中文摘要：

目的随着医学图像数据的急剧增长,建立从医学图像中自动分割特定解剖结构的算法.方法首先,获取的脑图像体数据集通过与参考体数据集的配准,使对应层图像包含与参考数据相似的解剖结构;然后利用训练得到的统计形状模型自动定位、分割指定的解剖结构.结果实验表明这种算法能取得良好的分割结果.结论本文提出的基于互信息的图像配准和统计形状模型的分割算法,能够实现从体数据中自动定位解剖结构所在的图像位置并分割出目标结构.

英文摘要：

Objective To design an automatic segmentation algorithm of certain anatomy structure, with the rapid increasing of medical image data. Methods Firstly, image registration is applied between the acquired brain datasets and the reference datasets, which making the corresponding slice of the two datasets contain the same anatomy; Secondly, the active shape model is adopted to locate and segment the target. Results Datasets have been used to test our algorithm, and results show that our method is effective. Conclusion In this paper, we describe a segmentation method, integrating image registration based on mutual information and active shape model, which can find the corresponding image where the anatomy structure is located and segment the target automatically.