The Research On Registration And Segmentation Algorithm Of Multi-mode Breast MRI Image

Breast tumors has become the first killer of women health.The technique of medical image processing and analysis plays a critical role in the diagnosis and treatment of breast cancer,optimization of treatment plan and evaluation of treatment effect.As the two key techniques of medical image processing and analysis,medical image registration and segmentation,which are widely applied in clinical diagnosis,surgical navigation,anatomy teaching and other aspects.Due to the pathological diversity and complexity of breast tumors,the results of medical image segmentation and registration can not fully reflect the pathological information of breast tumors,which seriously affects the diagnosis of breast tumors by doctors.Consequently,the existing segmentation and registration algorithms need to be further optimized.The specific work of this paper is as follows:(1)The background and significance of multi-mode breast MRI tumor segmentation and registration are described in detail,besides representing the research status at home and abroad.The concept,classification,method and evaluation index of medical image segmentation and registration are represented.(2)In the registration of multi-mode breast MRI tumor images,there are some problems in the SURF algorithm,such as less feature points and low accuracy of image registration.In the SURF algorithm,Harris process is used to extract feature points.In addition,the feature descriptors generated by 64 dimensional vectors with the circular area.Euclidean distance matching is applied to enhance the extraction of feature points.The PSO algorithm is used to find the global optimal solution,the mutual information is used as the fitness function to enhance the global optimal solution of the objective function,and the average maximum is applied to prevent the algorithm from falling into premature phenomenon.The simulation results indicated that,compared with SIFT algorithm and SURF algorithm,this algorithm can effectively strengthen the extraction of feature points and the accuracy of image registration.(3)After the multi-mode breast MRI tumor image registration,there are some problems that lower matching degree of weak boundary of registration image and location deviation of breast tumor block.The improved edge stop function and energy penalty term are integrated into the classic CV model to speed up the curve evolution.Compared with the traditional CV model,HLCV model,GLDS model and GLCV model,the simulation results indicate,this algorithm can quickly segment the whole multi-mode breast MRI tumor area.(4)Aiming at the problems of overlapping in breast tumor edge image segmentation,the classical CV model segmentation results in a large number of redundant contours,so that the multi-mode breast MRI tumor image can not segment the edge region,fat region and focusregion.In this paper,the multi-phase level set segmentation algorithm(MLS)is used to segment the image intonnon overlapping and independent regions,and the multi-mode breast MRI tumor image is divided into three different regions.A narrow band region is constructed by NBLS,which is an improved algorithm of level set algorithm.In the process of curve evolution,the narrow-band region is used for iterative calculation.In this paper,NBLS is integrated into MLS model to reduce the complexity of MLS algorithm,improving the efficiency of calculation.Simulation results show that,compared with MLS model,the algorithm in this paper has higher accuracy,and the segmentation results are better,fast and effective to segment different regions of breast tumor.