基于两阶段区域生长的肝内血管分割算法

Liver vessel segmentation algorithm using two-stage region growing

提出一种基于两阶段区域生长法的肝内血管分割算法。在第一阶段,使用三维区域生长法从CT图像序列中提取出肝脏,区域生长法的生长准则由均值和方差双阈值来决定。分别通过形态学运算和中值滤波对分割出的肝脏序列进行降噪处理。在第二阶段,再次利用区域生长法从已经得到的三维肝脏图像序列中分割出血管,区域生长法的生长准则仅由均值作为单阈值决定。对两阶段区域生长法的实验结果进行三维重建,与传统区域生长法的三维重建结果相比较,能够体现出该算法的准确性。

The algorithm for liver vessel segmentation is proposed by using the two-stage region growing. In the first stage, the 3D region growing constrained by intensity and variance is performed to isolate liver from CT sequence. Then the fol-lowing morphologic operations and median filter are employed to eliminate noise. In the second stage, another region growing restricted with intensity is used to extract vessels from the resulted 3D liver data. Compared to single region growing, the experimental results of 3D reconstruction by the method depict the accuracy of the proposed algorithm.