Low-Dose Simulation and Sparse Sampling with Statistical Iterative Reconstruction: Dose Reduction in MDCT-Based Bone Mineral Density and Microstructure Assessment

Introduction: The clinical use of multidetector computed tomography (MDCT) to assess bone strength and fracture risk in subjects with osteoporosis is limited by the relatively high radiation doses of established protocols. In this study, the effects of simulated dose reduction and sparse sampling combined with statistical iterative reconstruction (SIR) on bone mineral density (BMD) measurements and microstructure analyses of trabecular bone were assessed. We also explored whether analyses based on low-dose protocols were still able to differentiate between subjects with and without vertebral fractures.

Materials and Methods: In routine chest and abdominal MDCT scans of 12 subjects with prevalent osteoporotic vertebral fractures and 12 controls, pairwise matched for age and sex, lower radiation doses were simulated by virtual lower tube current (10%, 25%, and 50% of the original current) and sparse sampling (10%, 25%, and 50% of the original raw data). All images were reconstructed with SIR. BMD and trabecular bone microstructure parameters (including bone fraction and trabecular thickness) were extracted from the reconstructed images in T10 to L5.

Results: Effective dose of the original scan was 10 mSv, and 5, 3, and 1 mSv for the 50%, 25%, and 10% sparse sampled/low-dose simulated data, respectively. All BMD measurements calculated from different virtual lower tube currents and different sparse samplings were significantly lower in subjects with fractures compared with controls (range: 89–110 versus 125–187 mg/mL; p < 0.003 for all). Analogously, bone fraction (0.31–0.47 versus 0.50–0.52; p < 0.004 for all) and trabecular thickness were consistently lower in subjects with fractures (1.1–1.3 versus 1.7–2.4 mm; p < 0.011 for all).

Conclusion: After simulated dose reduction and sparse sampling, BMD and microstructure parameters obtained from MDCT examinations and processed with statistical iterative reconstruction were still significantly different in subjects with and without vertebral fractures, suggesting an osteoporosis diagnosis to be feasible in low-dose protocols.