Biomechanics Section, Mechanical Engineering, KU Leuven, Leuven, Belgium; AO Research Institute Davos, Davos, Switzerland.
Muscles & Movement, Department of Development and Regeneration, KU Leuven Campus Kulak, Kortrijk, Belgium.
Bone. 2022 Jan;154:116225. doi: 10.1016/j.bone.2021.116225. Epub 2021 Oct 9.
High-resolution peripheral quantitative computed tomography (HR-pQCT) devices can scan extremities at bone microstructural level in vivo and are used mainly in research of bone diseases. Two HR-pQCT scanners are commercially available to date: XtremeCT (first generation) and XtremeCT-II (second generation) from Scanco Medical AG (Switzerland). Recently, we have proposed an adaptive local thresholding (AT) technique and showed that it can improve quantification accuracy of bone microstructural parameters, with visually less sharp cone-beam CT (CBCT) images providing a similar accuracy than XtremeCT. The aim of this study was to evaluate whether the AT segmentation technique could enhance the accuracy of HR-pQCT in quantifying bone microstructural images and to assess whether the agreement between XtremeCT and XtremeCT-II could be improved. Nineteen radii were scanned with three scanners from Scanco Medical AG: (1) XtremeCT at 82 μm, (2) XtremeCT-II at 60.7 μm and (3) the small animal microCT scanner VivaCT40 at 19 μm voxel size. The scans were segmented applying two different methods, once following the manufacturer standard technique (ST), and once by means of AT. Three-dimensional (3D) morphological analysis was performed on the trabecular volume of the segmented images using the manufacturer's standard software to calculate bone volume fraction (BV/TV), trabecular thickness (Tb.Th), separation (Tb.Sp) and number (Tb.N). The average accuracy of XtremeCT improved from R = 0.76 (ST) to 0.85 (AT) and reached the same level of accuracy as XtremeCT-II with ST (R = 0.86). The largest improvements were obtained for BV/TV and Tb.Th. For XtremeCT-II, mean accuracy improved slightly from R = 0.86 (ST) to 0.89 (AT). For both segmentations and both scanners, the standard section was quantified slightly more accurate than the subchondral section. The agreement between the scanners was enhanced from R = 0.89 (ST) to 0.98 (AT). In conclusion, AT can enhance the accuracy of XtremeCT to quantify distal radius bone microstructural parameters close to XtremeCT-II level and increases the agreement between the two HR-pQCT scanners. High-resolution peripheral quantitative computed tomography, segmentation, bone microstructural parameters.
高分辨率外周定量计算机断层扫描（HR-pQCT）设备可在体内对四肢骨骼微结构进行扫描，主要用于骨骼疾病的研究。目前有两种 HR-pQCT 扫描仪可供使用：第一代的 XtremeCT 和第二代的 XtremeCT-II，均来自于 Scanco Medical AG（瑞士）。最近，我们提出了一种自适应局部阈值（AT）技术，结果表明，该技术可以提高骨微结构参数的定量准确性，与 XtremeCT 相比，视觉上不太清晰的锥束 CT（CBCT）图像可以提供相似的准确性。本研究旨在评估 AT 分割技术是否可以提高 HR-pQCT 对骨骼微结构图像的定量准确性，并评估 XtremeCT 和 XtremeCT-II 之间的一致性是否可以得到改善。我们使用 Scanco Medical AG 的三种扫描仪对 19 根桡骨进行了扫描：（1）XtremeCT，分辨率为 82μm；（2）XtremeCT-II，分辨率为 60.7μm；（3）小动物 microCT 扫描仪 VivaCT40，体素大小为 19μm。扫描结果应用两种不同的方法进行分割，一种是按照制造商的标准技术（ST），另一种是使用 AT。使用制造商的标准软件对分割后的图像进行三维（3D）形态学分析，以计算骨体积分数（BV/TV）、骨小梁厚度（Tb.Th）、分离度（Tb.Sp）和数量（Tb.N）。XtremeCT 的平均准确性从 R=0.76（ST）提高到 0.85（AT），与 ST 下的 XtremeCT-II 达到相同的准确性（R=0.86）。BV/TV 和 Tb.Th 的准确性提高最大。对于 XtremeCT-II，平均准确性从 R=0.86（ST）略有提高到 0.89（AT）。对于两种分割方法和两种扫描仪，标准节段的定量准确性均略高于软骨下节段。扫描仪之间的一致性从 R=0.89（ST）提高到 0.98（AT）。总之，AT 可以提高 XtremeCT 对桡骨远端骨微结构参数的定量准确性，使其接近 XtremeCT-II 水平，并提高了两种 HR-pQCT 扫描仪之间的一致性。高分辨率外周定量计算机断层扫描，分割，骨微结构参数。
