Optimal scanning parameters of lumbar bone density measured by fast kilovoltage-switching dual-energy computed tomography (DECT).

ABSTRACT

Background: The technological innovation of fast kilovoltage (KV)-switching dual-energy computed tomography (DECT) has enabled the accurate measurement of vertebral bone density; however, it does not account for the effects of abdominal fat and ribs on the vertebral body. In our study, a European spine phantom (ESP) was used to establish an abdominal phantom for normal weight and obese people, and to explore the best scanning parameters for DECT to measure the bone mineral density (BMD) of the human lumbar spine. Methods: Revolution CT was used to conduct energy spectrum scanning for each body mode. A total of 20 sets of energy spectrum scans was conducted and each set of conditions was scanned 10 times. The data conformed to a normal distribution, and the differences between the measured and actual values of ESP L1-3 vertebrae were compared using a one-sample t-test, and quantitative data were described by x ¯ ± s . A P value <0.05 was considered statistically significant. Relative error (RE) and root mean square error (RMSE) of BMD measurements were calculated for different scanning conditions in normal and obese populations. Results: When simulating the upper abdominal condition (L1-2 level, fat area 140 cm2, with rib influence) in a normal weight population, there was no statistical difference (P>0.05) in BMD measurements for each vertebra at 0.8 s/rotation (rot) with different tube currents, the smallest RE at 0.8 s/rot, 190 mA condition, and the smallest RMSE for L1 and 2 vertebral BMD measurements at 190 mA; when simulating the abdominal condition at the L4 level in a normal weight population (fat area of 240 cm2, no rib influence), there were no statistical differences between the measurements at 0.8 s/rot, 190 and 275 mA conditions (P>0.05), and the RE and RMSE in the 190 mA condition was smaller than that in the 275 mA condition. Simulating the upper abdominal condition in the obese population (L1-2 level, fat area 340 cm2, with rib influence), there were no statistical difference between the measurements in the 0.8 s/rot, 315 and 355 mA conditions (P>0.05), the RE and RMSE in the 315 mA condition was less than those in the 355 mA; simulated obese abdominal condition at the L4 level in the population (fat area 450 cm2, no rib influence) resulted in 0.8 s/rot, no statistical difference in measurements between 315 mA (P>0.05), RE in 315 mA conditions were L1 3.75%, L2 -1.06%, L3 0.42%, and the RMSE under 315 mA condition were L1 2.13, L2 1.21, L3 1.66. Conclusions: When using Revolution CT to measure lumbar spine bone density, 0.8 s/rot at 190 mA may be the best scanning parameter for a normal weight population, and 0.8 s/rot at 315 mA may be the best scanning parameter for an obese population.