The Liver Tumor Segmentation Algorithm Of CT Image Based On Geometric Deformable Model

Liver tumor segmentation of CT image is the foundation and key of liver cancer and other liver disease computer-aided detection and diagnosis technology,has important research significance and application value.Has been in-depth study,and made a lot of research results.Among them,the segmentation method based on deformation model has been widely used.The traditional geometric model is generally applicable to images with relatively high contrast ratios.CT images with intensity in-homogeneity and low contrast characteristics,the segmentation effect is not very good.In view of this problem,based on the traditional geometric deformation model,a new CT images segmentation method of liver tumor is proposed.The specific research work is as follows:(1)Research about related basis theoretical and the proposed algorithm.Carefully studied the characteristics of liver tumor CT image,including the grayscale and geometric properties.Read a lot of literatures about CT image segmentation method,classify them and study these methods carefully.Finally,the CT images of liver tumor segmentation were determined.(2)Image preprocessing method selection.Due to the relationship between the acquisition path,CT image has a certain noise.If you split the original CT image directly,the result will be unsatisfactory.Therefore,according to the de-noising results,the pretreatment method of liver tumor segmentation which is suit for CT image is determined.(3)The determination of geometric deformation model is improved.In order to obtain better segmentation results,this paper first estimate and correct the deviation of the pretreatment of image to improve the quality of image.Then,based on intensity inhomogeneity and low contrast characteristics of CT image,a local intensity clustering attribute is proposed to illustrate the intensity in-homogeneity of image.Around each region of the liver tumor in the CT image,set a local clustering criterion function,as the core of the surrounding tissue of the segmentation area,given a uniform segmentation criterion.An energy function is set up according to this criterion,which is related to the level set function of the surrounding regions and the bias field representing the gray unevenness of the liver tumor in CT image.Finally,by minimizing the energy function,the segmentation of the interesting region in CT image and the estimation and correction of the deviation are achieved.(4)Image processing after segmentation.In order to obtain a better segmentation effect,it is necessary to optimize the segmented image.Aiming at the characteristics of posterior segmental liver tumor,the optimization method of closed operation was selected.(5)Experimental verification.The software experiment platform VS2010 and Matlab R2010 a and auxiliary software are used to verify the algorithm,and the comparison and quantification analysis of the experimental results are carried out to prove the feasibility and effectiveness of the algorithm.The innovation of the research is that(1)the local intensity clustering criterion function is set,which can deal with the situation of local gray scale unevenness better.(2)The energy function of bidirectional geometric deformation model is proposed,and the evolution direction is set in two directions,which shortens the processing time.The shortcomings of the study are that the number of iterations is relatively large for the images with more changes in the boundary,and the segmentation time does not reach the ideal state.At the same time,the sensitivity of the algorithm for contrast,remains to be enhanced.