New standards for phantom image quality and SUV harmonization range for multicenter oncology PET studies
New standards for phantom image quality and SUV harmonization range for multicenter oncology PET studies
复制标题
DOI：
10.1007/s12149-021-01709-1
复制
发表时间：
2022-01-14
影响因子：
2.6
通讯作者：
Tateishi, Ukihide
中科院分区：
文献类型：
--
作者：
Akamatsu, Go;Shimada, Naoki;Tateishi, Ukihide
F-18-FDG PET 在肿瘤学中的关键作用不仅是用于病变检测、分期和表征的视觉解释，而且定量治疗反应评估也是 F-18-FDG PET 的关键作用。在多中心肿瘤学 PET 研究中，图像质量标准化和 SUV 协调对于获得可靠的研究结果至关重要。图像质量和 SUV 协调范围的标准应根据扫描仪性能的进步定期更新。因此，本研究的首要目标是提出新的图像质量参考水平，以确保小病变的可检测性。第二个目标是提出一个新的 SUV 协调范围和图像噪声标准，以最大限度地减少扫描仪间和扫描仪内 SUV 的变异性。我们使用 NEMA NU2 图像质量模型从 23 个最新 PET/CT 扫描仪型号中收集了总共 37 个图像模式。对 30-300 秒和 1800 秒不同采集持续时间的 PET 图像进行视觉和定量分析，得出 10 毫米直径热球、噪声当量计数 (NECphantom)、10 毫米球对比度 (Q(H,10 mm))、背景变异性 (N-10 mm)、对比度噪声比 (Q(H,10 mm)/N-10毫米）， 图像噪声水平 (CVBG) 以及热球（直径 10-37 毫米）的 SUVmax 和 SUVpeak。我们计算了每个图像质量指标的参考水平，以便可以目视检测 10 毫米球体。考虑到点扩散函数（PSF）重建引起的超调，提出了 SUV 协调范围和图像噪声准则。我们提出的图像质量参考水平如下：Q(H,10 mm)/N-10 mm >= 2.5 和 CVBG
Not only visual interpretation for lesion detection, staging, and characterization, but also quantitative treatment response assessment are key roles for F-18-FDG PET in oncology. In multicenter oncology PET studies, image quality standardization and SUV harmonization are essential to obtain reliable study outcomes. Standards for image quality and SUV harmonization range should be regularly updated according to progress in scanner performance. Accordingly, the first aim of this study was to propose new image quality reference levels to ensure small lesion detectability. The second aim was to propose a new SUV harmonization range and an image noise criterion to minimize the inter-scanner and intra-scanner SUV variabilities. We collected a total of 37 patterns of images from 23 recent PET/CT scanner models using the NEMA NU2 image quality phantom. PET images with various acquisition durations of 30-300 s and 1800 s were analyzed visually and quantitatively to derive visual detectability scores of the 10-mm-diameter hot sphere, noise-equivalent count (NECphantom), 10-mm sphere contrast (Q(H,10 mm)), background variability (N-10 mm), contrast-to-noise ratio (Q(H,10 mm)/N-10 mm), image noise level (CVBG), and SUVmax and SUVpeak for hot spheres (10-37 mm diameters). We calculated a reference level for each image quality metric, so that the 10-mm sphere can be visually detected. The SUV harmonization range and the image noise criterion were proposed with consideration of overshoot due to point-spread function (PSF) reconstruction. We proposed image quality reference levels as follows: Q(H,10 mm)/N-10 mm >= 2.5 and CVBG