ABSTRACT
UNLABELLED: We used a standard DXA device equipped with a C-arm to do in vivo reconstruction of human vertebrae from two orthogonal scans. This new technique, called 3D-XA (three-dimensional X-ray absorptiometry), allows the direct measurement of geometric parameters of the vertebrae with a good accuracy and precision. INTRODUCTION: Geometric parameters are predictors of bone strength. A technique called three-dimensional X-ray absorptiometry (3D-XA) allows 3D reconstruction of bones from DXA scans. We used the 3D-XA method to reconstruct human vertebrae and to evaluate the method's in vitro accuracy and in vivo precision. METHODS: A standard DXA device equipped with a C-arm was used. Calibration of its environment and identification of different anatomical landmarks of the vertebrae allows personalized 3D geometric reconstruction of vertebrae. Accuracy was calculated by reconstructing 16 dry human vertebrae by 3D-XA and CT scanner. In vivo inter-observer precision was calculated using 20 human spines. RESULTS: The mean difference between 3D reconstruction by CT and 3D-XA was -0.2 +/- 1.3 mm. The in vivo mean difference of the 3D-XA method between the two rheumatologists was -0.1 +/- 0.8 mm. For geometric parameters, mean difference ranged from 0.4 to 0.9 mm. For cross-sectional area and vertebral body volume, it was 2.9% and 3.2%, respectively. CONCLUSION: This study shows the good accuracy and precision of 3D-XA using a standard DXA device. It yields complementary information on bone geometry. Further studies are needed to evaluate if, coupled with bone density, it improves vertebral fracture risk prediction.