Medical Image Segmentation Based On Adaptive Active Contour Model

With the rapid development of medical imaging technology and the advent of computer assisted system,it has become an important tool to assist doctors in diagnosis.The important ring of computer assisted system is image segmentation.The accuracy of image segmentation is related to whether the diagnosis of the lesion area is correct in the later stage.According to the characteristics of medical images,the segmentation of medical images by active contour model has become the focus of current scholars,because the active contour model introduces the level set function,which can process the complex topological structure changes of medical images,and the numerical calculation is relatively simple.This paper makes a detailed study on the processing of medical images by active contour model.The main research contents are as follows:First of all,this paper introduces the basic theory of active contour model and introduces the classical active contour model CV model.Aiming at the shortcomings of CV model in medical image segmentation,such as the inaccuracy of gray image segmentation and the weak noise robustness of medical image,it is improved.Improvement as follows: according to the weight coefficients of CV model is often neglected problems,this paper introduced the image entropy as the adaptive weight CV model,through the calculation of adaptive areas inside and outside the gray information,drive the changing of the fitted curve,to avoid the prior parameters on the influence of the segmentation results,thus improve the uneven gray-level image segmentation accuracy.Then the length term is improved,and the boundary information is added in the length term.The boundary indicating function can attract the curve to approach the boundary and improve the segmentation rate.New fitting terms are added to the model to construct new energy functional and improve the robustness of the model to noise.Finally,the improved model and CV model were used to analyze the segmentation accuracy of medical images.In view of the energy functional single CV model,the complex backgrounds and the multi-objective problem of low efficiency of image segmentation,this paper introduce local energy law in CV model LBF model,and based on this,to join theadaptive function,through the calculation of statistical properties of pixels in the area of global and local change during the curve evolution in the proportion of size,drive curve near the target boundary.Finally,the segmentation results of CV model,LBF model,LGIF model and the improved model in this paper were compared and analyzed through experiments.