重建方法对 DaTSCAN 图像定量分析的影响。
The impact of reconstruction method on the quantification of DaTSCAN images.
机构信息
Institute of Nuclear Medicine, UCLH NHS Foundation Trust and University College London, London, UK.
出版信息
Eur J Nucl Med Mol Imaging. 2010 Jan;37(1):23-35. doi: 10.1007/s00259-009-1212-z.
PURPOSE
Reconstruction of DaTSCAN brain studies using OS-EM iterative reconstruction offers better image quality and more accurate quantification than filtered back-projection. However, reconstruction must proceed for a sufficient number of iterations to achieve stable and accurate data. This study assessed the impact of the number of iterations on the image quantification, comparing the results of the iterative reconstruction with filtered back-projection data.
METHODS
A striatal phantom filled with (123)I using striatal to background ratios between 2:1 and 10:1 was imaged on five different gamma camera systems. Data from each system were reconstructed using OS-EM (which included depth-independent resolution recovery) with various combinations of iterations and subsets to achieve up to 200 EM-equivalent iterations and with filtered back-projection. Using volume of interest analysis, the relationships between image reconstruction strategy and quantification of striatal uptake were assessed.
RESULTS
For phantom filling ratios of 5:1 or less, significant convergence of measured ratios occurred close to 100 EM-equivalent iterations, whereas for higher filling ratios, measured uptake ratios did not display a convergence pattern. Assessment of the count concentrations used to derive the measured uptake ratio showed that nonconvergence of low background count concentrations caused peaking in higher measured uptake ratios. Compared to filtered back-projection, OS-EM displayed larger uptake ratios because of the resolution recovery applied in the iterative algorithm.
CONCLUSION
The number of EM-equivalent iterations used in OS-EM reconstruction influences the quantification of DaTSCAN studies because of incomplete convergence and possible bias in areas of low activity due to the nonnegativity constraint in OS-EM reconstruction. Nevertheless, OS-EM using 100 EM-equivalent iterations provides the best linear discriminatory measure to quantify the uptake in DaTSCAN studies.
目的
使用 OS-EM 迭代重建对 DaTSCAN 脑研究进行重建，比滤波反投影提供更好的图像质量和更准确的定量分析。然而，重建必须进行足够数量的迭代，以实现稳定和准确的数据。本研究评估了迭代次数对图像定量的影响，比较了迭代重建和滤波反投影数据的结果。
方法
使用（123）I 填充纹状体的纹状体体模，纹状体与背景的比值在 2:1 到 10:1 之间，在五种不同的伽马相机系统上进行成像。使用 OS-EM（包括深度独立的分辨率恢复）对每个系统的数据进行重建，其中迭代和子集的组合方式不同，最高可达 200 个 EM 等效迭代，同时也使用滤波反投影。通过使用感兴趣区域分析，评估图像重建策略与纹状体摄取定量之间的关系。
结果
对于填充比为 5:1 或更低的体模，在接近 100 个 EM 等效迭代时，测量比值明显收敛，而对于更高的填充比，测量的摄取比值没有显示出收敛模式。对用于推导测量摄取比值的计数浓度的评估表明，由于在迭代算法中应用了分辨率恢复，低背景计数浓度的非收敛导致较高的测量摄取比值出现峰值。与滤波反投影相比，OS-EM 显示出更大的摄取比值，因为迭代算法中应用了分辨率恢复。
结论
OS-EM 重建中使用的 EM 等效迭代次数会影响 DaTSCAN 研究的定量分析，原因是不完全收敛和 OS-EM 重建中的非负约束可能导致低活性区域的偏差。然而，使用 100 个 EM 等效迭代的 OS-EM 提供了量化 DaTSCAN 研究摄取的最佳线性判别测量。
Zotero 插件