Research On Medical Image Segmentation Algorithm And Design Of Segmentation System

With the development of modern computer technology and digital medical equipment,digital medical image processing has gained high attention.Digital medical image is the main basis in disease diagnosis for clinicians and experts,medical image segmentation is one of the key steps in the analysis of medical image processing,and segmentation of the liver in abdominal CT images plays an important role in CAD diagnosis of liver cancer.However,image segmentation itself is a pathological problem.Ideal segmentation result always requires much prior knowledge,while it affects the result of image analysis and image understanding as well.It is difficult to find a universal and ideal algorithm to match every image well.Abdominal CT image,with its complicated structure,is always a bottleneck of medical image segmentation.Therefore,research on image segmentation with its valuable significance is always a hot research point in the field of image processing.Based on the traditional image segmentation method,this article focuses on random field model image segmentation algorithm,and the main contents of this article are listed as followed:(1)We researched the image segmentation method based on K-means clustering,and proposed a SA-PSO based improved K-means clustering algorithm,which acts as an initial segmentation algorithm.(2)According to the characteristics of abdominal CT images,a rib fitting based image preprocessing algorithm was proposed,and it can effectively remove other organs adhered to liver in the image.(3)The image segmentation algorithm based on Markov random field model was researched.Liver segmentation in abdominal CT images was implemented respectively based on the single-scale and multi-scale Markov random field model.According to the multi-resolution analysis theory,an image lifting wavelet decomposition method was used to build a multi-scale representation model of abdominal CT images.A multi-scale Markov random field model was established for abdominal CT image segmentation,and an abdominal CT image segmentation was implemented based on MRF-ICM method.A variable weight multi-scale MRF model was proposed based on the improvement of the feature field and label field model.(4)Combining with the characteristics of the Snake model,the MRF model based image segmentation result was acted as an initial outline of the Snake model,and a Snake model based secondary fine segmentation was conducted to obtain a more accurate segmentation result.(5)An image segmentation system was designed and realized by the mix program of the MATLAB R2010a and Visual Studio 2008.The proposed algorithm was tested to verify the effectiveness.