Department of Orthodontics and Dentofacial Orthopedics, University of Bern, CH-3010, Bern, Switzerland.
Department of Orthodontics and Dentofacial Orthopedics, 251 Hellenic Air Force General Hospital, GR-11525, Athens, Greece.
Sci Rep. 2020 Apr 30;10(1):7361. doi: 10.1038/s41598-020-64383-9.
The use of a single grey intensity threshold is one of the most straightforward and widely used methods to segment cranial base surface models from a 3D radiographic volume. In this study we used thirty Cone Beam Computer Tomography (CBCT) scans from three different machines and ten CT scans of growing individuals to test the effect of thresholding on the subsequently produced anterior cranial base surface models. From each scan, six surface models were generated using a range of voxel intensity thresholds. The models were then superimposed on a manually selected reference surface model, using an iterative closest point algorithm. Multivariate tests showed significant effects of the machine type, threshold value, and superimposition on the spatial position and the form of the created models. For both, CT and CBCT machines, the distance between the models, as well as the variation within each threshold category, was consistently increasing with the magnitude of difference between thresholds. The present findings highlight the importance of accurate anterior cranial base segmentation for reliable assessment of craniofacial morphology through surface superimposition or similar methods that utilize this anatomical structure as reference.
使用单一灰度阈值是将颅底表面模型从 3D 射线照相体积中分割出来的最直接和广泛使用的方法之一。在这项研究中，我们使用了来自三台不同机器的三十个锥形束 CT（CBCT）扫描和十个生长个体的 CT 扫描，以测试阈值对随后产生的前颅底表面模型的影响。从每个扫描中，使用一系列体素强度阈值生成了六个表面模型。然后，使用迭代最近点算法将这些模型叠加在手动选择的参考表面模型上。多变量检验表明，机器类型、阈值和叠加对创建模型的空间位置和形状有显著影响。对于 CT 和 CBCT 两种机器，模型之间的距离以及每个阈值类别的变化，都随着阈值之间差异的增大而持续增加。本研究结果强调了准确的前颅底分割对于通过表面叠加或类似方法可靠评估颅面形态的重要性，这些方法将该解剖结构用作参考。
