Backgrou nd Dense titanium(Ti)fusion cages have been commonly used in transforaminal lumbar interbody fusion.However,the stiffness mismatch between cages and adjacent bone endplates increases the risk of stress shield...Backgrou nd Dense titanium(Ti)fusion cages have been commonly used in transforaminal lumbar interbody fusion.However,the stiffness mismatch between cages and adjacent bone endplates increases the risk of stress shielding and cage subsidence.Methods The current study presents a multiscale optimization approach for porous Ti fusion cage development,including microscale topology optimization based on homogenization theory that obtains a unit cell with prescribed mechanical properties,and macroscale topology optimization that determines the layout of framework structure over the porous cage while maintaining the desired stiffness.The biomechanical performance of the designed porous cage is assessed using numerical simulations of fusion surgery.Selective laser melting is employed to assists with fabricating the designed porous structure and porous cage.Results The simulations demonstrate that the designed porous cage increases the strain energy density of bone grafts and decreases the peak stress on bone endplates.The mechanical and morphological discrepancies between the as-designed and fabricated porous structures are also described.Conclusion From the perspective of biomechanics,it is demonstrated that the designed porous cage contributes to reducing the risk of stress shielding and cage subsidence.The optimization of processing parameters and post-treatments are required to fabricate the designed porous cage.The present multiscale optimization approach can be extended to the development of cages with other shapes or materials and further types of orthopedic implants.展开更多
针对海量数据聚类过程中,经典的K-均值聚类算法对其K个初始聚类中心点的选择以及数据集噪声十分敏感的问题,提出了一种针对海量数据考虑初始聚类中心点选择的聚类算法。该算法首先采用冒泡排序法对数据集进行排序,获取数据集的各维中心...针对海量数据聚类过程中,经典的K-均值聚类算法对其K个初始聚类中心点的选择以及数据集噪声十分敏感的问题,提出了一种针对海量数据考虑初始聚类中心点选择的聚类算法。该算法首先采用冒泡排序法对数据集进行排序,获取数据集的各维中心值组成第一个初始聚类中心点。其次,通过计算与第一个初始聚类中心点的欧式距离,对剩余候选初始聚类中心点进行优化选择,保证所有的聚类中心点均匀地分布在数据集密度较大的空间上,以此减少聚类过程中的迭代次数和提高聚类算法效率。最后,基于UCI(University of California,Irvine)中多个数据集,进行聚类算法对比实验。结果表明,在不降低聚类效果的前提下,该聚类算法的迭代次数平均降低到50%,所需的时间降低平均达10%,由实验结果还能推出,当点集的数目越多时,该算法就能表现出越明显的聚类优势效果。展开更多
基金financially supported by the National Natural Science Foundation of China(No.51975336)the Key Basic Research Project of Natural Science Foundation of Shandong Province,China(No.ZR2018ZB0106)the Key Research and Development Program of Shandong Province,China(No.2019JZZY010112)。
文摘Backgrou nd Dense titanium(Ti)fusion cages have been commonly used in transforaminal lumbar interbody fusion.However,the stiffness mismatch between cages and adjacent bone endplates increases the risk of stress shielding and cage subsidence.Methods The current study presents a multiscale optimization approach for porous Ti fusion cage development,including microscale topology optimization based on homogenization theory that obtains a unit cell with prescribed mechanical properties,and macroscale topology optimization that determines the layout of framework structure over the porous cage while maintaining the desired stiffness.The biomechanical performance of the designed porous cage is assessed using numerical simulations of fusion surgery.Selective laser melting is employed to assists with fabricating the designed porous structure and porous cage.Results The simulations demonstrate that the designed porous cage increases the strain energy density of bone grafts and decreases the peak stress on bone endplates.The mechanical and morphological discrepancies between the as-designed and fabricated porous structures are also described.Conclusion From the perspective of biomechanics,it is demonstrated that the designed porous cage contributes to reducing the risk of stress shielding and cage subsidence.The optimization of processing parameters and post-treatments are required to fabricate the designed porous cage.The present multiscale optimization approach can be extended to the development of cages with other shapes or materials and further types of orthopedic implants.
文摘针对海量数据聚类过程中,经典的K-均值聚类算法对其K个初始聚类中心点的选择以及数据集噪声十分敏感的问题,提出了一种针对海量数据考虑初始聚类中心点选择的聚类算法。该算法首先采用冒泡排序法对数据集进行排序,获取数据集的各维中心值组成第一个初始聚类中心点。其次,通过计算与第一个初始聚类中心点的欧式距离,对剩余候选初始聚类中心点进行优化选择,保证所有的聚类中心点均匀地分布在数据集密度较大的空间上,以此减少聚类过程中的迭代次数和提高聚类算法效率。最后,基于UCI(University of California,Irvine)中多个数据集,进行聚类算法对比实验。结果表明,在不降低聚类效果的前提下,该聚类算法的迭代次数平均降低到50%,所需的时间降低平均达10%,由实验结果还能推出,当点集的数目越多时,该算法就能表现出越明显的聚类优势效果。
