低光子数PET图像的投影空间统计重建方法

摘要

A new statistical method for low photon count positron emission tomography(PET)image reconstruction is proposed. Different from the conventional maximum a posterioigr (MAP) method, the cost functions, including likelihood and penalty term are defined in the projection space. It describes the global feature of the image. The penalty term is designed according to the local impulse response to provide uniformly distributed spatial resolution. Simulation results show that the reconstruction quality is improved for low photon count PET image reconstruction.%提出了一种适用于低光子数观测数据的正电子发射断层图像重建的最大后验概率密度(Maximum a posterior,MAP)方法.与传统的MAP算法不同,本文将投影空间看作Gibbs随机场,并在该空间构造投影空间先验模型.根据系统的局部冲击响应对先验项进行修正.使得该方法能够获得近似一致的空间分辨率.实验结果表明,该方法在观测数据光子数较低的情况下,能够获得较好的图像重建结果.