Quintiens Jilmen, Coudyzer Walter, Bevers Melissa, Vereecke Evie, van den Bergh Joop P, Manske Sarah L, van Lenthe G Harry
Department of Mechanical Engineering, KU Leuven, Celestijnenlaan 300, 3001 Heverlee, Belgium.
Department of Radiology, University Hospital Leuven, Herestraat 49, 3000 Leuven, Belgium.
J Bone Miner Res. 2024 Nov 29;39(12):1774-1782. doi: 10.1093/jbmr/zjae163.
HR-pQCT has become standard practice when quantifying volumetric BMD (vBMD) in vivo. Yet, it is only accessible to peripheral sites, with small fields of view and lengthy scanning times. This limits general applicability in clinical workflows. The goal of this study was to assess the potential of photon counting CT (PCCT) in quantitative bone imaging. Using the European Forearm Phantom, PCCT was calibrated to hydroxyapatite (HA) density. Eight cadaveric forearms were scanned twice with PCCT and once with HR-pQCT. The dominant forearm of two volunteers was scanned twice with PCCT. In each scan, the carpals were delineated. At bone level, accuracy was assessed with a paired measurement of total vBMD (Tt.vBMD) calculated with PCCT and HR-pQCT. At voxel-level, repeatability was assessed by image registration and voxel-wise subtraction of the ex vivo PCCT scans. In an ideal scenario, this difference would be zero; any deviation was interpreted as falsely detected remodeling. For clinical usage, the least detectable remodeling was determined by finding a threshold in the PCCT difference image that resulted in a classification of bone formation and resorption below acceptable noise levels (<0.5%). The paired measurement of Tt.vBMD had a Pearson correlation of 0.986. Compared to HR-pQCT, PCCT showed a bias of 7.46 mgHA/cm3. At voxel-level, the repeated PCCT scans showed a bias of 17.66 mgHA/cm3 and a standard error of 96.23 mgHA/cm3. Least detectable remodeling was found to be 250 mgHA/cm3, for which 0.37% of the voxels was incorrectly classified as newly added or resorbed bone. In vivo, this volume increased to 0.97%. Based on the cadaver data, we conclude that PCCT can be used to quantify vBMD and bone turnover. We provided proof of principle that this technique is also accurate in vivo, hence, that it has high potential for clinical applications.
在体内定量测定体积骨密度（vBMD）时，高分辨率外周定量计算机断层扫描（HR-pQCT）已成为标准方法。然而，它仅适用于外周部位，视野小且扫描时间长。这限制了其在临床工作流程中的普遍适用性。本研究的目的是评估光子计数计算机断层扫描（PCCT）在定量骨成像中的潜力。使用欧洲前臂体模，将PCCT校准至羟基磷灰石（HA）密度。对8具尸体前臂进行了两次PCCT扫描和一次HR-pQCT扫描。对两名志愿者的优势前臂进行了两次PCCT扫描。在每次扫描中，都对腕骨进行了勾勒。在骨水平上，通过对PCCT和HR-pQCT计算的总vBMD（Tt.vBMD）进行配对测量来评估准确性。在体素水平上，通过图像配准和离体PCCT扫描的体素级减法来评估重复性。在理想情况下，这种差异应为零；任何偏差都被解释为错误检测到的骨重塑。对于临床应用，通过在PCCT差异图像中找到一个阈值来确定最小可检测骨重塑，该阈值导致骨形成和骨吸收的分类低于可接受的噪声水平（<0.5%）。Tt.vBMD的配对测量的Pearson相关系数为0.986。与HR-pQCT相比，PCCT的偏差为7.46 mgHA/cm³。在体素水平上，重复的PCCT扫描显示偏差为17.66 mgHA/cm³，标准误差为96.23 mgHA/cm³。发现最小可检测骨重塑为250 mgHA/cm³，其中0.37%的体素被错误分类为新添加或吸收的骨。在体内，这个体积增加到0.97%。基于尸体数据，我们得出结论，PCCT可用于定量vBMD和骨转换。我们提供了原理证明：该技术在体内也很准确，因此，它具有很高的临床应用潜力。
