A Bone Milling Robot for Spinal Surgery1
Shuai Wang,Peng Gao,Luping Fan,Baoliang Zhao,Peng Zhang,Ying Hu,Bing Li
DOI: https://doi.org/10.1115/1.4033800
2016-01-01
Journal of Medical Devices
Abstract:Spinal surgery is believed to be one of the most high-risk surgeries since any damages to the spinal cord may cause paralysis or even death. Traditionally, the spinal surgery is manually performed. The long duration of surgery will cause the surgeon's fatigue, which will reduce the surgery quality. Some researchers have tried to use robot to assist surgeons to improve the surgery accuracy and efficiency. An Israel company has marketed a parallel robot to help surgeons to guide the tools and implants (Renaissance Guidance System, Mazor Robotics®, Caesarea, Israel) [1]; Ortmaier et al. designed a robot for accurate placement of pedicle screws with the help of an optical navigation system [2]; Chung et al. designed a robot to insert pedicle screws in the spinal fusion procedure [3]; and Hu et al. developed a spinal surgical robot and successfully recognized the different states during the pedicle screw insertion process with a real-time force sensing algorithm [4].In this paper, a four degrees-of-freedom (DOFs) bone milling robot for spinal decompression surgery is introduced. The robot is equipped with a force/torque sensor to feedback the contact force and an infrared optical system for navigation. This paper focuses on the design and kinematic analysis of this robot, and a safety control algorithm is discussed at the end.Figure 1 shows the relative position between the bone drill and the human vertebra in surgery. After discussion with surgeon, 4DOF are required to perform the bone milling (Fig. 2). The first DOF is along the spine, to allow the milling tip moving along the superior–inferior direction on certain vertebra; the second and third DOFs are used to control the position of the drill on the transverse plane; the abovementioned 3DOFs will let the robot have a 3D workspace. To make sure the bone mill always keeps a vertical posture to the bone vertebral lamina, a fourth DOF is required to adjust the orientation of the drill in the transverse plane.The 4DOFs construct a prismatic–revolute–revolute–revolute serial structure, which are shown in Fig. 2.With the modified D–H (Denavit–Hartenberg) method [5], the link frames of the robot are assigned, as shown in Fig. 3. Table 1 shows the D–H parameters for each joint (β represents the angle between two neighboring links). The forward kinematics is presented as the homogeneous transform matrix(1)T50=T10T21T32T43T54=[r11r12r13pxr21r22r23pyr31r32r33pz0001]in which(2)r11=−c1+2+3+4(3)r12=s1+2+3+4(4)r21=−s1+2+3+4(5)r22=c1+2+3+4(6)r31=r32=r13=r23=0(7)r33=1(8)px=L1c1−L2c1+2+L3c1+2+3−L4c1+2+3+4(9)py=L1s1+L2s1+2+L3s1+2+3+L4s1+2+3+4(10)pz=d1+L5Based on the forward kinematics, the workspace of the robot is obtained (with L1 = 60 mm, L2 = 120 mm, L3 = 100 mm, L4 = 202 mm, L5 = 0 mm, and β1 = 2π/3), as shown in Fig. 4. To control the robot to move along a desired trajectory, the inverse kinematics is needed. Assume that the end-effector posture is known as(11)Ttarget=[r11r12r13pxr21r22r23pyr31r32r33pz0001]the position of each joint can be calculated as(12)d1=pz−L5(13)β2=π−arcsin[L3s3(A−L1c1)+(L2−L3c3)(B−L1s1)(L3s3)2−(L2−L3c3)2]−β1(14)β3=arccos[L22+L32−(A−L1c1)2−(B−L1s1)22L2L3](15)β4=2π+arccos[A−pxL4]−β1−β2−β3in which(16)A=px+L4r21(17)B=py+L4r11Safety issue is always the most important concern for the robot-assisted surgery. Figure 5 shows the whole system components for the bone milling surgical system. The C-arm will first scan the patient to get the CT images; then, the 3D model of the vertebra is reconstructed. The surgeon makes surgical plan based on the 3D model and controls the robot to carry out the bone milling. Both the patient and the bone mill have infrared optical markers attached to allow their postures to be monitored in real-time, the movement of the surgical site due to patient's breathing can also be compensated by the robot system. The bone mill is also equipped with a force/torque sensor to feedback the tool–tissue interaction force, which can be further used to monitor the bone milling state [4]. With all these techniques combined, we believe that the safety of the surgery can be improved.This work was supported by the National Natural Science Foundation of China (Nos. 61573336 and 61210013) and Research and Development Project of Guangdong Key Laboratory of Robotics and Intelligent Systems (No. ZDSYS20140509174140672).