Haverfield Zachary A, Agnew Amanda M, Loftis Kathryn, Zhang Jun, Hayden Lauren E, Hunter Randee L
Injury Biomechanics Research Center, The Ohio State University, Columbus, Ohio 43210, United States.
United States Army Futures Command DEVCOM Analysis Center, Aberdeen Proving Ground, Maryland, 21005, United States.
JBMR Plus. 2024 Aug 3;8(10):ziae106. doi: 10.1093/jbmrpl/ziae106. eCollection 2024 Oct.
Volumetric bone mineral density (vBMD) is commonly assessed using QCT. Although standard vBMD calculation methods require phantom rods that may not be available, internal-reference phantomless (IPL) and direct measurements of Hounsfield units (HU) can be used to calculate vBMD in their absence. Yet, neither approach has been systemically assessed across skeletal sites, and HU need further validation as a vBMD proxy. This study evaluated the accuracy of phantomless methods, including IPL and regression-based phantomless (RPL) calibration using HU to calculate vBMD, compared to phantom-based (PB) methods. vBMD from QCT scans of 100 male post-mortem human subjects (PMHS) was calculated using site-specific PB calibration at multiple skeletal sites throughout the body. A development sample of 50/100 PMHS was used to determine site-specific reference material density for IPL calibration and RPL equations. Reference densities and equations from the development sample were used to calculate IPL and RPL vBMD on the remaining 50/100 PMHS for method validation. PB and IPL/RPL vBMD were not significantly different ( > .05). Univariate regressions between PB and IPL/RPL vBMD were universally significant ( < 0.05), except for IPL Rad-30 ( = 0.078), with a percent difference across all sites of 6.97% ± 5.95% and 5.22% ± 4.59% between PB and IPL/RPL vBMD, respectively. As vBMD increased, there were weaker relationships and larger differences between PB vBMD and IPL/RPL vBMD. IPL and RPL vBMD had strong relationships with PB vBMD across sites (R = 97.99, R = 99.17%, respectively), but larger residual differences were found for IPL vBMD. As the accuracy of IPL/RPL vBMD varied between sites, phantomless methods should be site-specific to provide values more comparable to PB vBMD. Overall, this study suggests that RPL calibration may better represent PB vBMD compared to IPL calibration, increases the utility of opportunistic QCT, and provides insight into bone quality and fracture risk.
容积骨密度（vBMD）通常使用定量计算机断层扫描（QCT）进行评估。尽管标准的vBMD计算方法需要可能无法获取的体模棒，但在没有体模棒的情况下，可以使用内部参考无体模（IPL）和亨氏单位（HU）的直接测量来计算vBMD。然而，这两种方法都尚未在全身骨骼部位进行系统评估，并且HU作为vBMD替代指标需要进一步验证。本研究评估了无体模方法的准确性，包括使用HU进行基于IPL和基于回归的无体模（RPL）校准来计算vBMD，并与基于体模（PB）的方法进行比较。使用针对全身多个骨骼部位的特定部位PB校准，计算了100名男性尸体解剖人类受试者（PMHS）的QCT扫描的vBMD。使用50/100名PMHS的开发样本确定用于IPL校准和RPL方程的特定部位参考材料密度。将开发样本中的参考密度和方程用于计算其余50/100名PMHS的IPL和RPL vBMD，以进行方法验证。PB和IPL/RPL vBMD无显著差异（>0.05）。PB与IPL/RPL vBMD之间的单变量回归普遍显著（<0.05），除了IPL Rad-30（=0.078），PB与IPL/RPL vBMD在所有部位的百分比差异分别为6.97%±5.95%和5.22%±4.59%。随着vBMD增加，PB vBMD与IPL/RPL vBMD之间的关系变弱且差异增大。IPL和RPL vBMD在各部位与PB vBMD有很强的关系（R分别为97.99、R为99.17%），但IPL vBMD的残差差异更大。由于IPL/RPL vBMD的准确性在不同部位有所不同，无体模方法应针对特定部位，以提供与PB vBMD更具可比性的值。总体而言，本研究表明，与IPL校准相比，RPL校准可能更好地代表PB vBMD，提高了机会性QCT的实用性，并为骨质量和骨折风险提供了见解。
