改进的B-Snake模型肝脏CT图像分割算法

Improved B-Snake segmentation method for liver CT images

肝脏模型的个性化是肝脏虚拟手术系统中的一个关键技术,而肝脏模型的个性化又是以肝脏CT图像的三维分割为前提的。针对B-Snake模型的特点,提出一种结合区域填充的改进B-Snake模型图像分割算法。将相邻的上一张切片的分割结果映射到当前切片上,根据一定的规则进行区域填充,并将填充后的结果与前一张切片的分割结果按一定的算法进行比较,进一步优化。得到的初始轮廓很接近肝脏的真实边界,而且大部分曲线已在边界上,将其作为改进的B-Snake模型算法的初始轮廓,只需对其进行部分控制点的优化调整,就可得到准确的分割结果。以此类推,直到处理完所有切片图。实验表明,该算法能有效提高分割的准确度,获得较满意的分割结果。

The personalization of liver models, which is premised on the 3d segmentation of liver CT images, is a key technology in the virtual surgery of liver. Considering the features of B-Snake model, this paper presents an improved B-Snake segmentation algorithm combined with Region Filling. The contour of the adjacent and processed section is mapped on the current section. Based on the contour, it gets a connected region according to Region Filling algorithm and compares the region with the liver region of the adjacent and processed section according to certain algorithm in order to obtain a more accurate contour. The resulting contour is close to the liver boundary, and large amount of the control points are on the right boundary. Then, the contour is served as the initial contour of the improved B-Snake algorithm for further processing, resulting in the final segmentation result after the evolution of part of the initial contour. The algorithm will not end untill all sections are processed. Experimental results show that the algorithm can obtain segmentaion result of liver CT images efficiently and accurately.