磁共振图像的随机场分割方法

Markov Random Field Magnetic Resonance Image Segmentation

基于 Gibbs分布的 Markov随机场是一个重要的先验模型 ,能够简单地通过势能形式表示图像像素之间的相互作用 ,从而把图像的先验知识和图像分割的数学模型相结合 .利用 Markov随机场方法提出了脑磁共振图像最大后验概率的分割模型 ,并通过迭代条件方法求解 ,与传统的 K均值算法作比较 。

A Markov random field segmentation algorithm was presented here to segment brain MR images. It is important to combine the medical prior knowledge with image segmentation model of medical image segmentation. Markov random field with Gibbs distribution is an important prior model in image segmentation. It can simply express the interaction between pixels via potential energy. The iterated conditional method (ICM) was used to solve the maximum a posteriori (MAP) question. Examples were presented for the validation of the algorithm.