Evaluation of precision and accuracy of bone mineral densitometry / 中华放射学杂志

ABSTRACT

Objective:To evaluate the short-term precision and accuracy of bone mineral density (BMD) measured with quantitative CT (QCT) and dual-energy X-ray absorptiometry (DXA) in multi-centre clinical research with a European spine phantom (ESP).Methods:From January 2016 to April 2020, totally 40 CT scanners (12 Siemens from Germany, 12 Philips from Netherlands, 9 GE from US, 5 Toshiba from Japan, 2 United Imaging from China) and 53 DXA instruments (34 GE Lunar from US, 14 Hologic from US, 5 Medlink from France) used in multi-centre in China were enrolled. The CT equipment came from 31 centers and DXA equipment from 32 centers.Using Mindways QCT software, the ESP was scanned 10 times on each instrument with repositioning using standardized spine protocols with QCT and DXA. The BMD value of the three lumbar vertebrae with low, medium, high density and the mean BMD value were measured. Accuracy was assessed by comparing BMD values measured on each device with the actual value of the phantom. Short-term precision was calculated as the root-mean-square standard deviation (RMS-SD) and root-mean-square standard deviation coefficient of variation (RMS-%CV) for the repeated measurements. The repeated measures variance analysis was used to compare the differences in BMD between different devices.Results:The differences in BMD values were statistically significant among different CT and DXA devices.The ranges of the accuracy measured by different QCT devices were 1.20% to 7.60% for Siemens, -1.83% to 0.20% for Philips, 1.18% to 13.20% for GE, -0.12% to 3.55% for Toshiba, -1.65% to 6.32% for United Imaging, 6.59% to 21.34% for GE Lunar, -6.65% to 5.45% for Hologic, and -6.97% to -0.68% for Medlink, respectively. The RMS-%CV of all vertebral BMD values measured by QCT and DXA ranged from 0.38% to 3.85%. The RMS-SD of QCT was 0.54 to 2.45 mg/cm 3, of DXA was 0.009 to 0.037 g/cm 2. The RMS-%CV values of each vertebral body measured by different QCT and DXA devices decreased with the increase of BMD, while the RMS-SD values showed the opposite tendency. Conclusions:Based on ESP, the QCT and DXA devices have significant differences in lumbar spine BMD measurement. Comparing the measurement results among different devices requires cross-calibration. Overall, the accuracy and short-term precision are within a reasonable range, which can be used for clinical follow-up observation. The short-term precision error and accuracy error range of QCT in evaluating the density of ESP were slightly smaller than those of DXA.