Research On The Navigation Method Of Assisted Robotic Arm For Epilepsy Surgery Based On CT Image

There are about 6.5 ～ 9.1 million epileptic patients in China.Patients in severe cases often need to detect and treat epileptic lesions by implanting intracranial electrodes.More than ten to dozens of electrodes were needed in each operation.It has a high requirement for speed and precision during the operation.The traditional electrode implantation surgery uses the head-mounted navigation frame to locate the implantation position.However,this method has some problems: it’s hard to registrate and locate,timeconsuming,and low accuracy.These factors limit the productivity of doctors and hospitals;and increase the possibility of surgery accidents.In order to solve the above problems,a new surgery-assisted navigation method is designed,a prototype is developed,and the navigation results and accuracy are verified by experiments.In order to meet the requirments of high-precision registration and intraoperative visualization,a 3D reconstruction method based on medical images is designed in this thesis.The medical image with noise is preprocessed by anisotropic diffusion filtering algorithm,and the medical image is reconstructed by Marching Cubes algorithm and Marching Tetrahedron algorithm respectively,and the reconstruction results are compared.The algorithm with better reconstruction quality is selected for reconstruction,and the Laplace algorithm and quadratic error measurement method are used to optimize the reconstructed 3D model to get a noise-free and smooth 3D model.Aiming at the realization of the registration method of the manipulator,this thesis designs a two-step registration method which uses the camera for rough positioning,and then the doctor guides the robotic arm to position the feature points to achieve highprecision registration.In the process of rough positioning,ARTool KIT is used to detect the corner points of QR code,and the position of the center of the two-dimensional code is calculated by the three-dimensional attitude estimation algorithm,which makes the manipulator reach the working area.In the part of high-precision registration,this thesis first solves the forward and inverse kinematics of the manipulator,designs the selfcollision detection algorithm suitable for the UR manipulator,and analyzes the optimal workspace of the manipulator.The gravity compensation method of the end tool of the manipulator is designed,and the mapping relationship between the force sensor and the joint velocity is solved according to the inversion of the Jacobian matrix,so as to realize the function of manually pulling the manipulator to move and calibrate the characteristic point cloud.The characteristic point cloud is collected in the three-dimensional model,and the quaternion method is used to register the point cloud,which realizes the registration of the manipulator before operation.In order to solve the problem of collision with environment during manipulator operation,a trajectory planning method based on sampling is proposed.According to the simplicity of obstacle avoidance environment in the surgical scene,this thesis proposes a path planning method based on random search tree and random search graph.On this basis,the random sampling algorithm is improved,the function of fast random expansion to the target node and the algorithm of optimizing sampling points are added,and the superiority of the algorithm is verified by simulation environment.The hardware platform and algorithm platform of the surgery-assisted manipulator navigation system are designed and built,the feasibility of the technology is proved by registration and positioning experiments,and the corresponding host computer software system is designed to ensure that the operation can be carried out smoothly.