张玄1, 2, 张亚盟3, 吴秀杰1, 4
1.中国科学院脊椎动物演化与人类起源重点实验室，中国科学院古脊椎动物与古人类研究所, 北京 100044; 2. 中国科学院大学, 北京 100049; 3. 山东大学青岛校区文化遗产研究院，青岛，266237；4中国科学院生物演化与环境卓越创新中心, 北京 100044
摘要：近年来，根据三维软件虚拟复原的头骨来获取测量数据的方法被越来越多地应用在古生物，特别是古人类学的研究中，然而对于三维软件不同精度虚拟复原的头骨，其测量数值是否有差异，研究者并不是很清楚。本文以Mimics软件为例，根据其复原模型简化规则，选择未精简的最佳精度模型作为标准进行配对t检验或非参数检验，通过对43例云南人头骨的顶骨矢状弦长、颅周长、头盖部面积、乳突小房表面积、颅容量、乳突小房体积等六个测量项目的对比和分析，对Mimics软件低、中、高、最佳四种精度3D虚拟复原头骨间的测量差异进行了研究。结果显示：颅周长、头盖部面积、颅容量、乳突小房体积四项的所有简化精度模型的测量数据均与最佳精度模型测量数据的差异具有显著性；而除高精度组外，顶骨矢状弦长及乳突小房表面积的其余精度组测量数据均与最佳精度组差异具有显著性；此外，顶骨矢状弦长、颅周长、头盖部面积、颅容量的简化精度与最佳精度的测量差异占比均小于3%.而乳突小房表面积的低精度与最佳精度测量差异占比可超过50%，乳突小房体积的低精度与最佳精度测量差异占比可超过120%。这一结果提示我们，在测量Mimics复原的三维模型时，体量大差异小的测量项可以在较低精度的复原模型上进行测量；而对头骨内部腔窦这样体量小表面粗糙的结构，复原模型的精度选择及测量数据比较需要格外谨慎。
关键词：Mimics; 电子计算机断层扫描；三维重建；头骨测量；差异检验；精度
The influence of quality parameter selection on 3D virtual reconstruction model precision based on dry skull
ZHANG Xuan1, 2; ZHANG Yameng3; WU Xiujie1, 4
1. Key Laboratory of Vertebrate Evolution and Human Origin of Chinese Academy of Sciences, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing 100044; 2. University of Chinese Academy of Sciences, Beijing 100049；3. Institute of Cultural Heritage, Shandong University, Qingdao 266237 4. CAS Center for Excellence in Life and Paleoenvironment, Beijing 100044;
Abstract: In anthropology, craniometry is an important means for obtaining human skull dimensional information and other characteristics. Computed tomography (CT) and three-dimensional (3D) reconstruction technologies offer significant advantages to craniometry specialists by facilitating both the collection of repeated measurements and analysis of inner structures without destroying specimens. However, the influence of 3D reconstruction precision based on dry skull measurements is unclear. MIMICS, one of several commonly used 3D reconstruction software packages, provides users with the choice to select from four quality settings during the 3D model reconstruction process: low, medium, high and optimal. Ultimately, lower quality corresponds with a smaller file size and faster modeling computing speeds. In this study, four models were generated from a single skull using each of the four quality settings. Measurements were made of the parietal sagittal chord, cranial horizontal circumference, cranial surface area, cranial capacity, mastoid cell system surface area, and mastoid cell system volume of 43 reconstructed Yunnan modern human cranial specimens modeled using MIMICS. According to matrix simplification rules of MIMICS, optimal quality models were chosen as the standard for paired t-tests or non-parametric tests followed by the calculation of measurement difference (expressed as a percentage). Results indicated that the high-quality modeling group, including the parietal sagittal chord and mastoid cell system surface area measurements exhibited no difference in optimal quality. Conversely, measurement data of the other four characteristics used to generate simplified quality models significantly differed from optimal quality model data. Notably, measurement differences between simplified and optimal quality models of sagittal chord, cranial horizontal circumference, surface area of cranium, and cranial capacity were below 3%, while absolute values of measurement differences between low and optimal quality measurements of mastoid cell system surface area and volume exceeded 50% and 120%, respectively. These results suggest that low-quality 3D reconstruction models can be useful for measurements of large-scale morphological features with smooth surfaces. As for small-scale morphological features with rough surfaces such as the internal cavity sinus of the skull, three-dimensional reconstruction quality parameters must be selected very carefully.
Keywords: MIMICS; CT; 3D reconstruction; Measurements; Biological anthropology
