基于MRI图像纹理特征的膀胱肿瘤浸润深度检测  

Detection of the Invasion Depth of Bladder Tumor Based on Textural Features from MRI Images

作　　者：吴智德[1] 史正星[1] 张国鹏[1] 卢虹冰[1] 

机构地区：[1]第四军医大学,西安710032

出　　处：《透析与人工器官》2010年第3期20-24,共5页Chinese Journal of Dialysis and Artificial Organs

摘　　要：肿瘤浸润深度对于判断膀胱肿瘤分期及随后治疗有重要意义。本研究基于膀胱MRI图像,利用尿液和膀胱壁对比强的特点,结合经统计分析表明在肿瘤及膀胱壁组织有统计差异的纹理特征,通过提取肿瘤及周围膀胱壁组织的均值、均匀度、标准差、粗糙度等纹理特征,结合分类器识别,实现了对膀胱肿瘤浸润膀胱壁平滑肌深度的标记和边界划分。临床患者MRI图像分析结果表明该方法能够有效标记肿瘤浸润深度,为膀胱肿瘤的无创检测提供了新的手段。The invasion depth of bladder tumor is of great importance for tumor staging and treatment planning.Since MRI bladder images could provide natural contrast between urine and bladder wall,some texture features were extracted from MRI images previously,which demonstrated statistically significant difference between tumor tissue and wall tissue.In this study,based on these features,such as mean,standard deviation,uniformity,covariance and contrast,a classification and labeling scheme were proposed for the detection of invasion depth of bladder tumor.Experimental results based on patients＇ MRI datasets show the feasibility of the proposed scheme for the labeling of bladder tumor,which has the potential for noninvasive detection of bladder tumor and its stage.

关 键 词：计算机应用技术 膀胱肿瘤 MRI成像S VM 

分 类 号：R737.14[医药卫生—肿瘤]