Issues in interstudy comparisons of bone microarchitecture
Issues in interstudy comparisons of bone microarchitecture
复制标题
骨微结构研究间比较中的问题
作者：
M. Djuric;P. Milovanović;D. Djonic;M. Hahn;R. Marshall;M. Amling
Tassani和Perilli [1]在微型计算机断层扫描（micro-CT）研究中提出了一个有趣的技术问题。我们同意他们的观点，即扫描体积的特征也可能导致所获得数据的差异，在比较不同研究时也应记住这一点。实际上，各种微CT研究使用许多不同的方法、各种扫描分辨率、不同的分割和阈值处理程序以及不同形状和大小的感兴趣体积（VOI）。特别是，标称分辨率和体素大小在研究中变化显著，并且已经认识到这种变化对骨微结构参数的影响[2]。相比之下，VOI位置、形状和大小的可变性通常没有得到很好的评价，尽管在改变这些参数后在灵长类动物骨骼中发现了显著差异[3]。迄今为止，尚未对考虑特定规则几何VOI（圆柱形、球形、立方体）与手动标记（自适应）扫描区域的相对受益（或风险）进行认真评价。然而，建议使用“适应性VOI”的某些优点，因为简单的几何VOI形状无法涵盖所需区域内的所有骨组织[4]。
在我们对骨关节炎患者与健康对照组的人股骨颈骨小梁的显微CT研究中[5]，整个股骨颈被细分为内侧下半部分和外侧上半部分。分割是半自动进行的，这意味着我们在每第25个CT切片上手动标记完整的松质骨感兴趣区域（ROI），然后使用micro-CT程序（瑞士Soho Medical）的Morph功能在所有其他切片上对标记的ROI进行插值。因此，基于每个样本总共200个切片获得厚度为3.6 mm的VOI，并进一步定量分析。下内侧或上外侧颈部的适配VOI的尺寸约为1 cm 3。在组织形态计量学方面，3.6 mm的厚度提供了至少四个或五个小梁层，连同1 cm 3的体积，应代表分析区域。我们从Tassani和Perilli [1]中了解到，除了各向异性程度的值外，其他数据（如H1、H2、H3和决定系数）可能有用，但我们的研究中未记录H值，而决定系数未在Scorio micro-CT评价原则中提供。
然而，除了方法学差异外，研究间比较应仔细考虑检查骨部位的确切位置（例如股骨头[6]与股骨颈亚区[5]），因为大量数据表明骨结构存在部位间差异[7，8]。
总之，由于显微CT结果反映了生物学差异以及方法学印记，因此应始终谨慎比较各种研究的结果，并适当考虑数据采集/分析中的方法学和技术差异。然而，只要以一致的方式分析标本，在单一研究中比较标本可能保持可靠性。
Tassani and Perilli [1] raised an interesting technical question in micro computed tomography (micro-CT) studies. We agree with them that characteristics of the scanned volume might also contribute to differences in obtained data, which should be also borne in mind when comparing different studies. Indeed, various micro-CT studies use a number of different approaches, various scanning resolutions, different segmentation and thresholding procedures and different shape and size of volumes of interest (VOI). In particular, the nominal resolution and voxel size varied significantly among studies, and the effects of that variability on bone microarchitectural parameters have been recognised already [2]. In contrast, variability in VOI location, shape and size is often not well appreciated, although significant differences were found in primate bones after varying these parameters [3]. Hitherto, there has been no serious evaluation of the relative benefits (or risks) of considering a particular regular geometric VOI (cylindrical, spheric, cubic) vs. manually marked (adapted) scanning zone. Yet, certain advantages of using “adapted VOIs” were suggested, as simple geometric VOI shapes cannot encompass all the bone tissue within a desired zone [4].
In our micro-CT study on the human femoral neck’s trabecular bone in osteoarthritic patients vs. healthy controls [5], the entire femoral neck was subdivided into inferomedial and superolateral halves. Segmentation was performed semiautomatically, meaning that we manually marked the complete cancellous region of interest (ROI) on every 25th CT slice, followed by interpolation of the marked ROI on all other slices using the Morph function of the micro-CT programme (Scanco Medical, Switzerland). Therefore, a VOI with a thickness of 3.6 mm was obtained based on a total of 200 slices per specimen and was further analysed quantitatively. Dimensions of an adapted VOI for inferomedial or superolateral neck were approximately 1 cm3. In terms of histomorphometry, 3.6 mm of thickness provides at least four or five trabecular layers that, together with a 1-cm3 volume, should be representative of the analysed region. We understand from Tassani and Perilli [1] that apart from the values of degree of anisotropy, additional data such as H1, H2, H3 and determination coefficients might be useful, but H values were not recorded in our study, whereas the coefficient of determination is not provided in Scanco micro-CT evaluation printouts.
However, in addition to methodological differences, interstudy comparisons should carefully consider exact positions of examined bone sites (e.g. femoral head [6] vs. subregions of the femoral neck [5]), as a substantial amount of data suggests intersite differences in bone structure [7, 8].
In summary, as micro-CT findings reflect biological differences together with methodological imprints, findings from various studies should always be compared with caution, giving due consideration to methodological and technical differences in data acquisition/analysis. However, comparison of specimens within a single study might maintain reliability as long as specimens are analysed in a consistent manner.
影响因子：
4.1
作者：
Kim, DG;Christopherson, GT;Yeni, YN
通讯作者：
Yeni, YN