The Effect of Quantitative Computed Tomography Acquisition Protocols on Bone Mineral Density Estimation

The Effect of Quantitative Computed Tomography Acquisition Protocols on Bone Mineral Density Estimation
复制标题

DOI:
10.1115/1.4031572
复制
发表时间:
2015-11-01
影响因子:
1.7
通讯作者:
Nassr, Ahmad
Nassr, Ahmad
中科院分区:
工程技术4区
文献类型:
--
作者:
Giambini, Hugo;Dragomir-Daescu, Dan;Nassr, Ahmad

文献摘要

被引文献

相似文献

骨质疏松症的特征是骨材料丢失和骨强度降低,导致骨折风险显着增加。患者特有的定量计算机断层扫描(QCT)有限元(FE)模型可用于预测生理载荷下的骨折。用于预测骨折的有限元模型的材料属性是通过使用校准模将CT灰度值转换为体积骨密度(VBMD)来获得的。如果CT采集协议出现任何变化,vBMD估计和材料特性分配都可能受到影响,从而影响骨折风险的预测。我们假设材料属性的分配可能依赖于扫描和后处理设置,包括电压、电流和重建核,因此可能会对骨折风险预测产生影响。采用不同方案对兔股骨和标准校准体模进行了QCT成像。对皮质区和松质区进行分割,得到它们的平均Hounsfield单位(HU)值并换算成vBMD。皮质和松质区的估计vBMD受电压和核仁的影响,但不受电流的影响。我们的研究表明,不同的扫描采集获得的vBMD值存在显著差异。此外,使用不同的扫描参数观察到的大的噪声差异可能会对包含较少体素的小子区域产生重要的负面影响。
Osteoporosis is characterized by bony material loss and decreased bone strength leading to a significant increase in fracture risk. Patient-specific quantitative computed tomography (QCT) finite element (FE) models may be used to predict fracture under physiological loading. Material properties for the FE models used to predict fracture are obtained by converting grayscale values from the CT into volumetric bone mineral density (vBMD) using calibration phantoms. If there are any variations arising from the CT acquisition protocol, vBMD estimation and material property assignment could be affected, thus, affecting fracture risk prediction. We hypothesized that material property assignments may be dependent on scanning and postprocessing settings including voltage, current, and reconstruction kernel, thus potentially having an effect in fracture risk prediction. A rabbit femur and a standard calibration phantom were imaged by QCT using different protocols. Cortical and cancellous regions were segmented, their average Hounsfield unit (HU) values obtained and converted to vBMD. Estimated vBMD for the cortical and cancellous regions were affected by voltage and kernel but not by current. Our study demonstrated that there exists a significant variation in the estimated vBMD values obtained with different scanning acquisitions. In addition, the large noise differences observed utilizing different scanning parameters could have an important negative effect on small subregions containing fewer voxels.