基于空间模糊C均值算法的MRI图像分割  

MRI Image Segmentation Based on Spatial Fuzzy C-means Algorithm

作　　者：张冬冬 王静 ZHANG Dong-dong;WANG Jing(Medical Equipment Department of Hengshui People's Hospital,Hengshui 053000 China;School of Electronic and Information Engineering,Ankang University,Ankang 725000 China)

机构地区：[1]衡水市人民医院医学装备处,河北衡水053000 [2]安康学院电子与信息工程学院,陕西安康725000

出　　处：《自动化技术与应用》2024年第6期15-18,59,共5页Techniques of Automation and Applications

基　　金：国家自然科学基金(61801005)。

摘　　要：针对传统模糊c均值聚类(FCM)算法对磁共振成像(MRI)图像分割效果不佳的缺点,提出一种基于空间模糊C均值(SpFCM)算法的MRI图像分割新方法。通过定义概率函数设计以每个像素邻域空间信息为中心的方形邻域窗口,将空间信息转化为新的隶属度函数,为每个像素生成局部隶属度值。通过构建加权联合隶属度函数,利用局部和全局隶属度值来计算最终的隶属度值和聚类中心,有效地解决了MRI图像分割对噪声和强度非均匀性(IIH)的敏感问题。实验结果表明,SpFCM算法在抑制噪声、聚类有效性函数和组织分割精度方面具有较好的性能。In view of the poor effect of traditional Fuzzy C-means clustering(FCM)algorithm on Magnetic Resonance Imaging(MRI)image segmentation,a new MRI image segmentation method based on spatial Fuzzy C-means(SpFCM)algorithm is proposed in this paper.By defining the probability function,a square neighborhood window centered on the spatial information of each pixel neighborhood is designed,and the spatial information is transformed into a new membership function,so as to generate local membership values for each pixel.By constructing the weighted joint membership function and using the local and global membership values to calculate the final membership value and clustering center,the sensitivity of MRI image segmentation to noise and intensity non-uniformity(IIH)is effectively solved.Experimental results show that SpFCM algorithm has good performance in noise suppression,clustering effectiveness function and tissue segmentation accuracy.

关 键 词：图像分割 磁共振成像 模糊C均值聚类 空间模糊 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]