Effects of different combinations of noise index and preset adaptive statistical iterative reconstruction Veo on the accuracy and image quality of bone mineral density measurements using fast kilovolt-switching dual-energy computed tomography: a phantom study

Background: Rapid kilovolt (kV)-switching dual-energy computed tomography (DECT) has been increasingly applied to the measurement of lumbar spine bone mineral density (BMD) in humans and animal models. The objective of this study was to investigate the optimal parameters for the measurement of vertebral BMD. The BMD of the spinal model was measured by means of DECT in combination with different noise index (NI) and preset adaptive statistical iterative reconstruction Veo (ASiR-V) levels. The results were verified through simulating diverse total adipose tissue (TAT) conditions.

Methods: A total of 54 sets of parameters, including NI ranging from 4 to 20 (at intervals of 2) and ASiR-V level ranging from 0% to 100% (20% increments), were used to perform rapid kV-switching DECT scans on the European Spine Phantom (ESP). Hydroxyapatite (HAP) (water) was used as the base material to measure the equivalent density of HAP in the vertebral body by fast kV-switching DECT scanning and defined as BMD. Fresh porcine fat was wrapped around the phantom and divided into four groups according to different TAT cross-sectional areas of (S=0 cm², S=100 cm², S=200 cm², and S=350 cm²) to simulate belly fat levels in people with different body mass index (BMI) values. HAP value measured from 54 sets of parameters were compared using the TAT (S=0 cm²) group to find the optimal combination of NI and ASiR-V. The comparisons between four TAT groups were conducted between the optimal and default combinations. A one-sample t test was applied to analyze the differences between the phantom BMD values measured at the L1-L3 vertebral level and the corresponding true BMDs. One-way analysis of variance was used to analyze the differences in measurements under the different TAT conditions. Root mean square error (RMSE) of BMD measurements and contrast-to-noise ratio (CNR) were calculated and compared.

Results: For all three investigated vertebrae sections, there were significant differences (P<0.001). No significant differences existed between the measured HAP in three investigated vertebrae sections and the true values of the phantom (P>0.05) when NI =14 and ASiR-V_80%, NI =16 and ASiR-V_100%, NI =18 and ASiR-V_60-80%, and NI =20 and ASiR-V_60-100%. For the default combination (NI =6 and ASiR-V_0%), there was no significant difference between the measurements and the true values only in the 0-cm² TAT group. However, the optimal combination (NI =18 and ASiR-V_60%) did not lead to significant differences between the four TAT groups. Moreover, both combinations showed that RMSE increased and CNR declined with the increase in TAT, and there was less variation when NI =18 and ASiR-V_60%.

Conclusions: During the rapid kV-switching DECT scans, NI and ASiR-V, when combined with image quality and radiation dose, can improve the accuracy of BMD measurement, with the optimal parameter combination being NI =18 and ASiR-V_60%.