Medical Image Segmentation Research And Application Of In Leg Musculature Segmentation

Medical image segmentation is an important part in the computer-aided treatment and diagnosis, and it is also an important area in medical image processing and analysis. This thesis tries to study and explore the musculature’s segmentation. Image segmentation is crucial for studies ranging from image analysis to image processing. With continuous improvement of the medical image technology, the quality, resolution and contained details of medical image have been improved and the traditional segmentation algorithm requires much more time, which facilitates the reduction of generic segmentation algorithm. How to make an efficient and correct segmentation has become an important and advanced academic area.This thesis designs a method for muscle tissue segmentation after studing the segmentation of the medical image and analyzing the image of leg sections. And for the segmentation of muscle tissues, this thesis will focus on the following aspects:1. The research based on the level set segmentation algorithm. The thesis gives a brief description of the Li level set model and the fundamental rules of C-V level model, and then respectively studies the advantages and disadvantages of the two models. And it offers an improved method after analyzing the characteristics of the leg image to be split. 2. Clustering research based on EM expectation maximization algorithm. This thesis makes an introduction of EM algorithm and the concept of GMM. By introducing the Mean Shift algorithm to initialize the parameters of EM, an estimation of the Gaussian mixture model parameter will be made. And the method of segmentation is applied to the clarification of the leg section image, and the result of the research is analyzed.3. Study the muscle tissues segmentation. It deigns a method to analyze the muscle tissues in the image of leg section. The steps are as follows:1. Bilateral filtering denoising,2. Clustering based on EM algorithm and Gaussian mixture model, which uses the method of Mean Shift to initialize the EM parameters.3. Get the muscle tissues which we are interested in through accurate image segmentation by level set.