置信连接的自动肝脏分割方法

Confidence Connected Method for Automatic Liver Segmentation

肝脏分割是计算机辅助肝病诊断和手术计划制定的基础,文中结合置信连接的区域生长算法实现了肝脏的全自动分割.首先利用改进的曲线各向异性扩散滤波对CT图像进行平滑预处理,以便在去除噪声的同时保存边缘信息,进而通过对预处理图像的灰度分析自动选取序列种子点;然后使用置信连接的区域生长算法对肝脏进行分割;最后采用空洞填充法填补区域生长中产生的空洞,改善分割效果.对10套腹部CT图像数据的实验结果表明,分割每幅图像的平均时间是1.46s,准确率为93.6%,其高效、准确性为临床诊断和手术导航提供了有利的信息.

Liver segmentation is the primary step for computer-aided liver disease diagnosis and surgery planning. In this paper, we present a fully automatic method for liver segmentation based on the confidence connected region growing. First, a modified curvature anisotropic diffusion filter is applied to CT images for noise reduction while preserving the liver structures, and then a series of seed points is selected automatically by intensity analysis. Then, the liver is segmented with a confidence connected region growing algorithm starting from the seed points. Finally, cavity filling method is used to improve the results of region growing. When tested on 10 abdominal CT image datasets, the average time for liver segmentation from one slice is 1.46 s, and the average segmentation accuracy is 93.6%. Experimental results show that the proposed approach is accurate and efficient enough for the applications in clinical diagnosis and surgical navigation.