Research On Key Technologies Of MRI-US Image Fusion Navigation For Prostate Intervention

In recent years,the incidence of prostate cancer has been increasing.For the diagnosis and treatment of prostate cancer,the current mainstream method is still the prostate intervention guided by medical imaging.As for medical imaging,ultrasound(US)and magnetic resonance imaging(MRI)are the two most commonly used medical imaging methods in prostate intervention,which have their own advantages and disadvantages.Therefore,researchers at home and abroad have proposed an image guidance scheme that combines preoperative MRI images with intraoperative US images.In this dissertation,the key technologies of segmentation,registration and fusion of MR and US images in prostate intervention and the methods of robots to complete surgery with electromagnetic positioning equipment are studied and discussed.Firstly,the requirements of the prostate interventional therapy system are analyzed,and the basic structure and working process of the system are determined.By comparing the characteristics of the popular medical image processing platforms at home and abroad,3D Slicer is chosen as the medical image processing platform of the system from the aspects of support for external systems,openness and scalability,and on the basis of this platform,module expansion is carried out to form the navigation software of this system.A PRR configuration robot body structure suitable for prostate intervention is designed,and the corresponding robot control system is built.Secondly,the contour of prostate is extracted from US and MRI images to prepare for image registration.In view of the lack of reliable regional homogeneity and texture in prostate MRI images and the lack of boundary segments in end slices,an adaptive priori shape model based segmentation method is proposed,which can obtain accurate and reliable prostatic MRI image contours using adaptive priori model.Considering the low signal-to-noise ratio,large speckle noise interference,serious uneven gray distribution and blurred edges of prostate US image,a method of US image segmentation based on deep learning convolution neural network is proposed.The advantage of this method is that only a few training samples are needed to obtain better feature extraction performance.Thirdly,one of the key steps of fusion navigation of US and MRI images is to register the two images in space.A non-rigid image registration method based on improved Active Demons is proposed to solve the problem of large deformation of prostate in preoperative MRI and intraoperative US images.The average error and root mean square error of this method are better than those of general Active Demons method,which can meet the higher accuracy requirements of prostate interventional surgery.In order to verify the effectiveness of this method,the prostatic image data of 6 patients are used to evaluate it.Finally,based on the methods and results presented in the previous chapters,a series of schemes and strategies to solve the problem of multi-sensor fusion of the system are designed.In order to determine the lesion area of prostate and enhance the image by using complementary information of different modal images,pyramid decomposition method and direct fusion method are combined and Laplace pyramid and pixel weighting method are used to fuse MRI and US data.To achieve the fusion of pose data of surgical instruments and image data,an improved N-line model method is used to calibrate the ultrasonic probe,and then the combination of the ultrasonic system and the electromagnetic positioning system is realized.A method of error compensation based on Bernstein polynomial is proposed by researching the positioning error of surgical needle.Based on the registration model of surgical needle and US image,the concept of virtual needle insertion path is introduced.In order to verify the effectiveness and feasibility of the system,the fusion experiments of MRI and US image data,the integration experiments of ultrasound equipment and navigation system,and the puncture experiments of prostate interventional robot guided by fusion image are carried out respectively.