基于网格的最小交叉熵算法在PET图像重建中的应用

Minimum Cross-Entropy Reconstruction of PET Images Based on a Content-Adaptive Mesh Model

基于内容的自适应三角形网格模型是描述图像的一种有效方法,本文将网格模型与最小交叉熵算法相结合,并加入先验解剖信息,用于PET图像重建.在本文提出的新算法中,先将投影数据用滤波反投影方法(FBP)生成参考图像,再对参考图像提取网格节点,用加入先验解剖信息的最小交叉熵算法对网格节点灰度值进行迭代计算,最后利用迭代后的网格节点灰度值对象素点进行插值得到重建后的图像.在仿真实验中,将该算法与最大似然方法(MLEM)等算法作比较,并分析了参数对重建结果的影响.

Content-adaptive mesh modeling is an efficient method for image representation. In this paper, the minimum crossentropy algorithm using prior anatomical information, combined with mesh model, was applied to the reconstruction of PET images. In the proposed algorithm, the nodes of mesh model were extracted from a reference image obtained with FBP method; then, the val- ues of the nodes were computed through the minimum cross-entropy algorithm with prior anatomical information. Finally, the whole image was reconstructed by interpolation from the values of the nodes. The performance of the proposed method was tested and compared with other algorithms using a set of simulated data. The effect of the parameters on the result was also studied.