Atlas-based automated segmentation and anatomical identification of tubular and sheet structures from 3D medical images

一个全自动的分割系统，从三维医学图像的管状和片状结构的解剖识别已经开发的基础上的统计图谱表示的形状和解剖结构的相互关系的约束，这是通过统计分析的患者解剖地面实况数据集的数量构建。利用该系统，可以从三维CT图像中自动分割和识别出腹部动、静脉和门静脉等管状结构以及叶间裂和关节软骨等片状结构。

A fully-automated segmentation system for anatomical identification of tube-like and sheet-like structures from 3D medical images has been developed based on statistical atlas representing the constraints of shapes and interrelations of the anatomical structures, which is constructed by statistical analysis of a number of patient anatomical ground truth datasets. Using the system, abdominal arteries/ veins and portal veins as tube-like structures and interlobar fissure and articular cartilage as sheet-like structures were automatically segmented and identified from 3D CT images.