Chappard Christine, Basillais Armelle, Benhamou Laurent, Bonassie Alexandra, Brunet-Imbault Barbara, Bonnet Nicolas, Peyrin Francoise
Inserm U 658, 1 rue Porte Madeleine, Orleans Centre 45000, France.
Med Phys. 2006 Sep;33(9):3568-77. doi: 10.1118/1.2256069.
Microcomputed tomography (microCT) produces three-dimensional (3D) images of trabecular bone. We compared conventional microCT (CmicroCT) with a polychromatic x-ray cone beam to synchrotron radiation (SR) microCT with a monochromatic parallel beam for assessing trabecular bone microarchitecture of 14 subchondral femoral head specimens from patients with osteoarthritis (n=10) or osteoporosis (n=4). SRmicroCT images with a voxel size of 10.13 microm were reconstructed from 900 2D radiographic projections (angular step, 0.2 degrees). CmicroCT images with a voxel size of 10.77 microm were reconstructed from 205, 413, and 825 projections obtained using angular steps of 0.9 degrees, 0.45 degrees, and 0.23 degrees, respectively. A single threshold was used to binarize the images. We computed bone volume/ tissue volume (BV/TV), bone surface/bone volume (BS/BV), trabecular number (Tb.N), trabecular thickness (Tb.Th and Tb.Th*), trabecular spacing (Tb.Sp), degree of anisotropy (DA), and Euler density. With the 0.9 degrees angular step, all CmicroCT values were significantly different from SRmicroCT values. With the 0.23 degrees and 0.45 degrees rotation steps, BV/TV, Tb.Th, and BS/BV by CmicroCT differed significantly from the values by SRmicroCT. The error due to slice matching (visual site matching +/- 10 slices) was within 1% for most parameters. Compared to SRmicroCT, BV/TV, Tb.Sp, and Tb.Th by CmicroCT were underestimated, whereas Tb.N and Tb. Th* were overestimated. A Bland and Altman plot showed no bias for Tb.N or DA. Bias was -0.8 +/- 1.0%, +5.0 +/- 1.1 microm, -5.9 +/- 6.3 microm, and -5.7 +/- 29.1 microm for BV/TV, Tb.Th*, Tb.Th, and Tb.Sp, respectively, and the differences did not vary over the range of values. Although systematic differences were noted between SRmicroCT and CmicroCT values, correlations between the techniques were high and the differences would probably not change the discrimination between study groups. CmicroCT provides a reliable 3D assessment of human defatted bone when working at the 0.23 degrees or 0.45 degrees rotation step; the 0.9 degrees rotation step may be insufficiently accurate for morphological bone analysis.
微计算机断层扫描（microCT）可生成小梁骨的三维（3D）图像。我们将传统的多色X射线锥束微CT（CmicroCT）与单色平行束同步辐射（SR）微CT进行了比较，以评估14例骨关节炎患者（n = 10）或骨质疏松症患者（n = 4）的股骨髁下股骨头标本的小梁骨微结构。从900个二维放射摄影投影（角度步长为0.2度）重建了体素大小为10.13微米的SRmicroCT图像。分别从使用0.9度、0.45度和0.23度角度步长获得的205、413和825个投影重建了体素大小为10.77微米的CmicroCT图像。使用单一阈值对图像进行二值化处理。我们计算了骨体积/组织体积（BV/TV）、骨表面积/骨体积（BS/BV）、小梁数量（Tb.N）、小梁厚度（Tb.Th和Tb.Th*）、小梁间距（Tb.Sp）、各向异性程度（DA）和欧拉密度。在0.9度角度步长下，所有CmicroCT值与SRmicroCT值均有显著差异。在0.23度和0.45度旋转步长下，CmicroCT的BV/TV、Tb.Th和BS/BV与SRmicroCT的值有显著差异。大多数参数的切片匹配误差（视觉位置匹配±10个切片）在1%以内。与SRmicroCT相比，CmicroCT的BV/TV、Tb.Sp和Tb.Th被低估，而Tb.N和Tb.Th被高估。Bland-Altman图显示Tb.N或DA无偏差。BV/TV、Tb.Th、Tb.Th和Tb.Sp的偏差分别为 -0.8 ± 1.0%、+5.0 ± 1.1微米、-5.9 ± 6.3微米和 -5.7 ± 29.1微米，且差异在数值范围内无变化。尽管SRmicroCT和CmicroCT值之间存在系统差异，但两种技术之间的相关性很高，这些差异可能不会改变研究组之间的区分。当以0.23度或0.45度旋转步长工作时，CmicroCT可为人体脱脂骨提供可靠的3D评估；0.9度旋转步长对于形态学骨分析可能不够准确。
