Design Of Robotic Positioning And Tracking System For Transcranial Magnetic Stimulation

Transcranial Magnetic Stimulation(TMS)is a non-invasive,painless green neuromodulation method.Due to its unique function,low cost of use and simple operation,the technology has been rapidly extended to various fields of clinical and scientific research.However,in practical applications,it is difficult for TMS to accurately determine the stimulation site,and the stimulation coil is in a fixed state,which cannot compensate for the movement of the patient’s head,all of this greatly reduces the stimulation effect.Based on the limitations of TMS,this article designed the positioning and tracking system of TMS robot.In response to the demand of TMS robotic system,this article has determined the guiding ideology of positioning and tracking system to improve the stimulation accuracy and reduce the pain caused by treatment of patients.The main research contents include TMS robotic system overall scheme design,head pose detection and tracking,the pose adjustment control of end-effector based on force sensor and TMS robot obstacle avoidance strategy research.The coordinated working process of each component of the positioning and tracking system is to detect the posture of the patient’s head in real time under the guidance of vision,and use the six-degree-of-freedom robot arm to calibrate the TMS coil and compensate the movement of the patient’s head.Meanwhile,according to the measurement of the force sensor adjusts the pose of the end-effector in real time,so that the coil is closer to the head and improves the safety of the system.In order to make the TMS robot operation process easier and reduce the cost of use,this article applies the 3D pose estimation technology of marker to the TMS robot system for the first time,and uses the marker and monocular camera to positioning the patient’s head in real time.During the treatment,the patient completes the positioning task by wearing the AR glasses attached with marker,and feels the AR virtual scene to further relax himself.In order to improvement the safety of the system,the end-effector’s pose of the TMS robot is adjusted in real time.Considering the inconsistency of the zero drift of the force sensor in different attitudes and the influence of kinematic modeling error on gravity compensation accuracy,a gravity compensation method based on BP neural network is proposed.The adaptive controller is used to make a decision on the pose correction of the robot arm so that the end of the arm can follow the movement of the head and prevent the head from hitting the coil.Due to the uncertainty of the treatment environment,this thesis uses deep reinforcement learning to determine themovement direction of the TMS robot,builds and trains the obstacle avoidance network model,and uses the model to guide the robot arm to avoid the obstacle and reach the target position through the optimal path.Finally,the positioning and tracking system designed in this article is used as the experimental platform to verify the feasibility of the function of patient’s head positioning and compensation,and carried out migration experiment of the obstacle avoidance model.In order to further facilitate the operation of medical staff,this article also designed a human-computer interaction interface for the positioning and tracking system under the QT platform.