Purpose: Our aim was to estimate the in vivo reproducibility of bone mineral density (BMD) at dual-energy X-ray absorptiometry (DXA) and to compare fast array, array, and high-definition scan modes.
Materials and methods: A total of 378 patients (38 males and 340 females; mean age 63 ± 9 years) underwent DXA using a QDR-Discovery A densitometer (Hologic). Considering the three scan modes on lumbar spine and right femur, six independent groups of 30 patients were examined twice (for a total of 180 patients). Least significant change (LSC) and smallest detectable difference (SDD) were calculated. The remaining 198 patients underwent three scans of the lumbar spine (n = 92) or of the right femur (n = 106), one for each scan mode. The student t test and Bland-Altman analysis used were. Scan times were recorded and radiation dose was estimated using the ICRP60 method.
Results: Intra-scan mode reproducibility was 98-99%, corresponding to an LSC of 1.49-2.08%. The SDD was 0.018-0.023 g/cm(2) (lumbar spine) and 0.017-0.019 g/cm(2) (right femur). All comparisons among scan modes were statistically significant (p < 0.001) but lower than SDDs, i.e. not clinically relevant. Considering lumbar spine and the right femur, scan times were 50 and 38 s for fast array, 98 and 74 s for array, and 195 and 148 s for high definition, respectively; radiation doses were 6.7 and 4.7 μSv for fast array, and 13.3 and 9.3 μSv for both array and high definition, respectively.
Conclusion: Since all BMD differences were lower than the SSDs, the three scan modes can be considered interchangeable. As a consequence, although the absolute reduction in time and radiation dose is relatively low, when BMD measurement is the aim of DXA, fast array can be generally preferred.