Research On Ultrasound-guided Prostate Percutaneous Intervention Robot System

The increasing incidence of prostate disease poses a huge threat to men’s health.Transrectal ultrasound image-guided prostate percutaneous intervention plays an important role in the diagnosis and treatment of prostate diseases.Faced with drawback of low accuracy during manual insertion,this paper designs a transrectal ultrasound image-guided prostate percutaneous intervention robot to improve the operation accuracy,stability and success rate.This paper also studies the kinematic modeling and parameter identification method,ultrasound image segmentation method based on preknowledge and probe calibration method,and bevel-tip needle-tissue interaction force model.The robot systems now for prostate insertion have huge volumn,low stiffness or insufficient joints.This paper proposes mechanism scheme based on realitiy requirments,and designs an eight-degree-of-freedom prostate intervention surgery robot system.The robot system mainly includes three parts: the two-degree-of-freedom transrectal ultrasound probe manipulator,the four-degree-of-freedom parallel mechanism for needle position and posture adjustment,and the two-degree-of-freedom needle manipulator.Based on the structural design,the system workspace and accuracy are analyzed,which lay the foundation for the robot control.Precise kinematic parameters are the basis for accurate robot control.In this paper,kinematics modeling and parameter calibration and identification are accomplished for robot system.To solve the kinematics modeling and parameter identification problem of the specific parallel structure,this paper builds up the parallel-structure kinematics model firstly,and modifies the model with non-linear error.An improved particle swarm algorithm based on parameter discernibility is designed to accomplish the kinematic parameter identification.By designing and carrying out calibration experiments and accuracy provement experiments,the accuracy and efficiency of modified kinematics model are proved,where one-stage error is less than 1 mm,and target position error is less than 2 mm.The automatic segmentation in the ultrasound image helps to identify the location of the insertion needle during the percutaneous intervention process and establish the mapping relationship between the image coordinate and the Cartesian coordinate.The ultrasound images erosed with speckle noises are difficult to be segmented automatically.This paper firstly proposes an anisotropic diffusion algorithm combined with Jdivergence for ultrasound image preprocessing,which removes noise and enhances the boundary.Then the object shape to be segmented in ultrasound image is described as normalized Fourier descriptors,and the descriptor are used as prior knowledge to be combined with the proposed improved Snake segmentation algorithm,to complete the feature segmentation in ultrasound images.In order to verify the effectiveness of the algorithm,this paper carries out experiments on silica gel models and in vitro animal tissues,and proposes a calibration method for ultrasound probes based on the segmentation method.During insertion process,the bevel-tip needle is deflected by the unbalanced force.The theoretical modeling in other studies leads to low fitness with real condition.This paper proposes a needle-tissue interaction model based on mechanical modelling method,and adopts the particle swarm optimization algorithm to reverse the distribution of support force according to the deformation of the insertion needle after entering the tissue.Then,a silicone phantom experimental platform based on one-dimensional force sensor is built up.By carrying out hard needle and soft needle insertion and friction experiments,the parameters in the model are identified,and the rationality and effectiveness of the force model are verified.Based on the established force model and puncture needle trajectory prediction,this paper proposes a strategy based on force model to improve insertion accuracy by controlling needle deformation.This paper accomplishs the design and production of the robot system prototype,builds up the experiment platform,and carries out phantom insertion experiment.Firstly,the calibration experiment of the ultrasonic probe based on the ultrasonic image segmentation method is carried out,and the mapping parameters of the bi-plane transrectal ultrasound probe are obtained;the calibration error is proved to be less than 1mm.Then based on the robot-assisted prostate insertion process flow chart,the whole robot system targeting experiment is carried out,and the results show that the proposed insertion strategy based on force model can reduce the targeting error from 2.73±1.24 mm to 0.91±0.51 mm.