北京生物医学工程

基于图谱的肝脏CT三维自动分割研究

Atlas Based Automatic Liver 3D CT Image Segmentation

作者：刘伟贾富仓胡庆茂王俊

单位：南京邮电大学，图像处理与图像通信江苏省重点实验室（南京210003）

关键词：图谱;肝脏;自动分割;配准

分类号：

出版年·卷·期（页码）：2011·30·5（457-461）

摘要：

目的在肝脏外科手术或肝脏病理研究中，计算肝脏体积是重要步骤。由于肝脏外形复杂、临近组织灰度
值与之接近等特点，肝脏的自动医学图像分割仍是医学图像处理中的难点之一。方法本文采用图谱结合
3D非刚性配准的方法，同时加入肝脏区域搜索算法，实现了鲁棒性较高的肝脏自动分割程序。首先，利
用20套训练图像创建图谱，然后程序自动搜索肝脏区域，最后将图谱与待分割CT图像依次进行仿射配准
和B样条配准。配准以后的图谱肝脏轮廓即可表示为目标肝脏分割轮廓，进而计算出肝脏体积。结果评
估结果显示，上述方法在肝脏体积误差方面表现出色，达到77分，但在局部（主要在肝脏尖端）出现较
大的误差。结论该方法分割临床肝脏CT图像具有可行性。

Objective Liver segmentation is an important step for the planning and
navigation in liver surgery.Accurate,fast and robust automatic segmentation methods for
clinical routine data are urgently needed.Because of the liver’s characteristics,such as
the complexity of the external form,the similarity between the intensities of the liver and
the tissues around it,automatic segmentation of the liver is one of the difficulties in
medical image processing.Methods In this paper,3D non-rigid registration from a refined
atlas to liver CT images is used for segmentation.Firstly,twenty sets of training images
are utilized to create an atlas.Then the liver initial region is searched and located
automatically.After that threshold filtering is used to enhance the robustness of
segmentation.Finally,this atlas is non-rigidly registered to the liver in CT images with
affine and B-spline in succession.The registered segmentation of liver’s atlas represented
the segmentation of the target liver，and then the liver volume was calculated.Results The
evaluation show that the proposed method works well in liver volume error,with the 77
score,yet appears greater error in local position (mostly in liver tips).Conclusions
Experimental results show that this method is feasible for clinical liver CT image
segmentation.

参考文献：

［1］田捷,包尚联.医学影像处理与分析［M］.北京：电子工业出版社,2003.
［2］夏德深,傅德胜.现代图像处理技术与应用［M］.第2版.南京:东南大学出版社,2001.
［3］Furukawa D,Shimizu A,Kobatake H.Automatic Liver Segmentation Method based on Maximum A
Posterior Probability Estimation and Level Set Method ［C］.The 10th International
Conference on Medical Image Computing and Computer Assisted Intervention,Brisbane,2007:117
-124.
［4］Rusko L,Bekes G,Nemeth G.Fully automatic liver segmentation for contrast-enhanced CT
images ［C］.The 10th International Conference on Medical Image Computing and Computer
Assisted Intervention,Brisbane,2007:143-150.
［5］Seghers D,Slagmolen P,Lambelin AY.Landmark based liver segmentation using local shape
and local intensity models ［C］.The 10th International Conference on Medical Image
Computing and Computer Assisted Intervention,Brisbane,2007:135-142.
［6］Kalinmueller D,Lange T,Lamecker H.Shape constrained automatic segmentation of the
liver based on a heuristic intensity model ［C］.The 10th International Conference on
Medical Image Computing and Computer Assisted Intervention,Brisbane,2007:109-116.
［7］Slagmolen P,Elen A,Seghers D.Atlas based liver segmentation using nonrigid
registration with a B-spline transformation model ［C］.The 10th International Conference
on Medical Image Computing and Computer Assisted Intervention,Brisbane,2007,197-206.
［8］Gerig G,Jomier M,Chakos HM.A new validation tool for assessing and improving 3D object
segmentation ［J］.Lecture Notes in Computer Science,2001,1(1):516-523.
［9］Niessen WJ,Bouma CJ.Error metrics for quantitative evaluation of medical image
segmentation ［J］.Performance Character-ization in Computer Vision,2000,1(1):275-284.
［10］Heimann T,Ginneken B,Styner MA,et al.Comparison and evaluation of methods for liver
segmentation from CT datasets ［J］.IEEE Transactions on Medical Imaging,2009,28(8)：1251-
1265.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录