基于灰度统计和快速步进的肝脏自动分割方法

Automatic Liver Segmentation Method Based on Intensity Statistics and Fast Marching

为了提高肝脏自动分割效率,解决传统手工选取种子点影响肝脏自动分割算法稳定性的问题,提出了一种基于灰度统计和快速步进的肝脏自动分割方法。该方法通过对肝脏CT图像序列进行灰度统计分析,能够快速定位到肝内种子点;然后,利用各向异性扩散滤波、梯度强度滤波和Sigmoid滤波对CT图像进行平滑滤噪、增强边缘的预处理操作,以提高肝脏边缘的提取效果;最后,在自动获取种子点和预处理图像的基础上,利用快速步进法高效准确的分割出肝脏轮廓。实验结果表明,快速步进法的肝脏分割结果不再因人工选取种子点位置变化而不同。

In order to increase the accuracy of liver segmentation, solve the problem of traditional manual initialization for seed point selection in liver segmentation process, we present an automatic liver segmentation method based un intensity statistics and fast marching. According to statistics and analysis of liver intensity on the sequence of CT images, the seed points can be obtained quickly. Then, to improve the result of liver segmentation, we use anisotropic diffusion filter, gradient intensity filter and Sigmoid filter for smoothing, de-noising and enhancing the edge of CT images. Finally, based on the seed points obtained automatically and the preprocessing, fast marching method can segment liver contour efficiently and accurately. The experimental results show that the liver segmentation result of automatic fast marching method will not change by different positions of the seed selected manually.