Research And Application Of Bilevel Set Model In Medical Image Segmentation

Medical image becomes more complex because of imaging equipment and external interference.Medical image plays an important role in medical diagnosis.Effective image is helpful to make a reasonable medical plan.Although new theories and methods have been used in image segmentation,compared with other images,medical images are more complex,so how to get more efficient and accurate segmentation results still needs a lot of research.The level set algorithm can be fitted with many algorithms,and has a good segmentation effect for medical images with noise,uneven gray level and multi-objective.The main research contents and achievements are as follows:1.Firstly,the mathematical knowledge involved in the level set algorithm is introduced,including the solution of partial differential equations,gradient downflow and so on.Then it introduces the theory of curve evolution and the method of solving the level set equation.Finally,it introduces several different level set segmentation models.2.In order to solve this problem,an improved fast segmentation algorithm is proposed.Firstly,wavelet transform is used to denoise,and then median filter is used to denoise the image,which can not only remove the noise well,but also retain the useful information in the image.Then add acceleration factor to level set,which can effectively solve the problem of slow segmentation speed.Level set function needs to be initialized constantly.In order to solve this problem,this paper adds energy penalty term.Experiments show that the new model can meet the segmentation requirements.3.In view of the problem of uneven gray level and complex boundary in medical image,this paper adds gradient template in 45 degree and 135 degree directions on the basis of Canny operator to improve the detection ability of complex boundary image,adds fuzzy membership function in offset field model to improve the accuracy of segmentation,and finally uses hydrazine potential function in distance regularization to improve the image Segmentation efficiency.The experimental results show the effectiveness of the algorithm.4.In view of the phenomenon of high noise,multi-target and irregular focus area in medical image,this paper reduces the noise in image by improved bilateral filtering,and improves the segmentation accuracy of irregular focus area image by adding spatial information in fuzzy clustering.Finally,by adding energy penalty term and edge indicator function in the two-level concentration,it improves the segmentation accuracy of irregular focus area image For multi-objective medical image segmentation.The experimental results also show that the algorithm in this chapter can achieve an ideal segmentation effect.