3D Segmentation Of Breast MRI Based On Inter-frame Correlations

Breast cancer is one of the most common malignant tumors in women.It is a serious threat to women’s physical and mental health and life.Early diagnosis and treatment of breast cancer is helpful to improve the survival rate and quality of life of patients.Magnetic Resonance Imaging(Magnetic Resonance Imaging,MRI)technology is recognized as high sensitivity,which is great helpful to the early diagnosis and treatment of breast cancer.Computer aided diagnosis(Computer-Aided Diagnosis CAD)is a kind of new technology which can help the physician for medical image screening.CAD greatly reduces the workload of physicians,and improves the accuracy and sensitivity of breast cancer diagnosis.Medical image segmentation is a key part of the computer aided diagnosis system.The accuracy of the segmentation results will have a significant impact on the judgment of the area of tumor,the feature classification and the visualization of 3D reconstruction.In this paper,we will solve the problem such as the low accuracy of breast MRI image segmentation,the unsatisfactory results of adaptive segmentation effect,and the poor segmentation on the start frame and end frame.According to the characteristics of breast MRI image sequence,we propose a method of automatic 3D segmentation of breast MRI image sequences based on the correlation between sequence images frames,which uses SLIC0 super pixel and improved C-V level set.In the coarse segmentation part,the algorithm improves the parameters of SLIC,which separates the tumor from the background tissue.In the refined segmentation part,the algorithm combines with the improved C-V level set model,which makes the determine the tumor contour more rapidly and accurately.The algorithm also combine with the related information of 3D frames in the 2D segmentation process,which solves the problem of the poor segmentation result on the start frame and end frame,lying the foundation for the three-dimensional structure of the lesions.The whole segmentation process of this paper does not need manual intervention,and it is suitable for the automatic batch processing of MRI image.In segmentation evaluation section,the proposed method and three competitive methods were applied to 90 cases of breast MRI sequential images.Compared with the contour of manual segmentation,the average overlap rate of the proposed method is 87.84%,and it is 58.9% for C-V model,76.36% for super pixel combined with level set and 83.62% for K-means.It is also demonstrated that the automatic segmentation results of the proposed method have a higher accuracy for the start and the end frame of the breast mass.