Prostate MR/TRUS Images Registration And Design Of Robot Biopsy

Prostate cancer,the most common cancer in the world,is diagnosed with many of biopsies in hospital.As the one of widely recognized ways,prostate biopsy guided with Trans-Rectal Ultrasound(TRUS)has been popular means to diagnose prostate cancer due to the advantages such as real-time,low cost,convenient operation,etc.However,it is fairly hard to realize the location of tumor because of the low image quality of prostate TRUS,which may result missed diagnosis,misdiagnosis or even redundant biopsies.Recent years,the fusion diagnosis technology of magnetic resonance and trans-rectal ultrasound is deeply developed to achieve the targeted biopsy of prostate tumor.However,there is image deformation between prostate MR and TRUS images when registered,which requires to perform rigid or non-rigid registration methods but most are based on manual registration which is inefficient or low accuracy.This paper studies the automatic registration method of prostate MR/TRUS images,mainly describes the automatic segmentation method of prostate TRUS images under the supervised learning framework and the non-rigid registration of MR/TRUS images and discusses the simulator application of robot technology in prostate needle biopsy based on ROS platform.With the parameter initial problem in active appearance model,the novel boundary-driven model based on image gradient is proposed and the supervised learning model is applied to predict the prostate region.The math optimization model is solved with genetic algorithm and then the parameters computed are used in contour segmentation.In next registration process,the shape information descriptor is built to construct the matching features of contour points in both MR and TRUS images.These non-matching points will be matched after the assign problem solved with improved KM algorithm.To improve the registration accuracy of prostate MR/TRUS images,regularity factor is used to restrict the location error of thin plate spline.The experimental result proves the better registration performance.Finally,this paper discusses the robot assistant biopsy under prostate TRUS and the simulator platform based on ROS system is designed after the significant analysis of crucial structures.In this simulator system,MoveIt is used to configure the controller of robotic arm and the robot subscribes the pose topic published by Marker node.After waiting the message some time,robot will carry out autonomous path planning and obstacle avoidance,and needle path will be planned in Cartesian coordinate space to achieve the purpose of prostate biopsy.