Automatic Liver Tumor Image Segmentation Based On Improved Factorization And B-spline Level Set

Image processing is to extract the region of interest in original image,according to the underlying relative features of initial image.While in the field of medical image segmentation of image segmentation,there are widely attraction from experts and researchers,because it has great practical significance,functional value and clinical needs.This paper is mainly studying the liver tumor segmentation problem based on CT images,accurate segmentation result has very important reference role in preoperative evaluation and surgical planning.How to use the algorithm may not only accurately segment the liver tumor,but also merely need less interaction.Therefore,this paper realizes the improved factorization segmentation algorithm of liver tumor,and proposes an automatic segmentation algorithm of liver tumor combined mixed Gaussian model with B-spline level set.Firstly,researched in the image segmentation methods which are commonly used in the world,and discussing the algorithms for the segmentation of liver tumors in this paper.Secondly,this paper simply introduces the imaging principle of CT images,and get a comprehensive interpretation of medical image format DICOM file,and studying the DICOM display principle and analyzing the idea of BMP file format interchange.Thirdly,this paper analyzes the relevant theories of the level set and compares the different types of level sets.Finally,we compare the classical DRLSE level set method and improved factorization method for liver tumor segmentation,and propose the automatic segmentation algorithm of liver tumors combined with mixed Gaussian model and B-spline level set algorithm.In this paper,we achieve a liver tumor segmentation algorithm with improved factorization,and use local pixel inhomogeneity factor(LPIF)method to perform image preprocessing.Unlike the level set method,this method avoids the operation of the level set in the image segmentation to reinitialize the initial contour,and it is an automatic segmentation method.Compared with DRLSE,this method is better,but because of the edge of the tumor pixel discontinuity and the noise has a certain impact on the image itself,so the segmentation results and the standard segmentation also has a certain gap.In this paper,the algorithm of automatic segmentation of liver tumors combined mixed Gaussian model with B-spline level set is proposed to avoid the operation of initial contours necessary in the traditional level set.The initial contour is automatically determined according to the result of the mixed Gaussian model,and the manual interaction is reduced.The experimental results show that the algorithm has a better effect of liver tumor segmentation,and has certain functional value.The above proposed algorithm is based on a number of theoretical demonstration and experimental proof,the experimental data provided by the hospital for the real case,fully verify the application of the algorithm value and robustness.