中文摘要：

由于肝脏核磁图像存在边缘模糊、噪声大、灰度分布不均匀等特点,一般分割算法效果不甚理想.为了提高分割效果,提出了一种基于先验形状信息的核图割（kernel graph cuts,KGC）模型.采用区域增长算法在待分割的肝脏区域进行预分割,再用形态学算子进行膨胀腐蚀操作,形成初始分割轮廓;将形状模板集和初始轮廓配准后进行核主成分分析（kernel principle component analysis,KPCA）,训练出先验形状信息;在kernel Graph cuts模型的能量函数中融入先验形状信息,并在初始轮廓的基础上进行进一步精确分割.实验结果表明,提出的方法能准确分割出核磁图像中边界模糊、噪声污染大的肝脏边界,且无边界泄露和相似组织误分割等现象.

英文摘要：

Since the MR images have the characteristics of blurred boundary,intensity inhomogeneity and noisy,the common segment algorithms are not efficient.A novel method for liver segmentation of MR images is proposed which incorporates the kernel Graph cuts with shape prior information.First,the region growing algorithm is used and the operation of dilate/erode to form initial contour for segmentation.Second,to get the shape priors,KPCA is used to train the shape template set after registration procedures with initial contour,and all the shape templates are collected from different patients and be segmented by exper：ts.Finally,the shape priors are integrated into the kernel Graph cuts energy function to form a new model,which is to achieve the accurate image segmentation.The experimental results show that we can get a satisfied result without boundary leakage and error segmentation for similar tissues.