OSEM子集及迭代次数对PET图像质量的影响

The Effects of the Subsets and Iterative Times in Reconstruction of OSEM on PET Image Quality

目的 :探讨有序子集最大期望重建算法 (OSEM )中子集及迭代次数对PET图像质量的影响。材料和方法 :圆柱模型 (长 18.5cm ,直径 2 1.5cm)内灌注比活度为 1.2 5kBq/ml18F 水溶液形成本底 ,平行内置的 7个不同直径空心小圆柱内灌注18F 水溶液 (比活度 12 5kBq/ml)形成热灶 ,以 2D方式采集数据、OSEM算法重建图像 ,子集水平 3 2、16,迭代次数 1～10。在图像横断面对应外围 7个热灶画与其同大小的ROI及本底ROI ,由此计算CV等参数。结果 :相同条件下 3 2子集重建图像的热灶活度比 16子集更接近真实值。变异系数CV ,在较小热灶及背景区中 3 2子集的大于 16子集 ,较大的热灶中 3 2子集的则小于 16子集。结论 :OSEM重建方式选择 3 2子集 2～ 3次迭代 ,或 16子集 3～ 4次迭代的图像质量较好。

Purpose: To evaluate the effects of the number of order subsets and times of iterative reconstruction on PET image quality. Materials and Methods: A circular cylinder phantom filled with 18 F solution of 1.25kbq/ml as background, with 7 small casts filled with 125kbq/ml 18 F as the hot lesions. The data acquired with 2D and images reconstructed with OSEM at subset 32 and 16 respectively, the range of iterative is from 1 to 10. The mean activity and standard deviation in each ROI, and the coeffective variance were calculated. Results: The activity of the hot lesion at 32 subset is closer to actual value than that at 16 subset. The coeffective variance(CV) is larger in 32 than in 16 subset on smaller lesion, but the result opposite over larger lesions. Conclusion: The images reconstructed with 32 subsets and 2 to 3 times of iteration, or with 16 subsets and 3 to 4 times of iterations, were with better qualities.