基于先验知识的肝脏轮廓线提取算法研究

准确有效地提取肝脏CT序列的轮廓线是腹部软组织三维模型重建与可视化的关键问题之一。针对肝脏轮廓线提取准确性不高的问题, 提出了一种基于先验知识的肝脏轮廓线提取算法。首先利用拉普拉斯算法进行CT图像增强, 再利用基于边缘先验知识的套索模型对感兴趣区域进行半自动的初始化, 最后通过改进的Snake算法准确地提取肝脏CT图像的边缘。针对序列CT肝脏的边缘提取, 提出根据CT图像序列之间的相关性, 将上一幅图像的轮廓线提取结果作为下一幅CT图像边缘提取的初始化点, 接着批处理地提取CT序列的肝脏边缘。实验结果表明：该算法大大减少了手动初始化结果对目标边缘轮廓准确提取的依赖性, 并有效地解决了肝脏轮廓线的提取问题。

Liver contour extraction based on priori knowledge model

Accurately and efficiently contour extraction for the liver CT image sequence is one of the key problems of the abdominal soft-tissue model reconstruction and visualization. This paper proposed a contour extraction method based on a priori knowledge model. Firstly, it employed the Laplace transformation algorithm for image enhancement, and then proposed lasso model based on a priori knowledge of edges for semi-automatic initialization. Finally, it extracted contours of the liver accurately by an improved Snake algorithm. Furthermore, after the first image extraction, it proposed to initialize other CT images automatically according to the correlation between sequence, i. e. extraction result of previous image was employed as the initialization points of the next image, and so on. In this way, it preformed the whole extraction in a very elegant batch fashion. The experimental results on both liver images and lung images show that the proposed algorithm greatly reduces the dependence on initializing accuracy while effectively solve the problem of the liver contour extraction.