • 21
    Khalafi A, Citak M, Kendoff D, et al. 2009. The accuracy and precision of computer assisted surgery in the assessment of frontal plane deviations of the lower extremity: a femoral fracture model. Arch Orthop Trauma Surg 129: 11831187.
  • 22
    Bouazza-Marouf K, Browbank I, Hewit JR. 1995. Robotic-assisted internal fixation of femoral fractures. Proc Inst Mech Eng [H] 209: 5158.
  • 23
    Gosling T, Westphal R, Hufner T, et al. 2005. Robot-assisted fracture reduction: a preliminary study in the femur shaft. Med Biol Eng Comput 43: 115120.
  • 24
    Westphal R, Winkelbach S, Gosling T, et al. 2006. A surgical telemanipulator for femur shaft fracture reduction. Int J Med Robotics Comput Assist Surg 2: 238249.
  • 25
    Graham AE, Xie SQ, Aw KC, et al. 2006. Design of a Parallel Long Bone Fracture Reduction Robot with Planning Treatment Tool. International Conference on Intelligent Robots and Systems: 12551260.
  • 26
    Hufner T, Pohlemann T, Tarte S, et al. 2002. Computer-assisted fracture reduction of pelvic ring fractures: an in vitro study. Clin Orthop Relat Res 399: 231239.
  • 27
    Müller M, Nazarian S, Koch P, et al. 1994. The comprehensive classification of fractures of long bones. Berlin, Heidelberg, New York: Springer; 220 p.
  • 28
    Graham AE, Xie SQ, Aw KC, et al. 2008. Bone-muscle interaction of the fractured femur. J Orthop Res 26: 11591165.
  • 29
    Fuchtmeier B, Egersdoerfer S, Mai R, et al. 2004. Reduction of femoral shaft fractures in vitro by a new developed reduction robot system ‘RepoRobo’. Injury 35(Suppl 1): S-9.
  • 30
    Joung S, Kamon H, Liao H, et al. 2008. A robot assisted hip fracture reduction with a navigation system. Comput Assist Interv 11: 501508.
  • 31
    Maeda Y, Sugano N, Saito M, et al. 2008. Robot-assisted femoral fracture reduction: preliminary study in patients and healthy volunteers. Comput Aided Surg 13: 148156.
  • 32
    Yaniv Z, Joskowicz L. 2005. Precise robot-assisted guide positioning for distal locking of intramedullary nails. IEEE Trans Med Imaging 24: 624635.
  • 33
    Seide K, Wolter D, Kortmann HR. 1999. Fracture reduction and deformity correction with the hexapod Ilizarov fixator. Clin Orthop Relat Res 363: 186195.