The Level Set Method Of Feature Extraction And Application In Medical Image Diagnosis

In recent years, level set method is a hot research in image segmentation domain. This method has been used in many fields and achieved good results, such as image processing, computer graphic, fluid dynamics and so on. An intensive study of its key technologies and theories is made in this paper and the traditional C-V model is improved to effectively diagnose medical images.Firstly, a deep research of active contour model and level set method is made and an improved C-V model is proposed. The initial contour of the model is quickly moved near the target border, greatly reducing the evolution time; an adaptive velocity reconciling item is added for velocity equation to make the model converge to the true border.Secondly, the research achievement of level set method is applied to benign and malignant liver tumor diagnosis. Based on the careful study of the Naive Bayesian classification methods, the improved C-V model is used to segment liver tumors in CT images and the classification is trained and tested with the extracted contour features. The experiment result is efficient.Finally, the architecture of Computer-Aided Diagnosis system of liver tumors in CT images is proposed. In matlab7.5 environment, the related system module and the whole system are developed. Experimental results show that the system can effectively segment and classify benign and malignant tumors of liver in CT images.