一种数字人脑部切片图像分割新方法  

A new segmentation method for slice images of digital human brain

作　　者：罗洪艳[1] 李敏[1] 谭立文[2] 郑小林[1] 张绍祥[2] 侯文生[1] 

机构地区：[1]重庆大学生物工程学院,重庆400030 [2]第三军医大学基础部解剖学教研室,重庆400038

出　　处：《中国医学影像技术》2009年第8期1488-1491,共4页Chinese Journal of Medical Imaging Technology

基　　金：国家自然科学基金(60771025)

摘　　要：目的提出一种人脑切片图像自动分割算法,以克服现有的方法对大量人工参与的依赖。方法针对人脑切片图像的特征,提出一种基于区域生长的灰度直方图阈值化分割算法。首先通过区域生长过程对图像进行初始的粗分割,再用直方图阈值化方法进行二次细分割提取目标区域。结果采用此方法准确有效地分割出了大脑白质和大脑皮质。结论此算法结合切片图像的全局信息和局部信息应用于分割,是一种比较好的分割方法。Objective To present a new automatic segmentation algorithm for human slice images,in order to reduce the massive manual intervention in the existing segmentation methods. Methods According to the features of slice images of digital human brain, a segmentation algorithm based on the theory of region growing and threshold in normal gray histogram was proposed. More exactly, the slice images were initially segmented coarsely by means of the region growing. Then the method of threshold in normal gray histogram was adopted to refine the segmentation. Results White matter and cerebral cortex were segmented accurately and effectively with this method. Conclusion This algorithm characterized by the combination of global information and local information of a slice image for automatic segmentation exhibits good performance.

关 键 词：切片图像分割 白质 皮质 区域生长 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]