Salo Zoryana, Beek Maarten, Wright David, Whyne Cari Marisa
University of Toronto Institute of Biomaterials and Biomedical Engineering, Toronto, ON, Canada.
Sunnybrook Research Institute, Holland Musculoskeletal Research Program, Toronto, ON, Canada.
J Biomech. 2015 Apr 13;48(6):1125-32. doi: 10.1016/j.jbiomech.2015.01.013. Epub 2015 Jan 21.
Current methods for the development of pelvic finite element (FE) models generally are based upon specimen specific computed tomography (CT) data. This approach has traditionally required segmentation of CT data sets, which is time consuming and necessitates high levels of user intervention due to the complex pelvic anatomy. The purpose of this research was to develop and assess CT landmark-based semi-automated mesh morphing and mapping techniques to aid the generation and mechanical analysis of specimen-specific FE models of the pelvis without the need for segmentation. A specimen-specific pelvic FE model (source) was created using traditional segmentation methods and morphed onto a CT scan of a different (target) pelvis using a landmark-based method. The morphed model was then refined through mesh mapping by moving the nodes to the bone boundary. A second target model was created using traditional segmentation techniques. CT intensity based material properties were assigned to the morphed/mapped model and to the traditionally segmented target models. Models were analyzed to evaluate their geometric concurrency and strain patterns. Strains generated in a double-leg stance configuration were compared to experimental strain gauge data generated from the same target cadaver pelvis. CT landmark-based morphing and mapping techniques were efficiently applied to create a geometrically multifaceted specimen-specific pelvic FE model, which was similar to the traditionally segmented target model and better replicated the experimental strain results (R(2)=0.873). This study has shown that mesh morphing and mapping represents an efficient validated approach for pelvic FE model generation without the need for segmentation.
目前用于开发骨盆有限元（FE）模型的方法通常基于特定标本的计算机断层扫描（CT）数据。传统上，这种方法需要对CT数据集进行分割，由于骨盆解剖结构复杂，这既耗时又需要高水平的用户干预。本研究的目的是开发和评估基于CT标志点的半自动网格变形和映射技术，以辅助生成骨盆特定标本的FE模型并进行力学分析，而无需进行分割。使用传统分割方法创建一个特定标本的骨盆FE模型（源模型），并使用基于标志点的方法将其变形到另一个（目标）骨盆的CT扫描上。然后通过将节点移动到骨边界进行网格映射来细化变形后的模型。使用传统分割技术创建第二个目标模型。将基于CT强度的材料属性分配给变形/映射模型和传统分割的目标模型。对模型进行分析以评估其几何一致性和应变模式。将双腿站立姿势配置中产生的应变与从同一目标尸体骨盆生成的实验应变片数据进行比较。基于CT标志点的变形和映射技术被有效地应用于创建一个几何多面的特定标本骨盆FE模型，该模型与传统分割的目标模型相似，并且能更好地复制实验应变结果（R(2)=0.873）。这项研究表明，网格变形和映射是一种无需分割即可有效验证的骨盆FE模型生成方法。
