| In recent years,the number of orthopaedic surgeries is increasing year by year with the aging of China’s population and change of people’ s traffic mode.In particular,lumbar disease has become the clinical main diseases.Robot-assisted spine surgery is the new surgery method that combines the robot technology,image processing technology,and spine minimally invasive surgery technology.This new method tends to high precision,more reliable and better clinical promotion.In addition,the dependence of traditional minimally invasive surgery on surgeon ’ s experience could be alleviated and the learning curve for surgeon in conputer-assisted surgery could be shortened.However,the application of robot system in clinic is limited due to the two reasons.One is that commercial robot systems only focus on positioning of pedicle screw path while the delicate and burdensome process of pedicle screw path drilling is performed manually by surgeons.The other is that the expensive 3D c-arm is the main equipment in navigation system.Therefore,the main objective of this paper is to promote the clinical application of robot system in the condition of meeting demands of lumbar surgery.With the support of chinese national high technology research and development program(863),robot system for spine surgery is designed and integrated,robot navigation positioning based on 2D fluoroscopy image,and robot following and operation control by force are realized in this paper.Based on clinical requirement of spine minimally invasive surgery,the spine surgery robot system is designed and integrated.The configuration and structure of hybrid manipulator are designed by dof(degree of freedom),precision and compact structure.The kinematic model of hybrid manipulator is established and the validity of model is validated by the simulation.In addition,the key link parameters of manipulator are optimized based on kinematic model and requirement of robot absolute workspace.Based on content above,the composition of spine surgery robot system is determined and the control platform and software platform are established.Finally,the hybrid manipulator which could realize positioning of pedicle screw path and drilling of pedicle screw path is developed.The motion precision of manipulator is validated by laser tracker and could meet demand of spine surgery.Based on the problem of control point extraction complexity influenced by bone tissue in 2D fluoroscopy navigation,control point automatic extraction technology is proposed and robot navigation positioning is realized.The bone structure of patient and ideal position of control point are restored accurately by fluoroscopy image distortion correction using thin-plate spline.Then,the control point extraction technology is realized by SVM(Support Vector Machine),IHVCD(Improved Horizontal and Vertical Search for Circle Detection)and clustering.Based on control point position,2D-2D registration and optical tracker,the closed-loop positioning of robot is realized.The precision of closed-loop positioning is higher than that of open-loop positioning by experiment.Based on the intervention problem between patient and robot during robot positioning process,and problems of k-wire bending deformation and bone overheating caused by constant velocity drilling,the on-line parameter identification model is established to realize the robot following motion decided by force intention of operator,and drilling force control operation of pedicle screw path is realized by two-layer adaptive fuzzy controller and strategy of state recognition.The principle of robot following motion is analyzed based on admittance model,and the influence of parameters in admittance model on robot following motion is obtained.The influence of end-tool weight on force implemented by operator is removed by weight compensation model.On-line admittance parameter identification model using DDPG(Deep Deterministic Policy Gradient)is proposed to realize unknown environment robot following motion.The convergence performance superiority of DDPG is validated based on simulations involving DDPG,SARSA and F uzzy SARSA.In addition,drilling force control using two-layer adaptive fuzzy controller could be used to solve nonlinear time varying problems during cortical bone operation.Constant velocity control is applied in realizing drilling operation for cancellous bone.Finally,the safety of drilling bone using robot system could be guaranteed by state recognition strategy involving force signal and ae(acoustic emission)signal.To verify the validity of robot navigation positioning technology by 2D fluoroscopy image,the control point extraction experiments containing spine model bone,swine bone and swine are implemented.Experimental results show that control point extraction algorithm proposed in this paper has obvious advantage in strong background interference compared with EDCircles.The experiments containing planning precision and robot system positioning based on 2D fluoroscopy image are carried out.Experimental results are evaluated by Gertzbein-Robbins,and meet clinical safety requirement.To test the convergence performance of DDPG in robot following motion,the drag experiment is implemented for different operators.Experimental results show the advantages of convergence velocity and stability.Finally,to verify the clinical effect of robot system,the experiment of robot-assisted cadaver is implemented.Experimental result is evaluated by 2D c-arm and CT,and meets surgery requirement by clinical evaluation index. |
PDF Full Text Request