Joint Segmentation And Registration Method For Medical Images

In recent years,with the vigorous development of computer technology,image processing technology has been widely used in the work of various industries.Especially in the medical industry,due to the rapid development of medical imaging technology,the research and application of medical image processing technology has been greatly promoted.In the field of medical image processing,segmentation and registration of medical images are the prerequisites for subsequent image fusion,reconstruction,and target recognition operations.It is of great significance for clinical diagnosis and surgical positioning.Therefore,in order to improve the accuracy and effectiveness of these two technologies,scholars at home and abroad have launched a series of in-depth studies and proposed many new methods.But most of the studies are independent segmentation methods and independent registration methods.The requirements for image processing are becoming more and more stringent,and the limitations of independent segmentation and registration methods are becoming increasingly prominent.This thesis begins with a discussion of B-spline registration methods and level set-based segmentation methods.Among the methods of image registration,B-spline transformation is the most widely used non-rigid registration method at home and abroad because of its good local transformation ability.Although the B-spline registration method can effectively process images with large deformations among each other,for large and complex image deformations,the problem of local extreme values may become highly unconstrained,and the implementation of the optimal transformation requires considerable the amount of calculation.The most widely used image segmentation method is the active contour model segmentation method based on level set.Although the level set segmentation method can segment complex structures in an image,if there is noisy data in the image,the segmentation result is always poor.Considering the limitations of independent segmentation and registration methods,this thesis studies a joint segmentation and registration method that combines these two methods.Aiming at the shortcomings of B-spline non-rigid registration and level set segmentation,this thesis proposes a medical image joint segmentation and registration method based on locally updated hierarchical B-spline bidirectional transformation and level set method.The level set outline is segmented from the image to be registered first,and then it is deformed to match the reference image using the locally updated hierarchical B-spline transformation model,so as to achieve the segmentation of the reference image;the locally updated hierarchical Bspline transformation is used At the same time of registration,the feature in the image is accurately deformed by using the segmented target boundary information,and a two-way transformation is introduced in the registration to improve the accuracy and smoothness of the registration.The energy function of the level set method is combined with the energy function of the hierarchical B-spline bidirectional transformation to construct a joint energy function for registration and segmentation,and a gradient descent method is used to minimize the energy function to achieve segmentation and registration.The experimental results show that compared with the image registration method alone,the mean square error(MSE)of the method in this paper is reduced by 30%.Compared with the independent segmentation method,the Dice metric(DS)of this method is higher than that of the independent segmentation method.The joint method in this paper can effectively improve the accuracy of image registration,and it has high robustness when segmenting noisy images.