基于level sets的医学图像分割

Medical Image Segmentation Based on Level Sets Method

医学图像分割是一个非常重要的研究领域。它主要应用于病人诊断、图像引导手术,以及医学数据可视化。解决这个问题的一个常用方法就是利用活动轮廓或“snake”来分割感兴趣的物体。文中给出两种活动轮廓模型,其中一种基于边缘停止函数,而另一种是一个能量最小化算法。两种方法都采用level-sets模型,利用一个Lipschitz函数来进行自动拓扑变化。实验表明第一种方法仅仅只能检测边缘梯度较大的物体,而第二种方法没有这样的限制。

Segmentation of structure in medical images is an important research topic. It is used in patient diagnose, image-guided surgery, and medical data visulization. One common approach to solve this problem is to segment objects of interest with active contours or ＂snake＂. Two active contour models, one based on an edge-stopping function, while the other is an energy minimization algorithm, are demonstrated. Both methods can be put into a level-set framework using a Lipschitz function Ф for automatic topology changes. The experiments show that the first method can only detect object defined by a strong gradient, while the second method does not have this constraint.