全视野数字乳腺x线摄影中微钙化检测方法研究计算机科学与技术专业论文.docxVIP

武汉科技大学硕士学位论文Abstract 武汉科技大学硕士学位论文 Abstract As the top death-causing factor for women，breast cancer IS one of the most common malignancy tumor,which tends to affect young and elderly ladies both physically and mentally．Its prevalence has increased in recent decades，and early diagnosis is the key to reducing the mortality of breast cancer．In this paper,we presented an integrated workflow for microcalcification detection，which consider microcalcification’S characteristics including small and different sizes，variable shapes， minor gray value difference between the suspicious lesion and its surrounding tissue． The proposed method is a comprehensive detection technology,combining image enhancement，level set segmentation，possibilistic fuzzy c-means and weighted support vector machine technology． Detection algorithm used in this paper is divided into two phases：training and testing．The training phase includes microcalcification segmentation based on level set segmentation，feature selection based on mutual information，possibilistic fuzzy c-means clustering of features，calculating sample weights and training model under weighted nonlinear SVM．The testing phase includes microcalcification segmentation， feature extraction，microcalcifications classification and detection of microcalcification clusters．Aimed at the problem of accurately segmenting complete microcalcification，a calcification segmentation method is proposed，which combine edge point search and level set segmentation．Considering that some breast tissue is similar to microcalcification，leading to numerous false positive calcifications using traditional approaches，sample weights are obtained by considering both the probability and typicality values from clustering results of possibilistic fuzzy c-means．And microcalcification classification based on weighted support vector machine is performed．The proposed method Can distinguish microcalcification and suspected microcalcifications more precisely． Microcalcification d