Yan Jianhua, Planeta-Wilson Beata, Carson Richard E
PET center, Yale University, New Haven, CT 06520 USA.
IEEE Nucl Sci Symp Conf Rec (1997). 2008;4774103:3625-3628. doi: 10.1109/NSSMIC.2008.4774103.
We present a direct method for producing images of kinetic parameters from list mode PET data. The time-activity curve for each voxel is described by a one-tissue compartment, 2-parameter model. Extending previous EM algorithms, a new spatiotemporal complete data space was introduced to optimize the maximum likelihood function. This leads to a straightforward parametric image update equation with moderate additional computation requirements compared to the conventional algorithm. Qualitative and quantitative evaluations were performed using 2D (x,t) and 4D (x,y,z,t) simulated list mode data for a brain receptor study. Comparisons with the two-step approach (frame-based reconstruction followed by voxel-by-voxel parameter estimation) show that the proposed method can lead to accurate estimation of the parametric image values with reduced variance, especially for the volume of distribution (V(T)).
我们提出了一种从列表模式PET数据生成动力学参数图像的直接方法。每个体素的时间-活度曲线由单组织隔室二参数模型描述。在扩展先前的期望最大化(EM)算法的基础上,引入了一个新的时空完全数据空间来优化最大似然函数。这产生了一个直接的参数图像更新方程,与传统算法相比,其额外的计算需求适中。使用用于脑受体研究的二维(x,t)和四维(x,y,z,t)模拟列表模式数据进行了定性和定量评估。与两步法(基于帧的重建,随后逐体素进行参数估计)的比较表明,所提出的方法能够以降低的方差准确估计参数图像值,特别是对于分布容积(V(T))。