结合聚类和改进的C-V演化方程在医学图像分割中的应用  

Application of clustering and improved C-V evolution equation in medical image segmentation

作　　者：罗志宏[1] 冯国灿[2] 成秋生[3] 

机构地区：[1]中山大学计算机科学系,广州510275 [2]中山大学数学与计算科学学院,广州510275 [3]广州市第一人民医院脑内科,广州510180

出　　处：《计算机应用》2008年第9期2288-2291,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(60675007);教育部重点基金资助项目(104145)

摘　　要：在传统的C-V模型的基础上提出一类改进的C-V演化方程,利用它们与聚类技术相结合对病理状态脑部CT和MR图像进行分割,能获得较理想的实验结果。特别是利用提出的演化方程对一个合成的几何图像进行分割,结果显示能比一些传统的演化方程具有更好的稳健性和准确性。A class of improved C-V evolution equation based on classical C-V model was proposed in this paper. It integrated with k-means clustering to segment pathological tissue of Brain CT and MRI, and satisfactory results were obtained. Specially, experimental result of a synthetic image shows that the proposed method is more accurate and robust than some traditional evolution equations.

关 键 词：活动轮廓模型 水平集 图像分割 K-MEANS聚类 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]